1 <!-- $PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.573 2008/01/01 19:36:20 momjian Exp $ -->
7 [A-Z][A-Z ]+[A-Z] <command>
8 [A-Za-z_][A-Za-z0-9_]+() <function>
9 [A-Za-z_]/[A-Za-z_]+ <filename>
10 PostgreSQL <productname>
11 pg_[A-Za-z0-9_] <application>
12 [A-Z][A-Z] <type>, <envar>, <literal>
14 non-ASCII characters convert to HTML4 entity (&) escapes
16 official: http://www.w3.org/TR/html4/sgml/entities.html
17 one page: http://www.zipcon.net/~swhite/docs/computers/browsers/entities_page.html
18 other lists: http://www.zipcon.net/~swhite/docs/computers/browsers/entities.html
19 http://www.zipcon.net/~swhite/docs/computers/browsers/entities_page.html
20 http://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references
22 we cannot use UTF8 because SGML Docbook
24 http://www.pemberley.com/janeinfo/latin1.html#latexta
28 For new features, add links to the documentation sections. Use </link>
29 so that Perl can remove it so HISTORY.html can be created with no
30 links to the main documentation. This was added only in 8.2, so don't
31 do it for earlier branch release files.
35 <appendix id="release">
36 <title>Release Notes</title>
38 <sect1 id="release-introduction">
39 <title>Introduction</title>
42 The release notes contain the significant changes in each PostgreSQL
43 release, with major features and migration issues often listed at the
44 top. The release notes do not contain changes that affect only a few
45 users or changes that are internal and therefore not user-visible. For
46 example, the optimizer is improved in almost every release, but the
47 improvements are usually observed by users as simply faster queries.
51 A complete list of changes for each release can be obtained by
52 viewing the <link linkend="cvs">CVS</link> logs for each release. The <ulink
53 url="http://archives.postgresql.org/pgsql-committers/">committers email
54 list</ulink> contains all source code changes as well. There is also
55 a <ulink url="http://developer.postgresql.org/cvsweb.cgi/pgsql/">web
56 interface</ulink> that shows changes to specific files.
57 <!-- we need a file containing the CVS logs for each release, and something
58 like the SVN web interface that groups commits but has branches -->
62 The name appearing next to each item represents the major developer for
63 that item. Of course all changes involve community discussion and patch
64 review so each item is truly a community effort.
69 <sect1 id="release-8-3">
70 <title>Release 8.3</title>
73 <title>Release date</title>
74 <simpara>2008-01-??</simpara>
75 <para>CURRENT AS OF 2008-01-01</>
79 <title>Overview</title>
82 With significant new functionality and performance enhancements,
83 this release represents a major leap forward for
84 <productname>PostgreSQL</>. This was made possible by a growing
85 community that has dramatically accelerated the pace of
86 development. This release adds the following major features:
93 Full text search is integrated into the core database system
99 Support for the SQL/XML standard, including new operators and an
100 <type>XML</type> data type
106 Enumerated data types (<type>ENUM</type>)
112 Arrays of composite types
118 Universally Unique Identifier (<type>UUID</>) data type
124 Add control over whether <literal>NULL</>s sort first or last
136 Server configuration parameters can now be set on a per-function
143 User-defined types can now have type modifiers
149 Automatically re-plan cached queries when table
150 definitions change or statistics are updated
156 Numerous improvements in logging and statistics collection
162 Support Security Service Provider Interface (<acronym>SSPI</>) for
163 authentication on Windows
169 Support multiple concurrent autovacuum processes, and other
170 autovacuum improvements
176 Allow the whole <productname>PostgreSQL</> distribution to be compiled
177 with <productname>Microsoft Visual C++</>
184 Major performance improvements are listed below. Most of
185 these enhancements are automatic and do not require user changes or
193 Asynchronous commit delays writes to WAL during transaction commit
199 Checkpoint writes can be spread over a longer time period to smooth
200 the I/O spike during each checkpoint
206 Heap-Only Tuples (<acronym>HOT</>) accelerate space reuse for
207 most <command>UPDATE</>s and <command>DELETE</>s
213 Just-in-time background writer strategy improves disk write
220 Using non-persistent transaction IDs for read-only transactions
221 reduces overhead and <command>VACUUM</> requirements
227 Per-field and per-row storage overhead has been reduced
233 Large sequential scans no longer force out frequently used
240 Concurrent large sequential scans can now share disk reads
246 <literal>ORDER BY ... LIMIT</> can be done without sorting
253 The above items are explained in more detail in the sections below.
259 <title>Migration to Version 8.3</title>
262 A dump/restore using <application>pg_dump</application> is
263 required for those wishing to migrate data from any previous
268 Observe the following incompatibilities:
272 <title>General</title>
277 Non-character data types are no longer automatically cast to
278 <type>TEXT</> (Peter, Tom)
282 Previously, if a non-character value was supplied to an operator or
283 function that requires <type>text</> input, it was automatically
284 cast to <type>text</>, for most (though not all) built-in data types.
285 This no longer happens: an explicit cast to <type>text</> is now
286 required for all non-character-string types. For example, these
287 expressions formerly worked:
290 substr(current_date, 1, 4)
294 but will now draw <quote>function does not exist</> and <quote>operator
295 does not exist</> errors respectively. Use an explicit cast instead:
298 substr(current_date::text, 1, 4)
302 (Of course, you can use the more verbose <literal>CAST()</> syntax too.)
303 The reason for the change is that these automatic casts too often caused
304 surprising behavior. An example is that in previous releases, this
305 expression was accepted but did not do what was expected:
308 current_date < 2017-11-17
311 This is actually comparing a date to an integer, which should be
312 (and now is) rejected — but in the presence of automatic
313 casts both sides were cast to <type>text</> and a textual comparison
314 was done, because the <literal>text < text</> operator was able
315 to match the expression when no other <literal><</> operator could.
319 Types <type>char(<replaceable>n</>)</type> and
320 <type>varchar(<replaceable>n</>)</type> still cast to <type>text</>
321 automatically. Also, automatic casting to <type>text</> still works for
322 inputs to the concatenation (<literal>||</>) operator, so long as least
323 one input is a character-string type.
329 Full text search features from <filename>contrib/tsearch2</> have
330 been moved into the core server, with some minor syntax changes
334 <filename>contrib/tsearch2</> now contains a compatibility
341 <literal>ARRAY(SELECT ...)</literal>, where the <command>SELECT</>
342 returns no rows, now returns an empty array, rather than NULL
349 The array type name for a base data type is no longer always the base
350 type's name with an underscore prefix
354 The old naming convention is still honored when possible, but
355 application code should no longer depend on it. Instead
356 use the new <literal>pg_type.typarray</literal> column to
357 identify the array data type associated with a given type.
363 <literal>ORDER BY ... USING</> <replaceable>operator</> must now
364 use a less-than or greater-than <replaceable>operator</> that is
365 defined in a btree operator class
369 This restriction was added to prevent inconsistent results.
375 <command>SET LOCAL</command> changes now persist until
376 the end of the outermost transaction, unless rolled back (Tom)
380 Previously <command>SET LOCAL</command>'s effects were lost
381 after subtransaction commit (<command>RELEASE SAVEPOINT</>
382 or exit from a PL/pgSQL exception block).
388 Commands rejected in transaction blocks are now also rejected in
389 multiple-statement query strings (Tom)
393 For example, <literal>"BEGIN; DROP DATABASE; COMMIT"</> will now be
394 rejected even if submitted as a single query message.
400 <command>ROLLBACK</> outside a transaction block now
401 issues <literal>NOTICE</> instead of <literal>WARNING</> (Bruce)
407 Prevent <command>NOTIFY</command>/<command>LISTEN</command>/<command>UNLISTEN</command>
408 from accepting schema-qualified names (Bruce)
412 Formerly, these commands accepted <literal>schema.relation</> but
413 ignored the schema part, which was confusing.
419 <command>ALTER SEQUENCE</> no longer affects the sequence's
420 <function>currval()</> state (Tom)
426 Foreign keys now must match indexable conditions for
427 cross-data-type references (Tom)
431 This improves semantic consistency and helps avoid
432 performance problems.
438 Restrict object size functions to users who have reasonable
439 permissions to view such information (Tom)
443 For example, <function>pg_database_size()</function> now requires
444 <literal>CONNECT</> permission, which is granted to everyone by
445 default. <function>pg_tablespace_size()</function> requires
446 <literal>CREATE</> permission in the tablespace, or is allowed if
447 the tablespace is the default tablespace for the database.
453 Remove the undocumented <literal>!!=</> (not in) operator (Tom)
457 <literal>NOT IN (SELECT ...)</literal> is the proper way to
458 perform this operation.
464 Internal hashing functions are now more uniformly-distributed (Tom)
468 If application code was computing and storing hash values using
469 internal <productname>PostgreSQL</> hashing functions, the hash
470 values must be regenerated.
476 C-code conventions for handling variable-length data values
477 have changed (Greg Stark, Tom)
481 The new <function>SET_VARSIZE()</> macro <emphasis>must</> be used
482 to set the length of generated <type>varlena</> values. Also, it
483 might be necessary to expand (<quote>de-TOAST</quote>) input values
490 Continuous archiving no longer reports each successful archive
491 operation to the server logs unless <literal>DEBUG</> level is used
501 <title>Configuration Parameters</title>
507 Numerous changes in administrative server parameters
511 <varname>bgwriter_lru_percent</>,
512 <varname>bgwriter_all_percent</>,
513 <varname>bgwriter_all_maxpages</>,
514 <varname>stats_start_collector</>, and
515 <varname>stats_reset_on_server_start</> are removed.
516 <varname>redirect_stderr</> is renamed to
517 <varname>logging_collector</>.
518 <varname>stats_command_string</> is renamed to
519 <varname>track_activities</>.
520 <varname>stats_block_level</> and <varname>stats_row_level</>
521 are merged into <varname>track_counts</>.
522 A new boolean configuration parameter, <varname>archive_mode</>,
523 controls archiving. Autovacuum's default settings have changed.
529 Remove <varname>stats_start_collector</varname> parameter (Tom)
533 We now always start the collector process, unless <acronym>UDP</>
534 socket creation fails.
540 Remove <varname>stats_reset_on_server_start</varname> parameter (Tom)
544 This was removed because <function>pg_stat_reset()</function>
545 can be used for this purpose.
551 Commenting out a parameter in <filename>postgresql.conf</> now
552 causes it to revert to its default value (Joachim Wieland)
556 Previously, commenting out an entry left the parameter's value unchanged
557 until the next server restart.
567 <title>Character Encodings</title>
573 Add more checks for invalidly-encoded data (Andrew)
577 This change plugs some holes that existed in literal backslash
578 escape string processing and <command>COPY</command> escape
579 processing. Now the de-escaped string is rechecked to see if the
580 result created an invalid multi-byte character.
586 Disallow database encodings that are inconsistent with the server's
591 On most platforms, <literal>C</> locale is the only locale that
592 will work with any database encoding. Other locale settings imply
593 a specific encoding and will misbehave if the database encoding
594 is something different. (Typical symptoms include bogus textual
595 sort order and wrong results from <function>upper()</> or
596 <function>lower()</>.) The server now rejects attempts to create
597 databases that have an incompatible encoding.
603 Ensure that <function>chr()</function> cannot create
604 invalidly-encoded values (Andrew)
608 In UTF8-encoded databases the argument of <function>chr()</function> is
609 now treated as a Unicode code point. In other multi-byte encodings
610 <function>chr()</function>'s argument must designate a 7-bit ASCII
611 character. Zero is no longer accepted.
612 <function>ascii()</function> has been adjusted to match.
618 Adjust <function>convert()</function> behavior to ensure encoding
623 The two argument form of <function>convert()</function> has been
624 removed. The three argument form now takes a <type>bytea</type>
625 first argument and returns a <type>bytea</type>. To cover the
626 loss of functionality, three new functions have been added:
632 <function>convert_from(bytea, name)</function> returns
633 <type>text</> — converts the first argument from the named
634 encoding to the database encoding
640 <function>convert_to(text, name)</function> returns
641 <type>bytea</> — converts the first argument from the
642 database encoding to the named encoding
648 <function>length(bytea, name)</function> returns
649 <type>integer</> — gives the length of the first
650 argument in characters in the named encoding
658 Remove <literal>convert(argument USING conversion_name)</literal>
663 Its behavior did not match the SQL standard.
669 Make JOHAB encoding client-only (Tatsuo)
673 JOHAB is not safe as a server-side encoding.
684 <title>Changes</title>
687 Below you will find a detailed account of the
688 changes between <productname>PostgreSQL</productname> 8.3 and
689 the previous major release.
693 <title>Performance</title>
698 Asynchronous commit delays writes to WAL during transaction commit
703 This feature dramatically increases performance for short data-modifying
704 transactions. The disadvantage is that because disk writes are
705 delayed, if the operating system crashes before data is written to
706 the disk, committed data will be lost. This feature is useful for
707 applications that can accept some data loss. Unlike turning off
708 <varname>fsync</varname>, asynchronous commit does not put database
709 consistency at risk; the worst case is that after a database or system
710 crash the last few reportedly-committed transactions might be missing.
711 This feature is enabled by turning off <varname>synchronous_commit</>
712 (which can be done per-session or per-transaction, if some transactions
713 are critical and others are not).
714 <varname>wal_writer_delay</> can be adjusted to control the maximum
715 delay before transactions actually reach disk.
721 Checkpoint writes can be spread over a longer time period to smooth
722 the I/O spike during each checkpoint (Itagaki Takahiro and Heikki
727 Previously all modified buffers were forced to disk as quickly as
729 checkpoint, causing an I/O spike that decreased server performance.
730 This new approach spreads out disk writes during checkpoints,
731 reducing peak I/O usage. (User-requested and shutdown checkpoints
732 are still written as quickly as possible.)
738 Heap-Only Tuples (<acronym>HOT</>) accelerate space reuse for most
739 <command>UPDATE</>s and <command>DELETE</>s (Pavan Deolasee, with
740 ideas from many others)
744 <command>UPDATE</>s and <command>DELETE</>s leave dead tuples
745 behind, as do failed <command>INSERT</>s. Previously only
746 <command>VACUUM</> could reclaim space taken by dead tuples. With
747 <acronym>HOT</> dead tuple space can be reclaimed at the time of
748 <command>INSERT</> or <command>UPDATE</> if no changes are made to
749 indexed columns. This allows for more consistent performance.
750 Also, <acronym>HOT</> avoids adding duplicate index entries.
756 Just-in-time background writer strategy improves disk write
757 efficiency (Greg Smith, Itagaki Takahiro)
761 This greatly reduces the need for manual tuning of the background
768 Per-field and per-row storage overhead have been reduced
769 (Greg Stark, Heikki Linnakangas)
773 Variable-length data types with data values less than 128 bytes long
774 will see a storage decrease of 3 to 6 bytes. For example, two adjacent
775 <type>char(1)</type> fields now use 4 bytes instead of 16. Row headers
776 are also 4 bytes shorter than before.
782 Using non-persistent transaction IDs for read-only transactions
783 reduces overhead and <command>VACUUM</> requirements (Florian Pflug)
787 Non-persistent transaction IDs do not increment the global
788 transaction counter. Therefore, they reduce the load on
789 <structname>pg_clog</> and increase the time between forced
790 vacuums to prevent transaction ID wraparound.
792 improvements were also made that should improve concurrency.
798 Avoid incrementing the command counter after a read-only command (Tom)
802 There was formerly a hard limit of 2<superscript>32</>
803 (4 billion) commands per transaction. Now only commands that
804 actually changed the database count, so while this limit still
805 exists, it should be significantly less annoying.
811 Create a dedicated <acronym>WAL</> writer process to off-load
812 work from backends (Simon)
818 Skip unnecessary WAL writes for <command>CLUSTER</command> and
819 <command>COPY</command> (Simon)
823 Unless WAL archiving is enabled, the system now avoids WAL writes
824 for <command>CLUSTER</command> and just <function>fsync()</>s the
825 table at the end of the command. It also does the same for
826 <command>COPY</command> if the table was created in the same
833 Large sequential scans no longer force out frequently used
834 cached pages (Simon, Heikki, Tom)
840 Concurrent large sequential scans can now share disk reads (Jeff Davis)
844 This is accomplished by starting the new sequential scan in the
845 middle of the table (where another sequential scan is already
846 in-progress) and wrapping around to the beginning to finish. This
847 can affect the order of returned rows in a query that does not
848 specify <literal>ORDER BY</>.
854 <literal>ORDER BY ... LIMIT</> can be done without sorting
859 This is done by sequentially scanning the table and tracking just
860 the <quote>top N</> candidate rows, rather than performing a
861 full sort of the entire table. This is useful when there is no
862 matching index and the <literal>LIMIT</> is not large.
868 Put a rate limit on messages sent to the statistics
869 collector by backends
874 This reduces overhead for short transactions, but might sometimes
875 increase the delay before statistics are tallied.
881 Improve hash join performance for cases with many NULLs (Tom)
887 Speed up operator lookup for cases with non-exact datatype matches (Tom)
896 <title>Server</title>
901 Autovacuum is now enabled by default (Alvaro)
905 Several changes were made to eliminate disadvantages of having
906 autovacuum enabled, thereby justifying the change in default.
907 Several other autovacuum parameter defaults were also modified.
913 Support multiple concurrent autovacuum processes (Alvaro, Itagaki
918 This allows multiple vacuums to run concurrently. This prevents
919 vacuuming of a large table from delaying vacuuming of smaller tables.
925 Automatically re-plan cached queries when table
926 definitions change or statistics are updated (Tom)
930 Previously PL/PgSQL functions that referenced temporary tables
931 would fail if the temporary table was dropped and recreated
932 between function invocations, unless <literal>EXECUTE</> was
933 used. This improvement fixes that problem and many related issues.
939 Add a <varname>temp_tablespaces</varname> parameter to control
940 the tablespaces for temporary tables and files (Jaime Casanova,
941 Albert Cervera, Bernd Helmle)
945 This parameter defines a list of tablespaces to be used. This
946 enables spreading the I/O load across multiple tablespaces. A random
947 tablespace is chosen each time a temporary object is created.
948 Temporary files are no longer stored in per-database
949 <filename>pgsql_tmp/</filename> directories but in per-tablespace
956 Place temporary tables' TOAST tables in special schemas named
957 <literal>pg_toast_temp_<replaceable>nnn</></literal> (Tom)
961 This allows low-level code to recognize these tables as temporary,
962 which enables various optimizations such as not WAL-logging changes
963 and using local rather than shared buffers for access. This also
964 fixes a bug wherein backends unexpectedly held open file references
965 to temporary TOAST tables.
971 Fix problem that a constant flow of new connection requests could
972 indefinitely delay the postmaster from completing a shutdown or
973 a crash restart (Tom)
979 Guard against a very-low-probability data loss scenario by preventing
980 re-use of a deleted table's relfilenode until after the next
987 Fix <command>CREATE CONSTRAINT TRIGGER</>
988 to convert old-style foreign key trigger definitions into regular
989 foreign key constraints (Tom)
993 This will ease porting of foreign key constraints carried forward from
994 pre-7.3 databases, if they were never converted using
995 <filename>contrib/adddepend</>.
1001 Fix <literal>DEFAULT NULL</> to override inherited defaults (Tom)
1005 <literal>DEFAULT NULL</> was formerly considered a noise phrase, but it
1006 should (and now does) override non-null defaults that would otherwise
1007 be inherited from a parent table or domain.
1013 Add new encodings EUC_JIS_2004 and SHIFT_JIS_2004 (Tatsuo)
1017 These new encodings can be converted to and from UTF-8.
1023 Change server startup log message from <quote>database system is
1024 ready</quote> to <quote>database system is ready to accept
1025 connections</quote>, and adjust its timing
1029 The message now appears only when the postmaster is really ready
1030 to accept connections.
1039 <title>Monitoring</title>
1044 Add <varname>log_autovacuum_min_duration</varname> parameter to
1045 support configurable logging of autovacuum activity (Simon, Alvaro)
1051 Add <varname>log_lock_waits</varname> parameter to log lock waiting
1058 Add <varname>log_temp_files</varname> parameter to log temporary
1059 file usage (Bill Moran)
1065 Add <varname>log_checkpoints</varname> parameter to improve logging
1066 of checkpoints (Greg Smith, Heikki)
1072 <varname>log_line_prefix</varname> now supports
1073 <literal>%s</literal> and <literal>%c</literal> escapes in all
1078 Previously these escapes worked only for user sessions, not for
1079 background database processes.
1085 Add <varname>log_restartpoints</varname> to control logging of
1086 point-in-time recovery restart points (Simon)
1092 Last transaction end time is now logged at end of recovery and at
1093 each logged restart point (Simon)
1099 Autovacuum now reports its activity start time in
1100 <literal>pg_stat_activity</literal> (Tom)
1106 Allow server log output in comma-separated value (CSV) format (Arul
1107 Shaji, Greg Smith, Andrew Dunstan)
1111 CSV-format log files can easily be loaded into a database table for
1112 subsequent analysis.
1118 Use PostgreSQL-supplied timezone support for formatting timestamps
1119 displayed in the server log (Tom)
1123 This avoids Windows-specific problems with localized time zone
1124 names that are in the wrong encoding. There is a new
1125 <varname>log_timezone</> parameter that controls the timezone
1126 used in log messages, independently of the client-visible
1127 <varname>timezone</> parameter.
1133 New system view <literal>pg_stat_bgwriter</literal> displays
1134 statistics about background writer activity (Magnus)
1140 Add new columns for database-wide tuple statistics to
1141 <literal>pg_stat_database</literal> (Magnus)
1147 Add an <literal>xact_start</literal> (transaction start time) column to
1148 <literal>pg_stat_activity</literal> (Neil)
1152 This makes it easier to identify long-running transactions.
1158 Add <literal>n_live_tuples</> and <literal>n_dead_tuples</> columns
1159 to <literal>pg_stat_all_tables</literal> and related views (Glen
1166 Merge <varname>stats_block_level</> and <varname>stats_row_level</>
1167 parameters into a single parameter <varname>track_counts</>, which
1168 controls all messages sent to the statistics collector process
1175 Rename <varname>stats_command_string</varname> parameter to
1176 <varname>track_activities</varname> (Tom)
1182 Fix statistical counting of live and dead tuples to recognize that
1183 committed and aborted transactions have different effects (Tom)
1192 <title>Authentication</title>
1197 Support Security Service Provider Interface (<acronym>SSPI</>) for
1198 authentication on Windows (Magnus)
1204 Support GSSAPI authentication (Henry Hotz, Magnus)
1208 This should be preferred to native Kerberos authentication because
1209 GSSAPI is an industry standard.
1215 Support a global SSL configuration file (Victor Wagner)
1221 Add <varname>ssl_ciphers</> parameter to control accepted SSL ciphers
1228 Add a Kerberos realm parameter, <varname>krb_realm</> (Magnus)
1237 <title>Write-Ahead Log (<acronym>WAL</>) and Continuous Archiving</title>
1242 Change the timestamps recorded in transaction WAL records from
1243 time_t to TimestampTz representation (Tom)
1247 This provides sub-second resolution in WAL, which can be useful for
1248 point-in-time recovery.
1254 Reduce WAL disk space needed by warm standby servers (Simon)
1258 This change allows a warm standby server to pass the name of the earliest
1259 still-needed WAL file to the recovery script, allowing automatic removal
1260 of no-longer-needed WAL files. This is done using <literal>%r</> in
1261 the <varname>restore_command</varname> parameter of
1262 <filename>recovery.conf</filename>.
1268 New boolean configuration parameter, <varname>archive_mode</>,
1269 controls archiving (Simon)
1273 Previously setting <varname>archive_command</> to an empty string
1274 turned off archiving. Now <varname>archive_mode</> turns archiving
1275 on and off, independently of <varname>archive_command</>. This is
1276 useful for stopping archiving temporarily.
1285 <title>Queries</title>
1290 Full text search is integrated into the core database
1291 system (Teodor, Oleg)
1295 Text search has been improved, moved into the core code, and is now
1296 installed by default. <filename>contrib/tsearch2</> now contains
1297 a compatibility interface.
1303 Add control over whether <literal>NULL</>s sort first or last (Teodor, Tom)
1307 The syntax is <literal>ORDER BY ... NULLS FIRST/LAST</>.
1313 Allow per-column ascending/descending (<literal>ASC</>/<literal>DESC</>)
1314 ordering options for indexes (Teodor, Tom)
1318 Previously a query using <literal>ORDER BY</> with mixed
1319 <literal>ASC</>/<literal>DESC</> specifiers could not fully use
1320 an index. Now an index can be fully used in such cases if the
1321 index was created with matching
1322 <literal>ASC</>/<literal>DESC</> specifications.
1323 <literal>NULL</> sort order within an index can be controlled, too.
1329 Allow <literal>col IS NULL</> to use an index (Teodor)
1335 Updatable cursors (Arul Shaji, Tom)
1339 This eliminates the need to reference a primary key to
1340 <command>UPDATE</> or <command>DELETE</> rows returned by a cursor.
1341 The syntax is <literal>UPDATE/DELETE WHERE CURRENT OF</>.
1347 Allow <literal>FOR UPDATE</literal> in cursors (Arul Shaji, Tom)
1353 Create a general mechanism that supports casts to and from the
1354 standard string types (<type>TEXT</type>, <type>VARCHAR</type>,
1355 <type>CHAR</type>) for <emphasis>every</emphasis> datatype, by
1356 invoking the datatype's I/O functions (Tom)
1360 Previously, such casts were available only for types that had
1361 specialized function(s) for the purpose.
1362 These new casts are assignment-only in the to-string direction,
1363 explicit-only in the other direction, and therefore should create no
1364 surprising behavior.
1370 Allow <literal>UNION</> and related constructs to return a domain
1371 type, when all inputs are of that domain type (Tom)
1375 Formerly, the output would be considered to be of the domain's base
1382 Allow limited hashing when using two different data types (Tom)
1386 This allows hash joins, hash indexes, hashed subplans, and hash
1387 aggregation to be used in situations involving cross-data-type
1388 comparisons, if the data types have compatible hash functions.
1389 Currently, cross-data-type hashing support exists for
1390 <type>smallint</type>/<type>integer</type>/<type>bigint</type>,
1391 and for <type>float4</type>/<type>float8</type>.
1397 Improve optimizer logic for detecting when variables are equal
1398 in a <literal>WHERE</> clause (Tom)
1402 This allows mergejoins to work with descending sort orders, and
1403 improves recognition of redundant sort columns.
1409 Improve performance when planning large inheritance trees in
1410 cases where most tables are excluded by constraints (Tom)
1419 <title>Object Manipulation</title>
1425 Arrays of composite types (David Fetter, Andrew, Tom)
1429 In addition to arrays of explicitly-declared composite types,
1430 arrays of the rowtypes of regular tables and views are now
1431 supported, except for rowtypes of system catalogs, sequences, and TOAST
1439 Server configuration parameters can now be set on a per-function
1444 For example, functions can now set their own
1445 <varname>search_path</> to prevent unexpected behavior if a
1446 different <varname>search_path</> exists at run-time. Security
1447 definer functions should set <varname>search_path</varname> to
1448 avoid security loopholes.
1454 <command>CREATE/ALTER FUNCTION</command> now supports
1455 <literal>COST</literal> and <literal>ROWS</literal> options (Tom)
1459 <literal>COST</literal> allows specification of the cost of a
1460 function call. <literal>ROWS</literal> allows specification of
1461 the average number or rows returned by a set-returning function.
1462 These values are used by the optimizer in choosing the best plan.
1468 Allow triggers and rules to be deactivated in groups using a
1469 configuration parameter, for replication purposes (Jan)
1473 This allows replication systems to disable triggers and rewrite
1474 rules as a group without modifying the system catalogs directly.
1475 The behavior is controlled by <command>ALTER TABLE</> and a new
1476 parameter <varname>session_replication_role</varname>.
1482 User-defined types can now have type modifiers (Teodor, Tom)
1486 This allows a user-defined type to take a modifier, like
1487 <type>ssnum(7)</>. Previously only built-in
1488 data types could have modifiers.
1497 <title>Utility Commands</title>
1502 Non-superuser database owners now are able to add trusted procedural
1503 languages to their databases by default (Jeremy Drake)
1507 While this is reasonably safe, some administrators might wish to
1508 revoke the privilege. It is controlled by
1509 <structname>pg_pltemplate</>.<structfield>tmpldbacreate</>.
1515 Allow a session's current parameter setting to be used as the
1516 default for future sessions (Tom)
1520 This is done with <literal>SET ... FROM CURRENT</literal> in
1521 <command>CREATE/ALTER FUNCTION</command>, <command>ALTER
1522 DATABASE</command>, or <command>ALTER ROLE</command>.
1528 Implement new commands <command>DISCARD ALL</command>,
1529 <command>DISCARD PLANS</command>, <command>DISCARD
1530 TEMPORARY</command>, <command>CLOSE ALL</command>, and
1531 <command>DEALLOCATE ALL</command> (Marko Kreen, Neil)
1535 These commands simplify resetting a database session to its initial
1536 state, and are particularly useful for connection-pooling software.
1542 Add <command>ALTER VIEW ... RENAME TO</command> and <command>ALTER
1543 SEQUENCE ... RENAME TO</command> (David Fetter, Neil)
1547 Previously this could only be done via <command>ALTER TABLE ...
1548 RENAME TO</command>.
1554 Implement <command>CREATE TABLE LIKE ... INCLUDING
1555 INDEXES</command> (Trevor Hardcastle, Nikhil Sontakke, Neil)
1561 Allow <command>CREATE INDEX CONCURRENTLY</command> to ignore
1562 transactions in other databases (Simon)
1568 Make <command>CLUSTER</command> MVCC-safe (Heikki Linnakangas)
1572 Formerly, <command>CLUSTER</command> would discard all tuples
1573 that were committed dead, even if there were still transactions
1574 that should be able to see them under MVCC visibility rules.
1580 Add new <command>CLUSTER</command> syntax: <literal>CLUSTER
1581 <replaceable>table</> USING <replaceable>index</></literal>
1586 The old <command>CLUSTER</command> syntax is still supported, but
1587 the new form is considered more logical.
1593 Fix <command>EXPLAIN</command> so it can show complex plans
1594 more accurately (Tom)
1598 References to subplan outputs are now always shown correctly,
1599 instead of using <literal>?column<replaceable>N</>?</literal>
1600 for complicated cases.
1606 Make <command>CREATE/DROP/RENAME DATABASE</> wait briefly for
1607 conflicting backends to exit before failing (Tom)
1611 This increases the likelihood that these commands will succeed.
1617 Limit the amount of information reported when a user is dropped
1622 Previously, dropping (or attempting to drop) a user who owned many
1623 objects could result in large <literal>NOTICE</literal> or
1624 <literal>ERROR</literal> messages listing all these objects; this
1625 caused problems for some client applications. The length of the
1626 message is now limited, although a full list is still sent to the
1636 <title>Data Types</title>
1641 Support for the SQL/XML standard, including new operators and an
1642 <type>XML</type> data type (Nikolay Samokhvalov, Pavel Stehule, Peter)
1648 Enumerated data types (<type>ENUM</type>) (Tom Dunstan)
1652 This feature provides convenient support for fields that have a
1653 small, fixed set of allowed values. An example of creating an
1654 <literal>ENUM</> type is
1655 <literal>CREATE TYPE mood AS ENUM ('sad', 'ok', 'happy')</>.
1661 Universally Unique Identifier (<type>UUID</>) data type (Gevik
1666 This closely matches <acronym>RFC</> 4122.
1672 Widen the <type>MONEY</type> data type to 64 bits (D'Arcy Cain)
1676 This greatly increases the range of supported <type>MONEY</>
1683 Fix <type>float4</type>/<type>float8</type> to handle
1684 <literal>Infinity</> and <literal>NAN</> (Not A Number)
1685 consistently (Bruce)
1689 The code formerly was not consistent about distinguishing
1690 <literal>Infinity</> from overflow conditions.
1696 Allow leading and trailing whitespace during input of
1697 <type>boolean</type> values (Neil)
1703 Prevent <command>COPY</> from using digits and lowercase letters as
1713 <title>Functions</title>
1718 Add new regular expression functions
1719 <function>regexp_matches()</function>,
1720 <function>regexp_split_to_array()</function>, and
1721 <function>regexp_split_to_table()</function> (Jeremy Drake, Neil)
1725 These functions provide extraction of regular expression
1726 subexpressions and allow splitting a string using a POSIX regular
1733 Add <function>lo_truncate()</function> for large object truncation
1740 Implement <function>width_bucket()</function> for the <type>float8</>
1747 Add <function>pg_stat_clear_snapshot()</function> to discard
1748 statistics snapshots collected during the current transaction
1753 The first request for statistics in a transaction takes a statistics
1754 snapshot that does not change during the transaction. This function
1755 allows the snapshot to be discarded and a new snapshot loaded during
1756 the next statistics query. This is particularly useful for PL/PgSQL
1757 functions, which are confined to a single transaction.
1763 Add <literal>isodow</> option to <function>EXTRACT()</> and
1764 <function>date_part()</> (Bruce)
1768 This returns the day of the week, with Sunday as seven.
1769 (<literal>dow</> returns Sunday as zero.)
1775 Add <literal>ID</> (ISO day of week) and <literal>IDDD</> (ISO
1776 day of year) format codes for <function>to_char()</>,
1777 <function>to_date()</>, and <function>to_timestamp()</> (Brendan
1784 Make <function>to_timestamp()</> and <function>to_date()</>
1785 assume <literal>TM</literal> (trim) option for potentially
1786 variable-width fields (Bruce)
1790 This matches <productname>Oracle</>'s behavior.
1796 Fix off-by-one conversion error in
1797 <function>to_date()</function>/<function>to_timestamp()</function>
1798 <literal>D</> (non-ISO day of week) fields (Bruce)
1804 Make <function>setseed()</function> return void, rather than a
1805 useless integer value (Neil)
1811 Add a hash function for <type>NUMERIC</type> (Neil)
1815 This allows hash indexes and hash-based plans to be used with
1816 <type>NUMERIC</type> columns.
1822 Improve efficiency of
1823 <literal>LIKE</literal>/<literal>ILIKE</literal>, especially for
1824 multi-byte character sets like UTF-8 (Andrew, Itagaki Takahiro)
1830 Make <function>currtid()</function> functions require
1831 <literal>SELECT</literal> privileges on the target table (Tom)
1837 Add several <function>txid_*()</function> functions to query
1838 active transaction IDs (Jan)
1842 This is useful for various replication solutions.
1851 <title>PL/PgSQL Server-Side Language</title>
1856 Add scrollable cursor support, including directional control in
1857 <command>FETCH</command> (Pavel Stehule)
1863 Allow <literal>IN</literal> as an alternative to
1864 <literal>FROM</literal> in PL/PgSQL's <command>FETCH</command>
1865 statement, for consistency with the backend's
1866 <command>FETCH</command> command (Pavel Stehule)
1872 Add <command>MOVE</command> to PL/PgSQL (Magnus, Pavel Stehule,
1879 Implement <command>RETURN QUERY</command> (Pavel Stehule, Neil)
1883 This adds convenient syntax for PL/PgSQL set-returning functions
1884 that want to return the result of a query. <command>RETURN QUERY</>
1885 is easier and more efficient than a loop
1886 around <command>RETURN NEXT</command>.
1892 Allow function parameter names to be qualified with the
1893 function's name (Tom)
1897 For example, <literal>myfunc.myvar</>. This is particularly
1898 useful for specifying variables in a query where the variable
1899 name might match a column name.
1905 Make qualification of variables with block labels work properly (Tom)
1909 Formerly, outer-level block labels could unexpectedly interfere with
1910 recognition of inner-level record or row references.
1916 Tighten requirements for <literal>FOR</literal> loop
1917 <literal>STEP</> values (Tom)
1921 Prevent non-positive <literal>STEP</> values, and handle
1928 Improve accuracy when reporting syntax error locations (Tom)
1937 <title>Other Server-Side Languages</title>
1942 Allow type-name arguments to PL/Perl
1943 <function>spi_prepare()</function> to be data type aliases in
1944 addition to names found in <literal>pg_type</literal> (Andrew)
1950 Allow type-name arguments to PL/Python
1951 <function>plpy.prepare()</function> to be data type aliases in
1952 addition to names found in <literal>pg_type</literal> (Andrew)
1958 Allow type-name arguments to PL/Tcl <function>spi_prepare</> to
1959 be data type aliases in addition to names found in
1960 <literal>pg_type</literal> (Andrew)
1966 Enable PL/PythonU to compile on Python 2.5 (Marko Kreen)
1972 Support a true PL/Python boolean type in compatible Python versions
1973 (Python 2.3 and later) (Marko Kreen)
1979 Fix PL/Tcl problems with thread-enabled <filename>libtcl</> spawning
1980 multiple threads within the backend (Steve Marshall, Paul Bayer,
1985 This caused all sorts of unpleasantness.
1994 <title><link linkend="APP-PSQL"><application>psql</></link></title>
1999 List disabled triggers separately in <literal>\d</literal> output
2006 In <literal>\d</literal> patterns, always match <literal>$</literal>
2013 Show aggregate return types in <literal>\da</literal> output
2014 (Greg Sabino Mullane)
2020 Add the function's volatility status to the output of
2021 <literal>\df+</literal> (Neil)
2027 Add <literal>\prompt</literal> capability (Chad Wagner)
2033 Allow <literal>\pset</literal>, <literal>\t</literal>, and
2034 <literal>\x</literal> to specify <literal>on</> or <literal>off</>,
2035 rather than just toggling (Chad Wagner)
2041 Add <literal>\sleep</> capability (Jan)
2047 Enable <literal>\timing</> output for <literal>\copy</> (Andrew)
2053 Improve <literal>\timing</literal> resolution on Windows
2060 Flush <literal>\o</> output after each backslash command (Tom)
2066 Correctly detect and report errors while reading a <literal>-f</>
2073 Remove <literal>-u</> option (this option has long been depricated)
2083 <title><link linkend="APP-PGDUMP"><application>pg_dump</></link></title>
2088 Add <literal>--tablespaces-only</> and <literal>--roles-only</>
2089 options to <application>pg_dumpall</application> (Dave Page)
2095 Add an output file option to
2096 <application>pg_dumpall</application> (Dave Page)
2100 This is primarily useful on Windows, where output redirection of
2101 child <application>pg_dump</application> processes does not work.
2107 Allow <application>pg_dumpall</> to accept an initial-connection
2108 database name rather than the default
2109 <literal>template1</literal> (Dave Page)
2115 In <literal>-n</> and <literal>-t</> switches, always match
2116 <literal>$</literal> literally (Tom)
2122 Improve performance when a database has thousands of objects (Tom)
2128 Remove <literal>-u</> option (this option has long been depricated)
2138 <title>Other Client Applications</title>
2143 In <application>initdb</>, allow the location of the
2144 <filename>pg_xlog</filename> directory to be specified
2145 (Euler Taveira de Oliveira)
2151 Enable server core dump generation in <application>pg_regress</>
2152 on supported operating systems (Andrew)
2158 Add a <literal>-t</> (timeout) parameter to <application>pg_ctl</>
2163 This controls how long <application>pg_ctl</> will wait when waiting
2164 for server startup or shutdown. Formerly the timeout was hard-wired
2171 Add a <application>pg_ctl</> option to control generation
2172 of server core dumps (Andrew)
2178 Allow Control-C to cancel <application>clusterdb</>,
2179 <application>reindexdb</>, and <application>vacuumdb</> (Itagaki
2186 Suppress command tag output for <application>createdb</>,
2187 <application>createuser</>, <application>dropdb</>, and
2188 <application>dropuser</> (Peter)
2192 The <literal>--quiet</> option is ignored and will be removed in 8.4.
2193 Progress messages when acting on all databases now go to stdout
2194 instead of stderr because they are not actually errors.
2203 <title><link linkend="libpq"><application>libpq</></link></title>
2208 Interpret the <literal>dbName</> parameter of
2209 <function>PQsetdbLogin()</> as a <literal>conninfo</> string if
2210 it contains an equals sign (Andrew)
2214 This allows use of <literal>conninfo</> strings in client
2215 programs that still use <literal>PQsetdbLogin()</>.
2221 Support a global <acronym>SSL</> configuration file (Victor
2228 Add environment variable <varname>PGSSLKEY</> to control
2229 <acronym>SSL</> hardware keys (Victor Wagner)
2235 Add <function>lo_truncate()</function> for large object
2236 truncation (Kris Jurka)
2242 Add <function>PQconnectionNeedsPassword()</function> that returns
2243 true if the server required a password but none was supplied
2248 If this returns true after a failed connection attempt, a client
2249 application should prompt the user for a password. In the past
2250 applications have had to check for a specific error message string to
2251 decide whether a password is needed; that approach is now
2258 Add <function>PQconnectionUsedPassword()</function> that returns
2259 true if the supplied password was actually used
2264 This is useful in some security contexts where it is important
2265 to know whether a user-supplied password is actually valid.
2274 <title><link linkend="ecpg"><application>ecpg</></link></title>
2279 Use V3 frontend/backend protocol (Michael)
2283 This adds support for server-side prepared statements.
2289 Use native threads, instead of pthreads, on Windows (Magnus)
2295 Improve thread-safety of ecpglib (Itagaki Takahiro)
2301 Make the ecpg libraries export only necessary API symbols (Michael)
2310 <title><application>Windows</> Port</title>
2315 Allow the whole <productname>PostgreSQL</> distribution to be compiled
2316 with <productname>Microsoft Visual C++</> (Magnus and others)
2320 This allows Windows-based developers to use familiar development
2321 and debugging tools.
2322 Windows executables made with Visual C++ might also have better
2323 stability and performance than those made with other tool sets.
2324 The client-only Visual C++ build scripts have been removed.
2330 Drastically reduce postmaster's memory usage when it has many child
2337 Allow regression tests to be started by an administrative
2344 Add native shared memory implementation (Magnus)
2353 <title>Server Programming Interface (<acronym>SPI</>)</title>
2358 Add cursor-related functionality in SPI (Pavel Stehule)
2362 Allow access to the cursor-related planning options, and add
2363 <command>FETCH</>/<command>MOVE</> routines.
2369 Allow execution of cursor commands through
2370 <function>SPI_execute</function> (Tom)
2374 The macro <literal>SPI_ERROR_CURSOR</> still exists but will
2381 SPI plan pointers are now declared as <literal>SPIPlanPtr</> instead of
2382 <literal>void *</> (Tom)
2386 This does not break application code, but switching is
2387 recommended to help catch simple programming mistakes.
2396 <title>Build Options</title>
2401 Add <application>configure</> option <literal>--enable-profiling</>
2402 to enable code profiling (works only with <application>gcc</>)
2403 (Korry Douglas and Nikhil Sontakke)
2409 Add <application>configure</> option <literal>--with-system-tzdata</>
2410 to use the operating system's time zone database (Peter)
2416 Fix <acronym>PGXS</> so extensions can be built against PostgreSQL
2417 installations whose <application>pg_config</> program does not
2418 appear first in the <varname>PATH</> (Tom)
2424 Support <command>gmake draft</command> when building the
2425 <acronym>SGML</> documentation (Bruce)
2429 Unless <literal>draft</> is used, the documentation build will
2430 now be repeated if necessary to ensure the index is up-to-date.
2439 <title>Source Code</title>
2444 Rename macro <literal>DLLIMPORT</> to <literal>PGDLLIMPORT</> to
2445 avoid conflicting with third party includes (like Tcl) that
2446 define <literal>DLLIMPORT</> (Magnus)
2452 Create <quote>operator families</quote> to improve planning of
2453 queries involving cross-data-type comparisons (Tom)
2459 Update GIN <function>extractQuery()</> API to allow signalling
2460 that nothing can satisfy the query (Teodor)
2466 Move <literal>NAMEDATALEN</> definition from
2467 <filename>postgres_ext.h</> to <filename>pg_config_manual.h</>
2474 Provide <function>strlcpy()</function> and
2475 <function>strlcat()</function> on all platforms, and replace
2476 error-prone uses of <function>strncpy()</function>,
2477 <function>strncat()</function>, etc (Peter)
2483 Create hooks to let an external plugin monitor (or even replace) the
2484 planner and create plans for hypothetical situations (Gurjeet
2491 Create a function variable <literal>join_search_hook</> to let plugins
2492 override the join search order portion of the planner (Julius
2499 Add <function>tas()</> support for Renesas' M32R processor
2506 <function>quote_identifier()</function> and
2507 <application>pg_dump</application> no longer quote keywords that are
2508 unreserved according to the grammar (Tom)
2514 Change the on-disk representation of the <type>NUMERIC</type>
2515 data type so that the <structfield>sign_dscale</> word comes
2516 before the weight (Tom)
2522 Use <acronym>SYSV</> semaphores rather than POSIX on Darwin
2523 >= 6.0, i.e., OS X 10.2 and up (Chris Marcellino)
2529 Add <link linkend="acronyms">acronym</link> and <link
2530 linkend="creating-cluster-nfs">NFS</link> documentation
2537 "Postgres" is now documented as an accepted alias for
2538 "PostgreSQL" (Peter)
2544 Add documentation about preventing database server spoofing when
2545 the server is down (Bruce
2554 <title>Contrib</title>
2559 Move <filename>contrib</> <filename>README</> content into the
2560 main <productname>PostgreSQL</> documentation (Albert Cervera i
2567 Add <filename>contrib/pageinspect</filename> module for low-level
2568 page inspection (Simon, Heikki)
2574 Add <filename>contrib/pg_standby</filename> module for controlling
2575 warm standby operation (Simon)
2581 Add <filename>contrib/uuid-ossp</filename> module for generating
2582 <type>UUID</> values using the OSSP UUID library (Peter)
2586 Use <application>configure</>
2587 <literal>--with-ossp-uuid</literal> to activate. This takes
2588 advantage of the new <type>UUID</type> builtin type.
2594 Add <filename>contrib/dict_int</filename>,
2595 <filename>contrib/dict_xsyn</filename>, and
2596 <filename>contrib/test_parser</filename> modules to provide
2597 sample add-on text search dictionary templates and parsers
2604 Allow <application>contrib/pgbench</> to set the fillfactor (Pavan
2611 Add timestamps to <application>contrib/pgbench</> <literal>-l</>
2618 Add usage count statistics to
2619 <filename>contrib/pgbuffercache</filename> (Greg Smith)
2625 Add GIN support for <filename>contrib/hstore</> (Teodor)
2631 Add GIN support for <filename>contrib/pg_trgm</> (Guillaume Smet, Teodor)
2637 Update OS/X startup scripts in
2638 <filename>contrib/start-scripts</filename> (Mark Cotner, David
2645 Restrict <function>pgrowlocks()</function> and
2646 <function>dblink_get_pkey()</function> to users who have
2647 <literal>SELECT</literal> privilege on the target table (Tom)
2653 Restrict <filename>contrib/pgstattuple</filename> functions to
2660 <filename>contrib/xml2</filename> is deprecated and planned for
2661 removal in 8.4 (Peter)
2665 The new XML support in core PostgreSQL supersedes this module.
2675 <sect1 id="release-8-2-5">
2676 <title>Release 8.2.5</title>
2679 <title>Release date</title>
2680 <simpara>2007-09-17</simpara>
2684 This release contains a variety of fixes from 8.2.4.
2688 <title>Migration to Version 8.2.5</title>
2691 A dump/restore is not required for those running 8.2.X.
2697 <title>Changes</title>
2703 Prevent index corruption when a transaction inserts rows and
2704 then aborts close to the end of a concurrent <command>VACUUM</>
2705 on the same table (Tom)
2711 Fix <literal>ALTER DOMAIN ADD CONSTRAINT</> for cases involving
2712 domains over domains (Tom)
2718 Make <command>CREATE DOMAIN ... DEFAULT NULL</> work properly (Tom)
2724 Fix some planner problems with outer joins, notably poor
2725 size estimation for <literal>t1 LEFT JOIN t2 WHERE t2.col IS NULL</>
2732 Allow the <type>interval</> data type to accept input consisting only of
2733 milliseconds or microseconds (Neil)
2739 Allow timezone name to appear before the year in <type>timestamp</> input (Tom)
2745 Fixes for <acronym>GIN</> indexes used by <filename>/contrib/tsearch2</> (Teodor)
2751 Speed up rtree index insertion (Teodor)
2757 Fix excessive logging of <acronym>SSL</> error messages (Tom)
2763 Fix logging so that log messages are never interleaved when using
2764 the syslogger process (Andrew)
2770 Fix crash when <varname>log_min_error_statement</> logging runs out
2777 Fix incorrect handling of some foreign-key corner cases (Tom)
2783 Fix <function>stddev_pop(numeric)</> and <function>var_pop(numeric)</> (Tom)
2789 Prevent <command>REINDEX</> and <command>CLUSTER</> from failing
2790 due to attempting to process temporary tables of other sessions (Alvaro)
2796 Update the time zone database rules, particularly New Zealand's upcoming changes (Tom)
2802 Windows socket and semaphore improvements (Magnus)
2808 Make <command>pg_ctl -w</> work properly in Windows service mode (Dave Page)
2814 Fix memory allocation bug when using <application>MIT Kerberos</> on Windows (Magnus)
2820 Suppress timezone name (<literal>%Z</>) in log timestamps on Windows
2821 because of possible encoding mismatches (Tom)
2827 Require non-superusers who use <filename>/contrib/dblink</> to use only
2828 password authentication, as a security measure (Joe)
2834 Restrict <filename>/contrib/pgstattuple</> functions to superusers, for security reasons (Tom)
2840 Do not let <filename>/contrib/intarray</> try to make its GIN opclass
2841 the default (this caused problems at dump/restore) (Tom)
2850 <sect1 id="release-8-2-4">
2851 <title>Release 8.2.4</title>
2854 <title>Release date</title>
2855 <simpara>2007-04-23</simpara>
2859 This release contains a variety of fixes from 8.2.3,
2860 including a security fix.
2864 <title>Migration to Version 8.2.4</title>
2867 A dump/restore is not required for those running 8.2.X.
2873 <title>Changes</title>
2879 Support explicit placement of the temporary-table schema within
2880 <varname>search_path</>, and disable searching it for functions
2885 This is needed to allow a security-definer function to set a
2886 truly secure value of <varname>search_path</>. Without it,
2887 an unprivileged SQL user can use temporary objects to execute code
2888 with the privileges of the security-definer function (CVE-2007-2138).
2889 See <command>CREATE FUNCTION</> for more information.
2895 Fix <varname>shared_preload_libraries</> for Windows
2896 by forcing reload in each backend (Korry Douglas)
2902 Fix <function>to_char()</> so it properly upper/lower cases localized day or month
2903 names (Pavel Stehule)
2909 <filename>/contrib/tsearch2</> crash fixes (Teodor)
2915 Require <command>COMMIT PREPARED</> to be executed in the same
2916 database as the transaction was prepared in (Heikki)
2922 Allow <command>pg_dump</> to do binary backups larger than two gigabytes
2929 New traditional (Taiwan) Chinese <acronym>FAQ</> (Zhou Daojing)
2935 Prevent the statistics collector from writing to disk too frequently (Tom)
2941 Fix potential-data-corruption bug in how <command>VACUUM FULL</> handles
2942 <command>UPDATE</> chains (Tom, Pavan Deolasee)
2948 Fix bug in domains that use array types (Tom)
2954 Fix <command>pg_dump</> so it can dump a serial column's sequence
2955 using <option>-t</> when not also dumping the owning table
2962 Planner fixes, including improving outer join and bitmap scan
2963 selection logic (Tom)
2969 Fix possible wrong answers or crash when a PL/pgSQL function tries
2970 to <literal>RETURN</> from within an <literal>EXCEPTION</> block
2977 Fix PANIC during enlargement of a hash index (Tom)
2983 Fix POSIX-style timezone specs to follow new USA DST rules (Tom)
2992 <sect1 id="release-8-2-3">
2993 <title>Release 8.2.3</title>
2996 <title>Release date</title>
2997 <simpara>2007-02-07</simpara>
3001 This release contains two fixes from 8.2.2.
3005 <title>Migration to Version 8.2.3</title>
3008 A dump/restore is not required for those running 8.2.X.
3014 <title>Changes</title>
3020 Remove overly-restrictive check for type length in constraints and
3021 functional indexes(Tom)
3027 Fix optimization so MIN/MAX in subqueries can again use indexes (Tom)
3036 <sect1 id="release-8-2-2">
3037 <title>Release 8.2.2</title>
3040 <title>Release date</title>
3041 <simpara>2007-02-05</simpara>
3045 This release contains a variety of fixes from 8.2.1, including
3050 <title>Migration to Version 8.2.2</title>
3053 A dump/restore is not required for those running 8.2.X.
3059 <title>Changes</title>
3065 Remove security vulnerabilities that allowed connected users
3066 to read backend memory (Tom)
3070 The vulnerabilities involve suppressing the normal check that a SQL
3071 function returns the data type it's declared to, and changing the
3072 data type of a table column (CVE-2007-0555, CVE-2007-0556). These
3073 errors can easily be exploited to cause a backend crash, and in
3074 principle might be used to read database content that the user
3075 should not be able to access.
3081 Fix not-so-rare-anymore bug wherein btree index page splits could fail
3082 due to choosing an infeasible split point (Heikki Linnakangas)
3088 Fix Borland C compile scripts (L Bayuk)
3094 Properly handle <function>to_char('CC')</> for years ending in
3095 <literal>00</> (Tom)
3099 Year 2000 is in the twentieth century, not the twenty-first.
3105 <filename>/contrib/tsearch2</> localization improvements (Tatsuo, Teodor)
3111 Fix incorrect permission check in
3112 <literal>information_schema.key_column_usage</> view (Tom)
3116 The symptom is <quote>relation with OID nnnnn does not exist</> errors.
3117 To get this fix without using <command>initdb</>, use <command>CREATE OR
3118 REPLACE VIEW</> to install the corrected definition found in
3119 <filename>share/information_schema.sql</>. Note you will need to do
3120 this in each database.
3126 Improve <command>VACUUM</> performance for databases with many tables (Tom)
3132 Fix for rare Assert() crash triggered by <literal>UNION</> (Tom)
3138 Fix potentially incorrect results from index searches using
3139 <literal>ROW</> inequality conditions (Tom)
3145 Tighten security of multi-byte character processing for UTF8 sequences
3146 over three bytes long (Tom)
3152 Fix bogus <quote>permission denied</> failures occurring on Windows
3153 due to attempts to fsync already-deleted files (Magnus, Tom)
3159 Fix bug that could cause the statistics collector
3160 to hang on Windows (Magnus)
3164 This would in turn lead to autovacuum not working.
3170 Fix possible crashes when an already-in-use PL/pgSQL function is
3177 Improve PL/pgSQL handling of domain types (Sergiy Vyshnevetskiy, Tom)
3183 Fix possible errors in processing PL/pgSQL exception blocks (Tom)
3192 <sect1 id="release-8-2-1">
3193 <title>Release 8.2.1</title>
3196 <title>Release date</title>
3197 <simpara>2007-01-08</simpara>
3201 This release contains a variety of fixes from 8.2.
3205 <title>Migration to Version 8.2.1</title>
3208 A dump/restore is not required for those running 8.2.
3214 <title>Changes</title>
3220 Fix crash with <literal>SELECT</> ... <literal>LIMIT ALL</> (also
3221 <literal>LIMIT NULL</>) (Tom)
3227 <filename>Several /contrib/tsearch2</> fixes (Teodor)
3233 On Windows, make log messages coming from the operating system use
3234 <acronym>ASCII</> encoding (Hiroshi Saito)
3238 This fixes a conversion problem when there is a mismatch between
3239 the encoding of the operating system and database server.
3245 Fix Windows linking of <application>pg_dump</> using
3246 <filename>win32.mak</>
3253 Fix planner mistakes for outer join queries (Tom)
3259 Fix several problems in queries involving sub-SELECTs (Tom)
3265 Fix potential crash in SPI during subtransaction abort (Tom)
3269 This affects all PL functions since they all use SPI.
3275 Improve build speed of <acronym>PDF</> documentation (Peter)
3281 Re-add <acronym>JST</> (Japan) timezone abbreviation (Tom)
3287 Improve optimization decisions related to index scans (Tom)
3293 Have <application>psql</> print multi-byte combining characters as
3294 before, rather than output as <literal>\u</> (Tom)
3300 Improve index usage of regular expressions that use parentheses (Tom)
3304 This improves <application>psql</> <literal>\d</> performance also.
3310 Make <application>pg_dumpall</> assume that databases have public
3311 <literal>CONNECT</> privilege, when dumping from a pre-8.2 server (Tom)
3315 This preserves the previous behavior that anyone can connect to a
3316 database if allowed by <filename>pg_hba.conf</>.
3325 <sect1 id="release-8-2">
3326 <title>Release 8.2</title>
3329 <title>Release date</title>
3330 <simpara>2006-12-05</simpara>
3334 <title>Overview</title>
3337 This release adds many functionality and performance improvements that
3338 were requested by users, including:
3344 Query language enhancements including <command>INSERT/UPDATE/DELETE
3345 RETURNING</command>, multirow <literal>VALUES</literal> lists, and
3346 optional target-table alias in
3347 <command>UPDATE</>/<command>DELETE</command>
3353 Index creation without blocking concurrent
3354 <command>INSERT</>/<command>UPDATE</>/<command>DELETE</>
3361 Many query optimization improvements, including support for
3362 reordering outer joins
3368 Improved sorting performance with lower memory usage
3374 More efficient locking with better concurrency
3380 More efficient vacuuming
3386 Easier administration of warm standby servers
3392 New <literal>FILLFACTOR</literal> support for tables and indexes
3398 Monitoring, logging, and performance tuning additions
3404 More control over creating and dropping objects
3410 Table inheritance relationships can be defined
3411 for and removed from pre-existing tables
3417 <command>COPY TO</command> can copy the output of an arbitrary
3418 <command>SELECT</command> statement
3424 Array improvements, including nulls in arrays
3430 Aggregate-function improvements, including multiple-input
3431 aggregates and SQL:2003 statistical functions
3437 Many <filename>contrib/</filename> improvements
3448 <title>Migration to Version 8.2</title>
3451 A dump/restore using <application>pg_dump</application> is
3452 required for those wishing to migrate data from any previous
3457 Observe the following incompatibilities:
3465 linkend="guc-escape-string-warning"><varname>escape_string_warning</></link>
3466 to <literal>on</> by default (Bruce)
3470 This issues a warning if backslash escapes are used in
3471 <link linkend="sql-syntax-strings">non-escape (non-<literal>E''</>)
3478 Change the <link linkend="sql-syntax-row-constructors">row
3479 constructor syntax</link> (<literal>ROW(...)</>) so that
3480 list elements <literal>foo.*</> will be expanded to a list
3481 of their member fields, rather than creating a nested
3482 row type field as formerly (Tom)
3486 The new behavior is substantially more useful since it
3487 allows, for example, triggers to check for data changes
3488 with <literal>IF row(new.*) IS DISTINCT FROM row(old.*)</>.
3489 The old behavior is still available by omitting <literal>.*</>.
3495 Make <link linkend="row-wise-comparison">row comparisons</link>
3496 follow <acronym>SQL</> standard semantics and allow them
3497 to be used in index scans (Tom)
3501 Previously, row = and <> comparisons followed the
3502 standard but < <= > >= did not. A row comparison
3503 can now be used as an index constraint for a multicolumn
3504 index matching the row value.
3510 Make <link linkend="functions-comparison">row <literal>IS <optional>NOT</> NULL</literal></link>
3511 tests follow <acronym>SQL</> standard semantics (Tom)
3515 The former behavior conformed to the standard for simple cases
3516 with <literal>IS NULL</>, but <literal>IS NOT NULL</> would return
3517 true if any row field was non-null, whereas the standard says it
3518 should return true only when all fields are non-null.
3524 Make <link linkend="SQL-SET-CONSTRAINTS"><command>SET
3525 CONSTRAINT</></link> affect only one constraint (Kris Jurka)
3529 In previous releases, <command>SET CONSTRAINT</> modified
3530 all constraints with a matching name. In this release,
3531 the schema search path is used to modify only the first
3532 matching constraint. A schema specification is also
3533 supported. This more nearly conforms to the SQL standard.
3539 Remove <literal>RULE</> permission for tables, for security reasons
3544 As of this release, only a table's owner can create or modify
3545 rules for the table. For backwards compatibility,
3546 <command>GRANT</>/<command>REVOKE RULE</> is still accepted,
3547 but it does nothing.
3553 Array comparison improvements (Tom)
3557 Now array dimensions are also compared.
3563 Change <link linkend="functions-array">array concatenation</link>
3564 to match documented behavior (Tom)
3568 This changes the previous behavior where concatenation
3569 would modify the array lower bound.
3575 Make command-line options of <application>postmaster</>
3576 and <link linkend="app-postgres"><application>postgres</></link>
3581 This allows the postmaster to pass arguments to each backend
3582 without using <literal>-o</>. Note that some options are now
3583 only available as long-form options, because there were conflicting
3584 single-letter options.
3590 Deprecate use of <application>postmaster</> symbolic link (Peter)
3594 <application>postmaster</> and <application>postgres</>
3595 commands now act identically, with the behavior determined
3596 by command-line options. The <application>postmaster</> symbolic link is
3597 kept for compatibility, but is not really needed.
3604 linkend="guc-log-duration"><varname>log_duration</></link>
3605 to output even if the query is not output (Tom)
3609 In prior releases, <varname>log_duration</> only printed if
3610 the query appeared earlier in the log.
3617 linkend="functions-formatting"><function>to_char(time)</></link>
3619 linkend="functions-formatting"><function>to_char(interval)</></link>
3620 treat <literal>HH</> and <literal>HH12</> as 12-hour
3625 Most applications should use <literal>HH24</> unless they
3626 want a 12-hour display.
3632 Zero unmasked bits in conversion from <link
3633 linkend="datatype-inet"><type>INET</></link> to <link
3634 linkend="datatype-inet"><type>CIDR</></link> (Tom)
3638 This ensures that the converted value is actually valid for
3645 Remove <varname>australian_timezones</> configuration variable
3650 This variable has been superseded by a more general facility
3651 for configuring timezone abbreviations.
3657 Improve cost estimation for nested-loop index scans (Tom)
3661 This might eliminate the need to set unrealistically small
3663 linkend="guc-random-page-cost"><varname>random_page_cost</></link>.
3664 If you have been using a very small <varname>random_page_cost</>,
3665 please recheck your test cases.
3671 Change behavior of <command>pg_dump</> <literal>-n</> and
3672 <literal>-t</> options. (Greg Sabino Mullane)
3675 See the <command>pg_dump</> manual page for details.
3681 Change <link linkend="libpq"><application>libpq</></link>
3682 <function>PQdsplen()</> to return a useful value (Martijn
3689 Declare <link linkend="libpq"><application>libpq</></link>
3690 <function>PQgetssl()</> as returning <literal>void *</>,
3691 rather than <literal>SSL *</> (Martijn van Oosterhout)
3695 This allows applications to use the function without including
3696 the OpenSSL headers.
3702 C-language loadable modules must now include a
3703 <link linkend="xfunc-c-dynload"><literal>PG_MODULE_MAGIC</></link>
3704 macro call for version compatibility checking
3705 (Martijn van Oosterhout)
3711 For security's sake, modules used by a PL/PerlU function are no
3712 longer available to PL/Perl functions (Andrew)
3716 This also implies that data can no longer be shared between a PL/Perl
3717 function and a PL/PerlU function.
3718 Some Perl installations have not been compiled with the correct flags
3719 to allow multiple interpreters to exist within a single process.
3720 In this situation PL/Perl and PL/PerlU cannot both be used in a
3721 single backend. The solution is to get a Perl installation which
3722 supports multiple interpreters.
3729 In <filename>contrib/xml2/</>, rename <function>xml_valid()</> to
3730 <function>xml_is_well_formed()</> (Tom)
3734 <function>xml_valid()</> will remain for backward compatibility,
3735 but its behavior will change to do schema checking in a future
3742 Remove <filename>contrib/ora2pg/</>, now at <ulink
3743 url="http://www.samse.fr/GPL/ora2pg"></ulink>
3749 Remove contrib modules that have been migrated to PgFoundry:
3750 <filename>adddepend</>, <filename>dbase</>, <filename>dbmirror</>,
3751 <filename>fulltextindex</>, <filename>mac</>, <filename>userlock</>
3757 Remove abandoned contrib modules:
3758 <filename>mSQL-interface</>, <filename>tips</>
3764 Remove <acronym>QNX</> and <acronym>BEOS</> ports (Bruce)
3768 These ports no longer had active maintainers.
3776 <title>Changes</title>
3779 Below you will find a detailed account of the
3780 changes between <productname>PostgreSQL</productname> 8.2 and
3781 the previous major release.
3785 <title>Performance Improvements</title>
3790 Allow the planner to reorder <link linkend="queries-join">outer
3791 joins</link> in some circumstances (Tom)
3795 In previous releases, outer joins would always be evaluated in
3796 the order written in the query. This change allows the
3797 query optimizer to consider reordering outer joins, in cases where
3798 it can determine that the join order can be changed without
3799 altering the meaning of the query. This can make a
3800 considerable performance difference for queries involving
3801 multiple outer joins or mixed inner and outer joins.
3807 Improve efficiency of <link
3808 linkend="functions-comparisons"><literal>IN</>
3809 (list-of-expressions)</link> clauses (Tom)
3815 Improve sorting speed and reduce memory usage (Simon, Tom)
3821 Improve subtransaction performance (Alvaro, Itagaki Takahiro,
3828 Add <literal>FILLFACTOR</> to <link
3829 linkend="SQL-CREATETABLE">table</link> and <link
3830 linkend="SQL-CREATEINDEX">index</link> creation (ITAGAKI
3835 This leaves extra free space in each table or index page,
3836 allowing improved performance as the database grows. This
3837 is particularly valuable to maintain clustering.
3843 Increase default values for <link
3844 linkend="guc-shared-buffers"><varname>shared_buffers</></link>
3846 linkend="guc-max-fsm-pages"><varname>max_fsm_pages</></link>
3853 Improve locking performance by breaking the lock manager tables into
3859 This allows locking to be more fine-grained, reducing
3866 Reduce locking requirements of sequential scans (Qingqing
3873 Reduce locking required for database creation and destruction
3880 Improve the optimizer's selectivity estimates for <link
3881 linkend="functions-like"><literal>LIKE</></link>, <link
3882 linkend="functions-like"><literal>ILIKE</></link>, and
3883 <link linkend="functions-posix-regexp">regular expression</link>
3890 Improve planning of joins to <link linkend="ddl-inherit">inherited
3891 tables</link> and <link linkend="queries-union"><literal>UNION
3892 ALL</></link> views (Tom)
3898 Allow <link linkend="guc-constraint-exclusion">constraint
3899 exclusion</link> to be applied to <link
3900 linkend="ddl-inherit">inherited</link> <command>UPDATE</> and
3901 <command>DELETE</> queries (Tom)
3905 <command>SELECT</> already honored constraint exclusion.
3911 Improve planning of constant <literal>WHERE</> clauses, such as
3912 a condition that depends only on variables inherited from an
3913 outer query level (Tom)
3919 Protocol-level unnamed prepared statements are re-planned
3920 for each set of <literal>BIND</> values (Tom)
3924 This improves performance because the exact parameter values
3925 can be used in the plan.
3931 Speed up vacuuming of B-Tree indexes (Heikki Linnakangas,
3938 Avoid extra scan of tables without indexes during <link
3939 linkend="SQL-VACUUM"><command>VACUUM</></link> (Greg Stark)
3945 Improve multicolumn <link linkend="GiST"><acronym>GiST</></link>
3946 indexing (Oleg, Teodor)
3952 Remove dead index entries before B-Tree page split (Junji
3962 <title>Server Changes</title>
3967 Allow a forced switch to a new transaction log file (Simon, Tom)
3971 This is valuable for keeping warm standby slave servers
3972 in sync with the master. Transaction log file switching now also happens
3973 automatically during <link
3974 linkend="functions-admin"><function>pg_stop_backup()</></link>.
3975 This ensures that all
3976 transaction log files needed for recovery can be archived immediately.
3982 Add <acronym>WAL</> informational functions (Simon)
3986 Add functions for interrogating the current transaction log insertion
3987 point and determining <acronym>WAL</> filenames from the
3988 hex <acronym>WAL</> locations displayed by <link
3989 linkend="functions-admin"><function>pg_stop_backup()</></link>
3990 and related functions.
3996 Improve recovery from a crash during <acronym>WAL</> replay (Simon)
4000 The server now does periodic checkpoints during <acronym>WAL</>
4001 recovery, so if there is a crash, future <acronym>WAL</>
4002 recovery is shortened. This also eliminates the need for
4003 warm standby servers to replay the entire log since the
4004 base backup if they crash.
4010 Improve reliability of long-term <acronym>WAL</> replay
4011 (Heikki, Simon, Tom)
4015 Formerly, trying to roll forward through more than 2 billion
4016 transactions would not work due to XID wraparound. This meant
4017 warm standby servers had to be reloaded
4018 from fresh base backups periodically.
4025 linkend="guc-archive-timeout"><varname>archive_timeout</></link>
4026 to force transaction log file switches at a given interval (Simon)
4030 This enforces a maximum replication delay for warm standby servers.
4036 Add native <link linkend="auth-ldap"><acronym>LDAP</></link>
4037 authentication (Magnus Hagander)
4041 This is particularly useful for platforms that do not
4042 support <acronym>PAM</>, such as Windows.
4048 Add <link linkend="sql-grant-description-objects"><literal>GRANT
4049 CONNECT ON DATABASE</></link> (Gevik Babakhani)
4053 This gives SQL-level control over database access. It works as
4054 an additional filter on top of the existing
4055 <link linkend="auth-pg-hba-conf"><filename>pg_hba.conf</></link>
4062 Add support for <link linkend="ssl-tcp"><acronym>SSL</>
4063 Certificate Revocation List</link> (<acronym>CRL</>) files
4064 (Libor Hohoš)
4068 The server and <application>libpq</> both recognize <acronym>CRL</>
4075 <link linkend="GiST"><acronym>GiST</></link> indexes are
4076 now clusterable (Teodor)
4082 Remove routine autovacuum server log entries (Bruce)
4087 linkend="monitoring-stats-views-table"><literal>pg_stat_activity</></link>
4088 now shows autovacuum activity.
4094 Track maximum XID age within individual tables, instead of whole databases (Alvaro)
4098 This reduces the overhead involved in preventing transaction
4099 ID wraparound, by avoiding unnecessary VACUUMs.
4105 Add last vacuum and analyze timestamp columns to the stats
4106 collector (Larry Rosenman)
4110 These values now appear in the <link
4111 linkend="monitoring-stats-views-table"><literal>pg_stat_*_tables</></link>
4118 Improve performance of statistics monitoring, especially
4119 <varname>stats_command_string</>
4124 This release enables <varname>stats_command_string</> by
4125 default, now that its overhead is minimal. This means
4127 linkend="monitoring-stats-views-table"><literal>pg_stat_activity</></link>
4128 will now show all active queries by default.
4134 Add a <literal>waiting</> column to <link
4135 linkend="monitoring-stats-views-table"><literal>pg_stat_activity</></link>
4140 This allows <structname>pg_stat_activity</> to show all the
4141 information included in the <application>ps</> display.
4147 Add configuration parameter <link
4148 linkend="guc-update-process-title"><varname>update_process_title</></link>
4149 to control whether the <application>ps</> display is updated
4150 for every command (Bruce)
4154 On platforms where it is expensive to update the <application>ps</>
4155 display, it might be worthwhile to turn this off and rely solely on
4156 <structname>pg_stat_activity</> for status information.
4162 Allow units to be specified in configuration settings
4167 For example, you can now set <link
4168 linkend="guc-shared-buffers"><varname>shared_buffers</></link>
4169 to <literal>32MB</> rather than mentally converting sizes.
4175 Add support for <link linkend="config-setting">include
4176 directives</link> in <filename>postgresql.conf</> (Joachim
4183 Improve logging of protocol-level prepare/bind/execute
4184 messages (Bruce, Tom)
4188 Such logging now shows statement names, bind parameter
4189 values, and the text of the query being executed. Also,
4190 the query text is properly included in logged error messages
4191 when enabled by <varname>log_min_error_statement</>.
4198 linkend="guc-max-stack-depth"><varname>max_stack_depth</></link>
4199 from being set to unsafe values
4203 On platforms where we can determine the actual kernel stack depth
4204 limit (which is most), make sure that the initial default value of
4205 <varname>max_stack_depth</> is safe, and reject attempts to set it
4206 to unsafely large values.
4212 Enable highlighting of error location in query in more
4217 The server is now able to report a specific error location for
4218 some semantic errors (such as unrecognized column name), rather
4219 than just for basic syntax errors as before.
4225 Fix <quote>failed to re-find parent key</> errors in
4226 <command>VACUUM</> (Tom)
4232 Clean out <filename>pg_internal.init</> cache files during server
4237 This avoids a hazard that the cache files might contain stale
4238 data after PITR recovery.
4244 Fix race condition for truncation of a large relation across a
4245 gigabyte boundary by <command>VACUUM</> (Tom)
4251 Fix bug causing needless deadlock errors on row-level locks (Tom)
4257 Fix bugs affecting multi-gigabyte hash indexes (Tom)
4263 Each backend process is now its own process group leader (Tom)
4267 This allows query cancel to abort subprocesses invoked from a
4268 backend or archive/recovery process.
4277 <title>Query Changes</title>
4282 Add <link linkend="SQL-INSERT"><command>INSERT</></link>/<link
4283 linkend="SQL-UPDATE"><command>UPDATE</></link>/<link
4284 linkend="SQL-DELETE"><command>DELETE</></link>
4285 <literal>RETURNING</> (Jonah Harris, Tom)
4289 This allows these commands to return values, such as the
4290 computed serial key for a new row. In the <command>UPDATE</>
4291 case, values from the updated version of the row are returned.
4297 Add support for multiple-row <link
4298 linkend="queries-values"><literal>VALUES</></link> clauses,
4299 per SQL standard (Joe, Tom)
4303 This allows <command>INSERT</> to insert multiple rows of
4304 constants, or queries to generate result sets using constants.
4305 For example, <literal>INSERT ... VALUES (...), (...),
4306 ....</>, and <literal>SELECT * FROM (VALUES (...), (...),
4307 ....) AS alias(f1, ...)</>.
4313 Allow <link linkend="SQL-UPDATE"><command>UPDATE</></link>
4314 and <link linkend="SQL-DELETE"><command>DELETE</></link>
4315 to use an alias for the target table (Atsushi Ogawa)
4319 The SQL standard does not permit an alias in these commands, but
4320 many database systems allow one anyway for notational convenience.
4326 Allow <link linkend="SQL-UPDATE"><command>UPDATE</></link>
4327 to set multiple columns with a list of values (Susanne
4332 This is basically a short-hand for assigning the columns
4333 and values in pairs. The syntax is <literal>UPDATE tab
4334 SET (<replaceable>column</>, ...) = (<replaceable>val</>, ...)</>.
4340 Make row comparisons work per standard (Tom)
4344 The forms <, <=, >, >= now compare rows lexicographically,
4345 that is, compare the first elements, if equal compare the second
4346 elements, and so on. Formerly they expanded to an AND condition
4347 across all the elements, which was neither standard nor very useful.
4353 Add <link linkend="SQL-TRUNCATE"><literal>CASCADE</></link>
4354 option to <command>TRUNCATE</> (Joachim Wieland)
4358 This causes <command>TRUNCATE</> to automatically include all tables
4359 that reference the specified table(s) via foreign keys. While
4360 convenient, this is a dangerous tool — use with caution!
4366 Support <literal>FOR UPDATE</> and <literal>FOR SHARE</>
4367 in the same <link linkend="SQL-INSERT"><literal>SELECT</></link>
4374 Add <link linkend="functions-comparisons"><literal>IS NOT
4375 DISTINCT FROM</></link> (Pavel Stehule)
4379 This operator is similar to equality (<literal>=</>), but
4380 evaluates to true when both left and right operands are
4381 <literal>NULL</>, and to false when just one is, rather than
4382 yielding <literal>NULL</> in these cases.
4388 Improve the length output used by <link
4389 linkend="queries-union"><literal>UNION</></link>/<literal>INTERSECT</>/<literal>EXCEPT</>
4394 When all corresponding columns are of the same defined length, that
4395 length is used for the result, rather than a generic length.
4401 Allow <link linkend="functions-like"><literal>ILIKE</></link>
4402 to work for multi-byte encodings (Tom)
4406 Internally, <literal>ILIKE</> now calls <function>lower()</>
4407 and then uses <literal>LIKE</>. Locale-specific regular
4408 expression patterns still do not work in these encodings.
4415 linkend="guc-standard-conforming-strings"><varname>standard_conforming_strings</></link>
4416 to be turned <literal>on</> (Kevin Grittner)
4420 This allows backslash escaping in strings to be disabled,
4421 making <productname>PostgreSQL</> more
4422 standards-compliant. The default is <literal>off</> for backwards
4423 compatibility, but future releases will default this to <literal>on</>.
4429 Do not flatten subqueries that contain <literal>volatile</>
4430 functions in their target lists (Jaime Casanova)
4434 This prevents surprising behavior due to multiple evaluation
4435 of a <literal>volatile</> function (such as <function>random()</>
4436 or <function>nextval()</>). It might cause performance
4437 degradation in the presence of functions that are unnecessarily
4438 marked as <literal>volatile</>.
4444 Add system views <link
4445 linkend="view-pg-prepared-statements"><literal>pg_prepared_statements</></link>
4447 linkend="view-pg-cursors"><literal>pg_cursors</></link>
4448 to show prepared statements and open cursors (Joachim Wieland, Neil)
4452 These are very useful in pooled connection setups.
4458 Support portal parameters in <link
4459 linkend="SQL-EXPLAIN"><command>EXPLAIN</></link> and <link
4460 linkend="SQL-EXECUTE"><command>EXECUTE</></link> (Tom)
4464 This allows, for example, <acronym>JDBC</> <literal>?</> parameters to
4465 work in these commands.
4471 If <acronym>SQL</>-level <link
4472 linkend="SQL-PREPARE"><command>PREPARE</></link> parameters
4473 are unspecified, infer their types from the content of the
4478 Protocol-level <command>PREPARE</> already did this.
4484 Allow <literal>LIMIT</> and <literal>OFFSET</> to exceed
4485 two billion (Dhanaraj M)
4494 <title>Object Manipulation Changes</title>
4499 Add <literal>TABLESPACE</> clause to <link
4500 linkend="SQL-CREATETABLEAS"><command>CREATE TABLE AS</></link>
4505 This allows a tablespace to be specified for the new table.
4511 Add <literal>ON COMMIT</> clause to <link
4512 linkend="SQL-CREATETABLEAS"><command>CREATE TABLE AS</></link>
4517 This allows temporary tables to be truncated or dropped on
4518 transaction commit. The default behavior is for the table
4519 to remain until the session ends.
4525 Add <literal>INCLUDING CONSTRAINTS</> to <link
4526 linkend="SQL-CREATETABLE"><command>CREATE TABLE LIKE</></link>
4531 This allows easy copying of <literal>CHECK</> constraints to a new
4538 Allow the creation of placeholder (shell) <link
4539 linkend="SQL-CREATETYPE">types</link> (Martijn van Oosterhout)
4543 A shell type declaration creates a type name, without specifying
4544 any of the details of the type. Making a shell type is useful
4545 because it allows cleaner declaration of the type's input/output
4546 functions, which must exist before the type can be defined <quote>for
4547 real</>. The syntax is <command>CREATE TYPE <replaceable
4548 class="parameter">typename</replaceable></>.
4554 <link linkend="SQL-CREATEAGGREGATE">Aggregate functions</link>
4555 now support multiple input parameters (Sergey Koposov, Tom)
4561 Add new aggregate creation <link
4562 linkend="SQL-CREATEAGGREGATE">syntax</link> (Tom)
4566 The new syntax is <command>CREATE AGGREGATE
4567 <replaceable>aggname</> (<replaceable>input_type</>)
4568 (<replaceable>parameter_list</>)</command>. This more
4569 naturally supports the new multi-parameter aggregate
4570 functionality. The previous syntax is still supported.
4577 linkend="SQL-ALTERROLE"><command>ALTER ROLE PASSWORD NULL</></link>
4578 to remove a previously set role password (Peter)
4584 Add <command>DROP</> object <literal>IF EXISTS</> for many
4585 object types (Andrew)
4589 This allows <command>DROP</> operations on non-existent
4590 objects without generating an error.
4596 Add <link linkend="SQL-DROP-OWNED"><literal>DROP OWNED</></link>
4597 to drop all objects owned by a role (Alvaro)
4603 Add <link linkend="SQL-REASSIGN-OWNED"><literal>REASSIGN
4604 OWNED</></link> to reassign ownership of all objects owned
4609 This, and <literal>DROP OWNED</> above, facilitate dropping
4616 Add <link linkend="SQL-GRANT"><command>GRANT ON SEQUENCE</></link>
4621 This was added for setting sequence-specific permissions.
4622 <literal>GRANT ON TABLE</> for sequences is still supported
4623 for backward compatibility.
4629 Add <link linkend="SQL-GRANT"><literal>USAGE</></link>
4630 permission for sequences that allows only <function>currval()</>
4631 and <function>nextval()</>, not <function>setval()</>
4636 <literal>USAGE</> permission allows more fine-grained
4637 control over sequence access. Granting <literal>USAGE</>
4638 allows users to increment
4639 a sequence, but prevents them from setting the sequence to
4640 an arbitrary value using <function>setval()</>.
4646 Add <link linkend="SQL-ALTERTABLE"><literal>ALTER TABLE
4647 [ NO ] INHERIT</></link> (Greg Stark)
4651 This allows inheritance to be adjusted dynamically, rather than
4652 just at table creation and destruction. This is very valuable
4653 when using inheritance to implement table partitioning.
4659 Allow <link linkend="SQL-COMMENT">comments</link> on global
4660 objects to be stored globally (Kris Jurka)
4664 Previously, comments attached to databases were stored in individual
4665 databases, making them ineffective, and there was no provision
4666 at all for comments on roles or tablespaces. This change adds a new
4667 shared catalog <link
4668 linkend="catalog-pg-shdescription"><structname>pg_shdescription</structname></link>
4669 and stores comments on databases, roles, and tablespaces therein.
4678 <title>Utility Command Changes</title>
4683 Add option to allow indexes to be created without blocking
4684 concurrent writes to the table (Greg Stark, Tom)
4688 The new syntax is <link linkend="SQL-CREATEINDEX"><command>CREATE
4689 INDEX CONCURRENTLY</></link>. The default behavior is
4690 still to block table modification while a index is being
4697 Provide <link linkend="functions-advisory-locks">advisory
4698 locking</link> functionality (Abhijit Menon-Sen, Tom)
4702 This is a new locking API designed to replace what used to be
4703 in /contrib/userlock. The userlock code is now on pgfoundry.
4709 Allow <link linkend="SQL-COPY"><command>COPY</></link> to
4710 dump a <command>SELECT</> query (Zoltan Boszormenyi, Karel
4715 This allows <command>COPY</> to dump arbitrary <acronym>SQL</>
4716 queries. The syntax is <literal>COPY (SELECT ...) TO</>.
4722 Make the <link linkend="SQL-COPY"><command>COPY</></link>
4723 command return a command tag that includes the number of
4724 rows copied (Volkan YAZICI)
4730 Allow <link linkend="SQL-VACUUM"><command>VACUUM</></link>
4731 to expire rows without being affected by other concurrent
4732 <command>VACUUM</> operations (Hannu Krossing, Alvaro, Tom)
4738 Make <link linkend="APP-INITDB"><application>initdb</></link>
4739 detect the operating system locale and set the default
4740 <varname>DateStyle</> accordingly (Peter)
4744 This makes it more likely that the installed
4745 <filename>postgresql.conf</> <varname>DateStyle</> value will
4752 Reduce number of progress messages displayed by <application>initdb</> (Tom)
4761 <title>Date/Time Changes</title>
4766 Allow full timezone names in <link
4767 linkend="datatype-datetime"><type>timestamp</></link> input values
4772 For example, <literal>'2006-05-24 21:11
4773 America/New_York'::timestamptz</>.
4779 Support configurable timezone abbreviations (Joachim Wieland)
4783 A desired set of timezone abbreviations can be chosen via the
4784 configuration parameter <link
4785 linkend="guc-timezone-abbreviations"><varname>timezone_abbreviations</></link>.
4792 linkend="view-pg-timezone-abbrevs"><varname>pg_timezone_abbrevs</></link>
4794 linkend="view-pg-timezone-names"><varname>pg_timezone_names</></link>
4795 views to show supported timezones (Magnus Hagander)
4802 linkend="functions-datetime-table"><function>clock_timestamp()</></link>,
4804 linkend="functions-datetime-table"><function>statement_timestamp()</></link>,
4806 linkend="functions-datetime-table"><function>transaction_timestamp()</></link>
4811 <function>clock_timestamp()</> is the current wall-clock time,
4812 <function>statement_timestamp()</> is the time the current
4813 statement arrived at the server, and
4814 <function>transaction_timestamp()</> is an alias for
4822 linkend="functions-formatting"><function>to_char()</></link>
4823 to print localized month and day names (Euler Taveira de
4831 linkend="functions-formatting"><function>to_char(time)</></link>
4833 linkend="functions-formatting"><function>to_char(interval)</></link>
4834 to output <acronym>AM</>/<acronym>PM</> specifications
4839 Intervals and times are treated as 24-hour periods, e.g.
4840 <literal>25 hours</> is considered <acronym>AM</>.
4846 Add new function <link
4847 linkend="functions-datetime-table"><function>justify_interval()</></link>
4848 to adjust interval units (Mark Dilger)
4854 Allow timezone offsets up to 14:59 away from GMT
4858 Kiribati uses GMT+14, so we'd better accept that.
4864 Interval computation improvements (Michael Glaesemann, Bruce)
4873 <title>Other Data Type and Function Changes</title>
4878 Allow arrays to contain <literal>NULL</> elements (Tom)
4884 Allow assignment to array elements not contiguous with the existing
4889 The intervening array positions will be filled with nulls.
4890 This is per SQL standard.
4896 New built-in <link linkend="functions-array">operators</link>
4897 for array-subset comparisons (<literal>@></>,
4898 <literal><@</>, <literal>&&</>) (Teodor, Tom)
4902 These operators can be indexed for many data types using
4903 <acronym>GiST</> or <acronym>GIN</> indexes.
4909 Add convenient arithmetic <link
4910 linkend="cidr-inet-operators-table">operations</link> on
4911 <type>INET</>/<type>CIDR</> values (Stephen R. van den
4916 The new operators are <literal>&</> (and), <literal>|</>
4917 (or), <literal>~</> (not), <type>inet</> <literal>+</> <type>int8</>,
4918 <type>inet</> <literal>-</> <type>int8</>, and
4919 <type>inet</> <literal>-</> <type>inet</>.
4926 linkend="functions-aggregate-statistics-table">aggregate functions</link>
4927 from SQL:2003 (Neil)
4931 The new functions are <function>var_pop()</>,
4932 <function>var_samp()</>, <function>stddev_pop()</>, and
4933 <function>stddev_samp()</>. <function>var_samp()</> and
4934 <function>stddev_samp()</> are merely renamings of the
4935 existing aggregates <function>variance()</> and
4936 <function>stddev()</>. The latter names remain available
4937 for backward compatibility.
4943 Add SQL:2003 statistical <link
4944 linkend="functions-aggregate-statistics-table">aggregates</link>
4949 New functions: <function>regr_intercept()</>,
4950 <function>regr_slope()</>, <function>regr_r2()</>,
4951 <function>corr()</>, <function>covar_samp()</>,
4952 <function>covar_pop()</>, <function>regr_avgx()</>,
4953 <function>regr_avgy()</>, <function>regr_sxy()</>,
4954 <function>regr_sxx()</>, <function>regr_syy()</>,
4955 <function>regr_count()</>.
4961 Allow <link linkend="SQL-CREATEDOMAIN">domains</link> to be
4962 based on other domains (Tom)
4968 Properly enforce domain <link
4969 linkend="ddl-constraints"><literal>CHECK</></link> constraints
4970 everywhere (Neil, Tom)
4974 For example, the result of a user-defined function that is
4975 declared to return a domain type is now checked against the
4976 domain's constraints. This closes a significant hole in the domain
4983 Fix problems with dumping renamed <link
4984 linkend="datatype-serial"><type>SERIAL</></link> columns
4989 The fix is to dump a <type>SERIAL</> column by explicitly
4990 specifying its <literal>DEFAULT</> and sequence elements,
4991 and reconstructing the <type>SERIAL</> column on reload
4992 using a new <link linkend="SQL-ALTERSEQUENCE"><command>ALTER
4993 SEQUENCE OWNED BY</></link> command. This also allows
4994 dropping a <type>SERIAL</> column specification.
5000 Add a server-side sleep function <link
5001 linkend="functions-datetime-delay"><function>pg_sleep()</></link>
5008 Add all comparison operators for the <link
5009 linkend="datatype-oid"><type>tid</></link> (tuple id) data
5010 type (Mark Kirkwood, Greg Stark, Tom)
5019 <title>PL/PgSQL Server-Side Language Changes</title>
5024 Add <literal>TG_table_name</> and <literal>TG_table_schema</> to
5025 trigger parameters (Andrew)
5029 <literal>TG_relname</> is now deprecated. Comparable
5030 changes have been made in the trigger parameters for the other
5037 Allow <literal>FOR</> statements to return values to scalars
5038 as well as records and row types (Pavel Stehule)
5044 Add a <literal>BY</> clause to the <literal>FOR</> loop,
5045 to control the iteration increment (Jaime Casanova)
5051 Add <literal>STRICT</> to <link
5052 linkend="plpgsql-statements-sql-onerow"><command>SELECT
5053 INTO</></link> (Matt Miller)
5057 <literal>STRICT</> mode throws an exception if more or less
5058 than one row is returned by the <command>SELECT</>, for
5059 <productname>Oracle PL/SQL</> compatibility.
5068 <title>PL/Perl Server-Side Language Changes</title>
5073 Add <literal>table_name</> and <literal>table_schema</> to
5074 trigger parameters (Adam Sjøgren)
5080 Add prepared queries (Dmitry Karasik)
5086 Make <literal>$_TD</> trigger data a global variable (Andrew)
5090 Previously, it was lexical, which caused unexpected sharing
5097 Run PL/Perl and PL/PerlU in separate interpreters, for security
5101 In consequence, they can no longer share data nor loaded modules.
5102 Also, if Perl has not been compiled with the requisite flags to
5103 allow multiple interpreters, only one of these languages can be used
5104 in any given backend process.
5113 <title>PL/Python Server-Side Language Changes</title>
5118 Named parameters are passed as ordinary variables, as well as in the
5119 <literal>args[]</> array (Sven Suursoho)
5125 Add <literal>table_name</> and <literal>table_schema</> to
5126 trigger parameters (Andrew)
5132 Allow returning of composite types and result sets (Sven Suursoho)
5138 Return result-set as <literal>list</>, <literal>iterator</>,
5139 or <literal>generator </>(Sven Suursoho)
5145 Allow functions to return <literal>void</> (Neil)
5151 Python 2.5 is now supported (Tom)
5160 <title><link linkend="APP-PSQL"><application>psql</></link> Changes</title>
5165 Add new command <literal>\password</> for changing role
5166 password with client-side password encryption (Peter)
5172 Allow <literal>\c</> to connect to a new host and port
5173 number (David, Volkan YAZICI)
5179 Add tablespace display to <literal>\l+</> (Philip Yarra)
5185 Improve <literal>\df</> slash command to include the argument
5186 names and modes (<literal>OUT</> or <literal>INOUT</>) of
5187 the function (David Fetter)
5193 Support binary <command>COPY</> (Andreas Pflug)
5199 Add option to run the entire session in a single transaction
5204 Use option <literal>-1</> or <literal>--single-transaction</>.
5210 Support for automatically retrieving <command>SELECT</>
5211 results in batches using a cursor (Chris Mair)
5215 This is enabled using <command>\set FETCH_COUNT
5216 <replaceable>n</></command>. This
5217 feature allows large result sets to be retrieved in
5218 <application>psql</> without attempting to buffer the entire
5219 result set in memory.
5225 Make multi-line values align in the proper column
5226 (Martijn van Oosterhout)
5230 Field values containing newlines are now displayed in a more
5237 Save multi-line statements as a single entry, rather than
5238 one line at a time (Sergey E. Koposov)
5242 This makes up-arrow recall of queries easier. (This is
5243 not available on Windows, because that platform uses the native
5244 command-line editing present in the operating system.)
5250 Make the line counter 64-bit so it can handle files with more
5251 than two billion lines (David Fetter)
5257 Report both the returned data and the command status tag
5258 for <command>INSERT</>/<command>UPDATE</>/<command>DELETE
5268 <title><link linkend="APP-PGDUMP"><application>pg_dump</></link> Changes</title>
5273 Allow complex selection of objects to be included or excluded
5274 by <application>pg_dump</> (Greg Sabino Mullane)
5278 <application>pg_dump</> now supports multiple <literal>-n</>
5279 (schema) and <literal>-t</> (table) options, and adds
5280 <literal>-N</> and <literal>-T</> options to exclude objects.
5281 Also, the arguments of these switches can now be wild-card expressions
5282 rather than single object names, for example
5283 <literal>-t 'foo*'</>, and a schema can be part of
5284 a <literal>-t</> or <literal>-T</> switch, for example
5285 <literal>-t schema1.table1</>.
5291 Add <link linkend="APP-PGRESTORE"><application>pg_restore</></link>
5292 <literal>--no-data-for-failed-tables</> option to suppress
5293 loading data if table creation failed (i.e., the table already
5294 exists) (Martin Pitt)
5300 Add <link linkend="APP-PGRESTORE"><application>pg_restore</></link>
5301 option to run the entire session in a single transaction
5306 Use option <literal>-1</> or <literal>--single-transaction</>.
5315 <title><link linkend="libpq"><application>libpq</></link> Changes</title>
5321 linkend="libpq-misc"><function>PQencryptPassword()</></link>
5322 to encrypt passwords (Tom)
5326 This allows passwords to be sent pre-encrypted for commands
5327 like <link linkend="SQL-ALTERROLE"><command>ALTER ROLE ...
5335 linkend="libpq-threading"><function>PQisthreadsafe()</></link>
5340 This allows applications to query the thread-safety status
5348 linkend="libpq-exec-main"><function>PQdescribePrepared()</></link>,
5350 linkend="libpq-exec-main"><function>PQdescribePortal()</></link>,
5351 and related functions to return information about previously
5352 prepared statements and open cursors (Volkan YAZICI)
5358 Allow <link linkend="libpq-ldap"><acronym>LDAP</></link> lookups
5360 linkend="libpq-pgservice"><filename>pg_service.conf</></link>
5367 Allow a hostname in <link
5368 linkend="libpq-pgpass"><filename>~/.pgpass</></link>
5369 to match the default socket directory (Bruce)
5373 A blank hostname continues to match any Unix-socket connection,
5374 but this addition allows entries that are specific to one of
5375 several postmasters on the machine.
5384 <title><link linkend="ecpg"><application>ecpg</></link> Changes</title>
5389 Allow <link linkend="SQL-SHOW"><command>SHOW</></link> to
5390 put its result into a variable (Joachim Wieland)
5396 Add <link linkend="SQL-COPY"><command>COPY TO STDOUT</></link>
5403 Add regression tests (Joachim Wieland, Michael)
5409 Major source code cleanups (Joachim Wieland, Michael)
5418 <title><application>Windows</> Port</title>
5423 Allow <acronym>MSVC</> to compile the <productname>PostgreSQL</>
5424 server (Magnus, Hiroshi Saito)
5430 Add <acronym>MSVC</> support for utility commands and <link
5431 linkend="APP-PGDUMP"><application>pg_dump</></link> (Hiroshi
5438 Add support for Windows code pages <literal>1253</>,
5439 <literal>1254</>, <literal>1255</>, and <literal>1257</>
5446 Drop privileges on startup, so that the server can be started from
5447 an administrative account (Magnus)
5453 Stability fixes (Qingqing Zhou, Magnus)
5459 Add native semaphore implementation (Qingqing Zhou)
5463 The previous code mimicked SysV semaphores.
5472 <title>Source Code Changes</title>
5477 Add <link linkend="GIN"><acronym>GIN</></link> (Generalized
5478 Inverted iNdex) index access method (Teodor, Oleg)
5484 Remove R-tree indexing (Tom)
5488 Rtree has been re-implemented using <link
5489 linkend="GiST"><acronym>GiST</></link>. Among other
5490 differences, this means that rtree indexes now have support
5491 for crash recovery via write-ahead logging (WAL).
5497 Reduce libraries needlessly linked into the backend (Martijn
5498 van Oosterhout, Tom)
5504 Add a configure flag to allow libedit to be preferred over
5505 <acronym>GNU</> readline (Bruce)
5510 linkend="configure"><literal>--with-libedit-preferred</></link>.
5516 Allow installation into directories containing spaces
5523 Improve ability to relocate installation directories (Tom)
5529 Add support for <productname>Solaris x86_64</> using the
5530 <productname>Solaris</> compiler (Pierre Girard, Theo
5531 Schlossnagle, Bruce)
5537 Add <application>DTrace</> support (Robert Lor)
5543 Add <literal>PG_VERSION_NUM</> for use by third-party
5544 applications wanting to test the backend version in C using >
5545 and < comparisons (Bruce)
5551 Add <literal>XLOG_BLCKSZ</> as independent from <literal>BLCKSZ</>
5558 Add <literal>LWLOCK_STATS</> define to report locking
5565 Emit warnings for unknown <application>configure</> options
5566 (Martijn van Oosterhout)
5572 Add server support for <quote>plugin</> libraries
5573 that can be used for add-on tasks such as debugging and performance
5574 measurement (Korry Douglas)
5578 This consists of two features: a table of <quote>rendezvous
5579 variables</> that allows separately-loaded shared libraries to
5580 communicate, and a new configuration parameter <link
5581 linkend="guc-local-preload-libraries"><varname>local_preload_libraries</></link>
5582 that allows libraries to be loaded into specific sessions without
5583 explicit cooperation from the client application. This allows
5584 external add-ons to implement features such as a PL/PgSQL debugger.
5590 Rename existing configuration parameter
5591 <varname>preload_libraries</> to <link
5592 linkend="guc-shared-preload-libraries"><varname>shared_preload_libraries</></link>
5597 This was done for clarity in comparison to
5598 <varname>local_preload_libraries</>.
5604 Add new configuration parameter <link
5605 linkend="guc-server-version-num"><varname>server_version_num</></link>
5606 (Greg Sabino Mullane)
5610 This is like <varname>server_version</varname>, but is an
5611 integer, e.g. <literal>80200</>. This allows applications to
5612 make version checks more easily.
5618 Add a configuration parameter <link
5619 linkend="guc-seq-page-cost"><varname>seq_page_cost</></link>
5626 Re-implement the <link linkend="regress">regression test</link> script as a C program
5633 Allow loadable modules to allocate shared memory and
5634 lightweight locks (Marc Munro)
5640 Add automatic initialization and finalization of dynamically
5641 loaded libraries (Ralf Engelschall, Tom)
5645 New <link linkend="xfunc-c-dynload">functions</link>
5646 <function>_PG_init()</> and <function>_PG_fini()</> are
5647 called if the library defines such symbols. Hence we no
5648 longer need to specify an initialization function in
5649 <varname>shared_preload_libraries</>; we can assume that
5650 the library used the <function>_PG_init()</> convention
5658 linkend="xfunc-c-dynload"><literal>PG_MODULE_MAGIC</></link>
5659 header block to all shared object files (Martijn van
5664 The magic block prevents version mismatches between loadable object
5671 Add shared library support for AIX (Laurenz Albe)
5677 New <link linkend="datatype-xml"><acronym>XML</></link>
5678 documentation section (Bruce)
5687 <title>Contrib Changes</title>
5692 Major tsearch2 improvements (Oleg, Teodor)
5699 multibyte encoding support, including <acronym>UTF8</>
5704 query rewriting support
5709 improved ranking functions
5714 thesaurus dictionary support
5719 Ispell dictionaries now recognize <application>MySpell</>
5720 format, used by <application>OpenOffice</>
5725 <acronym>GIN</> support
5735 Add adminpack module containing <application>Pgadmin</> administration
5740 These functions provide additional file system access
5741 routines not present in the default <productname>PostgreSQL</>
5748 Add sslinfo module (Victor Wagner)
5752 Reports information about the current connection's <acronym>SSL</>
5759 Add pgrowlocks module (Tatsuo)
5763 This shows row locking information for a specified table.
5769 Add hstore module (Oleg, Teodor)
5775 Add isn module, replacing isbn_issn (Jeremy Kronuz)
5779 This new implementation supports <acronym>EAN13</>, <acronym>UPC</>,
5780 <acronym>ISBN</> (books), <acronym>ISMN</> (music), and
5781 <acronym>ISSN</> (serials).
5787 Add index information functions to pgstattuple (ITAGAKI Takahiro,
5794 Add pg_freespacemap module to display free space map information
5801 pgcrypto now has all planned functionality (Marko Kreen)
5806 Include iMath library in pgcrypto to have the public-key encryption
5807 functions always available.
5812 Add SHA224 algorithm that was missing in OpenBSD code.
5817 Activate builtin code for SHA224/256/384/512 hashes on older
5818 OpenSSL to have those algorithms always available.
5823 New function gen_random_bytes() that returns cryptographically strong
5824 randomness. Useful for generating encryption keys.
5829 Remove digest_exists(), hmac_exists() and cipher_exists() functions.
5837 Improvements to cube module (Joshua Reich)
5841 New functions are <function>cube(float[])</>,
5842 <function>cube(float[], float[])</>, and
5843 <function>cube_subset(cube, int4[])</>.
5849 Add async query capability to dblink (Kai Londenberg,
5856 New operators for array-subset comparisons (<literal>@></>,
5857 <literal><@</>, <literal>&&</>) (Tom)
5861 Various contrib packages already had these operators for their
5862 datatypes, but the naming wasn't consistent. We have now added
5863 consistently named array-subset comparison operators to the core code
5864 and all the contrib packages that have such functionality.
5865 (The old names remain available, but are deprecated.)
5871 Add uninstall scripts for all contrib packages that have install
5872 scripts (David, Josh Drake)
5883 <sect1 id="release-8-1-10">
5884 <title>Release 8.1.10</title>
5887 <title>Release date</title>
5888 <simpara>2007-09-17</simpara>
5892 This release contains a variety of fixes from 8.1.9.
5896 <title>Migration to Version 8.1.10</title>
5899 A dump/restore is not required for those running 8.1.X.
5900 However, if you are upgrading from a version earlier than 8.1.2,
5901 see the release notes for 8.1.2.
5907 <title>Changes</title>
5913 Prevent index corruption when a transaction inserts rows and
5914 then aborts close to the end of a concurrent <command>VACUUM</>
5915 on the same table (Tom)
5921 Make <command>CREATE DOMAIN ... DEFAULT NULL</> work properly (Tom)
5927 Allow the <type>interval</> data type to accept input consisting only of
5928 milliseconds or microseconds (Neil)
5934 Speed up rtree index insertion (Teodor)
5940 Fix excessive logging of <acronym>SSL</> error messages (Tom)
5946 Fix logging so that log messages are never interleaved when using
5947 the syslogger process (Andrew)
5953 Fix crash when <varname>log_min_error_statement</> logging runs out
5960 Fix incorrect handling of some foreign-key corner cases (Tom)
5966 Prevent <command>REINDEX</> and <command>CLUSTER</> from failing
5967 due to attempting to process temporary tables of other sessions (Alvaro)
5973 Update the time zone database rules, particularly New Zealand's upcoming changes (Tom)
5979 Windows socket improvements (Magnus)
5985 Suppress timezone name (<literal>%Z</>) in log timestamps on Windows
5986 because of possible encoding mismatches (Tom)
5992 Require non-superusers who use <filename>/contrib/dblink</> to use only
5993 password authentication, as a security measure (Joe)
6002 <sect1 id="release-8-1-9">
6003 <title>Release 8.1.9</title>
6006 <title>Release date</title>
6007 <simpara>2007-04-23</simpara>
6011 This release contains a variety of fixes from 8.1.8,
6012 including a security fix.
6016 <title>Migration to Version 8.1.9</title>
6019 A dump/restore is not required for those running 8.1.X.
6020 However, if you are upgrading from a version earlier than 8.1.2,
6021 see the release notes for 8.1.2.
6027 <title>Changes</title>
6033 Support explicit placement of the temporary-table schema within
6034 <varname>search_path</>, and disable searching it for functions
6038 This is needed to allow a security-definer function to set a
6039 truly secure value of <varname>search_path</>. Without it,
6040 an unprivileged SQL user can use temporary objects to execute code
6041 with the privileges of the security-definer function (CVE-2007-2138).
6042 See <command>CREATE FUNCTION</> for more information.
6048 <filename>/contrib/tsearch2</> crash fixes (Teodor)
6054 Require <command>COMMIT PREPARED</> to be executed in the same
6055 database as the transaction was prepared in (Heikki)
6061 Fix potential-data-corruption bug in how <command>VACUUM FULL</> handles
6062 <command>UPDATE</> chains (Tom, Pavan Deolasee)
6068 Planner fixes, including improving outer join and bitmap scan
6069 selection logic (Tom)
6075 Fix PANIC during enlargement of a hash index (bug introduced in 8.1.6)
6082 Fix POSIX-style timezone specs to follow new USA DST rules (Tom)
6091 <sect1 id="release-8-1-8">
6092 <title>Release 8.1.8</title>
6095 <title>Release date</title>
6096 <simpara>2007-02-07</simpara>
6100 This release contains one fix from 8.1.7.
6104 <title>Migration to Version 8.1.8</title>
6107 A dump/restore is not required for those running 8.1.X.
6108 However, if you are upgrading from a version earlier than 8.1.2,
6109 see the release notes for 8.1.2.
6115 <title>Changes</title>
6121 Remove overly-restrictive check for type length in constraints and
6122 functional indexes(Tom)
6131 <sect1 id="release-8-1-7">
6132 <title>Release 8.1.7</title>
6135 <title>Release date</title>
6136 <simpara>2007-02-05</simpara>
6140 This release contains a variety of fixes from 8.1.6, including
6145 <title>Migration to Version 8.1.7</title>
6148 A dump/restore is not required for those running 8.1.X.
6149 However, if you are upgrading from a version earlier than 8.1.2,
6150 see the release notes for 8.1.2.
6156 <title>Changes</title>
6162 Remove security vulnerabilities that allowed connected users
6163 to read backend memory (Tom)
6166 The vulnerabilities involve suppressing the normal check that a SQL
6167 function returns the data type it's declared to, and changing the
6168 data type of a table column (CVE-2007-0555, CVE-2007-0556). These
6169 errors can easily be exploited to cause a backend crash, and in
6170 principle might be used to read database content that the user
6171 should not be able to access.
6177 Fix rare bug wherein btree index page splits could fail
6178 due to choosing an infeasible split point (Heikki Linnakangas)
6184 Improve <command>VACUUM</> performance for databases with many tables (Tom)
6190 Fix autovacuum to avoid leaving non-permanent transaction IDs in
6191 non-connectable databases (Alvaro)
6195 This bug affects the 8.1 branch only.
6201 Fix for rare Assert() crash triggered by <literal>UNION</> (Tom)
6207 Tighten security of multi-byte character processing for UTF8 sequences
6208 over three bytes long (Tom)
6214 Fix bogus <quote>permission denied</> failures occurring on Windows
6215 due to attempts to fsync already-deleted files (Magnus, Tom)
6221 Fix possible crashes when an already-in-use PL/pgSQL function is
6231 <sect1 id="release-8-1-6">
6232 <title>Release 8.1.6</title>
6235 <title>Release date</title>
6236 <simpara>2007-01-08</simpara>
6240 This release contains a variety of fixes from 8.1.5.
6244 <title>Migration to Version 8.1.6</title>
6247 A dump/restore is not required for those running 8.1.X.
6248 However, if you are upgrading from a version earlier than 8.1.2,
6249 see the release notes for 8.1.2.
6255 <title>Changes</title>
6261 Improve handling of <function>getaddrinfo()</> on AIX (Tom)
6265 This fixes a problem with starting the statistics collector,
6272 Fix <application>pg_restore</> to handle a tar-format backup
6273 that contains large objects (blobs) with comments (Tom)
6279 Fix <quote>failed to re-find parent key</> errors in
6280 <command>VACUUM</> (Tom)
6286 Clean out <filename>pg_internal.init</> cache files during server
6291 This avoids a hazard that the cache files might contain stale
6292 data after PITR recovery.
6298 Fix race condition for truncation of a large relation across a
6299 gigabyte boundary by <command>VACUUM</> (Tom)
6305 Fix bug causing needless deadlock errors on row-level locks (Tom)
6311 Fix bugs affecting multi-gigabyte hash indexes (Tom)
6317 Fix possible deadlock in Windows signal handling (Teodor)
6323 Fix error when constructing an <literal>ARRAY[]</> made up of multiple
6324 empty elements (Tom)
6330 Fix ecpg memory leak during connection (Michael)
6336 Fix for Darwin (OS X) compilation (Tom)
6342 <function>to_number()</> and <function>to_char(numeric)</>
6343 are now <literal>STABLE</>, not <literal>IMMUTABLE</>, for
6344 new <application>initdb</> installs (Tom)
6348 This is because <varname>lc_numeric</> can potentially
6349 change the output of these functions.
6355 Improve index usage of regular expressions that use parentheses (Tom)
6359 This improves <application>psql</> <literal>\d</> performance also.
6365 Update timezone database
6369 This affects Australian and Canadian daylight-savings rules in
6379 <sect1 id="release-8-1-5">
6380 <title>Release 8.1.5</title>
6383 <title>Release date</title>
6384 <simpara>2006-10-16</simpara>
6388 This release contains a variety of fixes from 8.1.4.
6392 <title>Migration to Version 8.1.5</title>
6395 A dump/restore is not required for those running 8.1.X.
6396 However, if you are upgrading from a version earlier than 8.1.2,
6397 see the release notes for 8.1.2.
6403 <title>Changes</title>
6406 <listitem><para>Disallow aggregate functions in <command>UPDATE</>
6407 commands, except within sub-SELECTs (Tom)</para>
6408 <para>The behavior of such an aggregate was unpredictable, and in 8.1.X
6409 could cause a crash, so it has been disabled. The SQL standard does not allow
6410 this either.</para></listitem>
6411 <listitem><para>Fix core dump when an untyped literal is taken as
6412 ANYARRAY</para></listitem>
6413 <listitem><para>Fix core dump in duration logging for extended query protocol
6414 when a <command>COMMIT</> or <command>ROLLBACK</> is
6415 executed</para></listitem>
6416 <listitem><para>Fix mishandling of AFTER triggers when query contains a SQL
6417 function returning multiple rows (Tom)</para></listitem>
6418 <listitem><para>Fix <command>ALTER TABLE ... TYPE</> to recheck
6419 <literal>NOT NULL</> for <literal>USING</> clause (Tom)</para></listitem>
6420 <listitem><para>Fix <function>string_to_array()</> to handle overlapping
6421 matches for the separator string</para>
6422 <para>For example, <literal>string_to_array('123xx456xxx789', 'xx')</>.
6424 <listitem><para>Fix <function>to_timestamp()</> for
6425 <literal>AM</>/<literal>PM</> formats (Bruce)</para></listitem>
6426 <listitem><para>Fix autovacuum's calculation that decides whether
6427 <command>ANALYZE</> is needed (Alvaro)</para></listitem>
6428 <listitem><para>Fix corner cases in pattern matching for
6429 <application>psql</>'s <literal>\d</> commands</para></listitem>
6430 <listitem><para>Fix index-corrupting bugs in /contrib/ltree
6431 (Teodor)</para></listitem>
6432 <listitem><para>Numerous robustness fixes in <application>ecpg</> (Joachim
6433 Wieland)</para></listitem>
6434 <listitem><para>Fix backslash escaping in /contrib/dbmirror</para></listitem>
6435 <listitem><para>Minor fixes in /contrib/dblink and /contrib/tsearch2</para>
6437 <listitem><para>Efficiency improvements in hash tables and bitmap index scans
6438 (Tom)</para></listitem>
6439 <listitem><para>Fix instability of statistics collection on Windows (Tom, Andrew)</para></listitem>
6440 <listitem><para>Fix <varname>statement_timeout</> to use the proper
6441 units on Win32 (Bruce)</para>
6442 <para>In previous Win32 8.1.X versions, the delay was off by a factor of
6443 100.</para></listitem>
6444 <listitem><para>Fixes for <acronym>MSVC</> and <productname>Borland C++</>
6445 compilers (Hiroshi Saito)</para></listitem>
6446 <listitem><para>Fixes for <systemitem class="osname">AIX</> and
6447 <productname>Intel</> compilers (Tom)</para></listitem>
6448 <listitem><para>Fix rare bug in continuous archiving (Tom)</para></listitem>
6454 <sect1 id="release-8-1-4">
6455 <title>Release 8.1.4</title>
6458 <title>Release date</title>
6459 <simpara>2006-05-23</simpara>
6463 This release contains a variety of fixes from 8.1.3,
6464 including patches for extremely serious security issues.
6468 <title>Migration to Version 8.1.4</title>
6471 A dump/restore is not required for those running 8.1.X.
6472 However, if you are upgrading from a version earlier than 8.1.2,
6473 see the release notes for 8.1.2.
6477 Full security against the SQL-injection attacks described in
6478 CVE-2006-2313 and CVE-2006-2314 might require changes in application
6479 code. If you have applications that embed untrustworthy strings
6480 into SQL commands, you should examine them as soon as possible to
6481 ensure that they are using recommended escaping techniques. In
6482 most cases, applications should be using subroutines provided by
6483 libraries or drivers (such as <application>libpq</>'s
6484 <function>PQescapeStringConn()</>) to perform string escaping,
6485 rather than relying on <foreignphrase>ad hoc</> code to do it.
6490 <title>Changes</title>
6493 <listitem><para>Change the server to reject invalidly-encoded multibyte
6494 characters in all cases (Tatsuo, Tom)</para>
6495 <para>While <productname>PostgreSQL</> has been moving in this direction for
6496 some time, the checks are now applied uniformly to all encodings and all
6497 textual input, and are now always errors not merely warnings. This change
6498 defends against SQL-injection attacks of the type described in CVE-2006-2313.
6501 <listitem><para>Reject unsafe uses of <literal>\'</> in string literals</para>
6502 <para>As a server-side defense against SQL-injection attacks of the type
6503 described in CVE-2006-2314, the server now only accepts <literal>''</> and not
6504 <literal>\'</> as a representation of ASCII single quote in SQL string
6505 literals. By default, <literal>\'</> is rejected only when
6506 <varname>client_encoding</> is set to a client-only encoding (SJIS, BIG5, GBK,
6507 GB18030, or UHC), which is the scenario in which SQL injection is possible.
6508 A new configuration parameter <varname>backslash_quote</> is available to
6509 adjust this behavior when needed. Note that full security against
6510 CVE-2006-2314 might require client-side changes; the purpose of
6511 <varname>backslash_quote</> is in part to make it obvious that insecure
6512 clients are insecure.
6515 <listitem><para>Modify <application>libpq</>'s string-escaping routines to be
6516 aware of encoding considerations and
6517 <varname>standard_conforming_strings</></para>
6518 <para>This fixes <application>libpq</>-using applications for the security
6519 issues described in CVE-2006-2313 and CVE-2006-2314, and also future-proofs
6520 them against the planned changeover to SQL-standard string literal syntax.
6521 Applications that use multiple <productname>PostgreSQL</> connections
6522 concurrently should migrate to <function>PQescapeStringConn()</> and
6523 <function>PQescapeByteaConn()</> to ensure that escaping is done correctly
6524 for the settings in use in each database connection. Applications that
6525 do string escaping <quote>by hand</> should be modified to rely on library
6529 <listitem><para>Fix weak key selection in pgcrypto (Marko Kreen)</para>
6530 <para>Errors in fortuna PRNG reseeding logic could cause a predictable
6531 session key to be selected by <function>pgp_sym_encrypt()</> in some cases.
6532 This only affects non-OpenSSL-using builds.
6535 <listitem><para>Fix some incorrect encoding conversion functions</para>
6536 <para><function>win1251_to_iso</>, <function>win866_to_iso</>,
6537 <function>euc_tw_to_big5</>, <function>euc_tw_to_mic</>,
6538 <function>mic_to_euc_tw</> were all broken to varying
6542 <listitem><para>Clean up stray remaining uses of <literal>\'</> in strings
6543 (Bruce, Jan)</para></listitem>
6545 <listitem><para>Make autovacuum visible in <structname>pg_stat_activity</>
6546 (Alvaro)</para></listitem>
6548 <listitem><para>Disable <literal>full_page_writes</> (Tom)</para>
6549 <para>In certain cases, having <literal>full_page_writes</> off would cause
6550 crash recovery to fail. A proper fix will appear in 8.2; for now it's just
6554 <listitem><para>Various planner fixes, particularly for bitmap index scans and
6555 MIN/MAX optimization (Tom)</para></listitem>
6557 <listitem><para>Fix incorrect optimization in merge join (Tom)</para>
6558 <para>Outer joins could sometimes emit multiple copies of unmatched rows.
6561 <listitem><para>Fix crash from using and modifying a plpgsql function in the
6562 same transaction</para></listitem>
6564 <listitem><para>Fix WAL replay for case where a B-Tree index has been
6565 truncated</para></listitem>
6567 <listitem><para>Fix <literal>SIMILAR TO</> for patterns involving
6568 <literal>|</> (Tom)</para></listitem>
6570 <listitem><para>Fix <command>SELECT INTO</> and <command>CREATE TABLE AS</> to
6571 create tables in the default tablespace, not the base directory (Kris
6572 Jurka)</para></listitem>
6574 <listitem><para>Fix server to use custom DH SSL parameters correctly (Michael
6575 Fuhr)</para></listitem>
6577 <listitem><para>Improve qsort performance (Dann Corbit)</para>
6578 <para>Currently this code is only used on Solaris.
6581 <listitem><para>Fix for OS/X Bonjour on x86 systems (Ashley Clark)</para></listitem>
6583 <listitem><para>Fix various minor memory leaks</para></listitem>
6585 <listitem><para>Fix problem with password prompting on some Win32 systems
6586 (Robert Kinberg)</para></listitem>
6588 <listitem><para>Improve <application>pg_dump</>'s handling of default values
6589 for domains</para></listitem>
6591 <listitem><para>Fix <application>pg_dumpall</> to handle identically-named
6592 users and groups reasonably (only possible when dumping from a pre-8.1 server)
6594 <para>The user and group will be merged into a single role with
6595 <literal>LOGIN</> permission. Formerly the merged role wouldn't have
6596 <literal>LOGIN</> permission, making it unusable as a user.
6599 <listitem><para>Fix <application>pg_restore</> <literal>-n</> to work as
6600 documented (Tom)</para></listitem>
6606 <sect1 id="release-8-1-3">
6607 <title>Release 8.1.3</title>
6610 <title>Release date</title>
6611 <simpara>2006-02-14</simpara>
6615 This release contains a variety of fixes from 8.1.2,
6616 including one very serious security issue.
6620 <title>Migration to Version 8.1.3</title>
6623 A dump/restore is not required for those running 8.1.X.
6624 However, if you are upgrading from a version earlier than 8.1.2,
6625 see the release notes for 8.1.2.
6630 <title>Changes</title>
6634 <listitem><para>Fix bug that allowed any logged-in user to <command>SET
6635 ROLE</> to any other database user id (CVE-2006-0553)</para>
6636 <para>Due to inadequate validity checking, a user could exploit the special
6637 case that <command>SET ROLE</> normally uses to restore the previous role
6638 setting after an error. This allowed ordinary users to acquire superuser
6639 status, for example.
6640 The escalation-of-privilege risk exists only in 8.1.0-8.1.2.
6641 However, in all releases back to 7.3 there is a related bug in <command>SET
6642 SESSION AUTHORIZATION</> that allows unprivileged users to crash the server,
6643 if it has been compiled with Asserts enabled (which is not the default).
6644 Thanks to Akio Ishida for reporting this problem.
6647 <listitem><para>Fix bug with row visibility logic in self-inserted
6649 <para>Under rare circumstances a row inserted by the current command
6650 could be seen as already valid, when it should not be. Repairs bug
6651 created in 8.0.4, 7.4.9, and 7.3.11 releases.
6654 <listitem><para>Fix race condition that could lead to <quote>file already
6655 exists</> errors during pg_clog and pg_subtrans file creation
6656 (Tom)</para></listitem>
6658 <listitem><para>Fix cases that could lead to crashes if a cache-invalidation
6659 message arrives at just the wrong time (Tom)</para></listitem>
6661 <listitem><para>Properly check <literal>DOMAIN</> constraints for
6662 <literal>UNKNOWN</> parameters in prepared statements
6663 (Neil)</para></listitem>
6665 <listitem><para>Ensure <command>ALTER COLUMN TYPE</> will process
6666 <literal>FOREIGN KEY</>, <literal>UNIQUE</>, and <literal>PRIMARY KEY</>
6667 constraints in the proper order (Nakano Yoshihisa)</para></listitem>
6669 <listitem><para>Fixes to allow restoring dumps that have cross-schema
6670 references to custom operators or operator classes (Tom)</para></listitem>
6672 <listitem><para>Allow <application>pg_restore</> to continue properly after a
6673 <command>COPY</> failure; formerly it tried to treat the remaining
6674 <command>COPY</> data as SQL commands (Stephen Frost)</para></listitem>
6676 <listitem><para>Fix <application>pg_ctl</> <literal>unregister</> crash
6677 when the data directory is not specified (Magnus)</para></listitem>
6679 <listitem><para>Fix <application>libpq</> <function>PQprint</> HTML tags
6680 (Christoph Zwerschke)</para></listitem>
6682 <listitem><para>Fix <application>ecpg</> crash on AMD64 and PPC
6683 (Neil)</para></listitem>
6685 <listitem><para>Allow <literal>SETOF</> and <literal>%TYPE</> to be used
6686 together in function result type declarations</para></listitem>
6688 <listitem><para>Recover properly if error occurs during argument passing
6689 in <application>PL/python</> (Neil)</para></listitem>
6691 <listitem><para>Fix memory leak in <function>plperl_return_next</>
6692 (Neil)</para></listitem>
6694 <listitem><para>Fix <application>PL/perl</>'s handling of locales on
6695 Win32 to match the backend (Andrew)</para></listitem>
6697 <listitem><para>Various optimizer fixes (Tom)</para></listitem>
6699 <listitem><para>Fix crash when <literal>log_min_messages</> is set to
6700 <literal>DEBUG3</> or above in <filename>postgresql.conf</> on Win32
6701 (Bruce)</para></listitem>
6703 <listitem><para>Fix <application>pgxs</> <literal>-L</> library path
6704 specification for Win32, Cygwin, OS X, AIX (Bruce)</para></listitem>
6706 <listitem><para>Check that SID is enabled while checking for Win32 admin
6707 privileges (Magnus)</para></listitem>
6709 <listitem><para>Properly reject out-of-range date inputs (Kris
6710 Jurka)</para></listitem>
6712 <listitem><para>Portability fix for testing presence of <function>finite</>
6713 and <function>isinf</> during configure (Tom)</para></listitem>
6715 <listitem><para>Improve speed of <command>COPY IN</> via libpq, by
6716 avoiding a kernel call per data line (Alon Goldshuv)</para></listitem>
6718 <listitem><para>Improve speed of <filename>/contrib/tsearch2</> index
6719 creation (Tom)</para></listitem>
6726 <sect1 id="release-8-1-2">
6727 <title>Release 8.1.2</title>
6730 <title>Release date</title>
6731 <simpara>2006-01-09</simpara>
6735 This release contains a variety of fixes from 8.1.1.
6739 <title>Migration to Version 8.1.2</title>
6742 A dump/restore is not required for those running 8.1.X.
6743 However, you might need to <command>REINDEX</> indexes on textual
6744 columns after updating, if you are affected by the locale or
6745 <application>plperl</> issues described below.
6750 <title>Changes</title>
6754 <listitem><para>Fix Windows code so that postmaster will continue rather
6755 than exit if there is no more room in ShmemBackendArray (Magnus)</para>
6756 <para>The previous behavior could lead to a denial-of-service situation if too
6757 many connection requests arrive close together. This applies
6758 <emphasis>only</> to the Windows port.</para></listitem>
6760 <listitem><para>Fix bug introduced in 8.0 that could allow ReadBuffer
6761 to return an already-used page as new, potentially causing loss of
6762 recently-committed data (Tom)</para></listitem>
6764 <listitem><para>Fix for protocol-level Describe messages issued
6765 outside a transaction or in a failed transaction (Tom)</para></listitem>
6767 <listitem><para>Fix character string comparison for locales that consider
6768 different character combinations as equal, such as Hungarian (Tom)</para>
6769 <para>This might require <command>REINDEX</> to fix existing indexes on
6770 textual columns.</para></listitem>
6772 <listitem><para>Set locale environment variables during postmaster startup
6773 to ensure that <application>plperl</> won't change the locale later</para>
6774 <para>This fixes a problem that occurred if the <application>postmaster</> was
6775 started with environment variables specifying a different locale than what
6776 <application>initdb</> had been told. Under these conditions, any use of
6777 <application>plperl</> was likely to lead to corrupt indexes. You might need
6778 <command>REINDEX</> to fix existing indexes on
6779 textual columns if this has happened to you.</para></listitem>
6781 <listitem><para>Allow more flexible relocation of installation
6782 directories (Tom)</para>
6783 <para>Previous releases supported relocation only if all installation
6784 directory paths were the same except for the last component.</para></listitem>
6786 <listitem><para>Prevent crashes caused by the use of
6787 <literal>ISO-8859-5</> and <literal>ISO-8859-9</> encodings
6788 (Tatsuo)</para></listitem>
6790 <listitem><para>Fix longstanding bug in strpos() and regular expression
6791 handling in certain rarely used Asian multi-byte character sets (Tatsuo)
6794 <listitem><para>Fix bug where COPY CSV mode considered any
6795 <literal>\.</> to terminate the copy data</para> <para>The new code
6796 requires <literal>\.</> to appear alone on a line, as per
6797 documentation.</para></listitem>
6799 <listitem><para>Make COPY CSV mode quote a literal data value of
6800 <literal>\.</> to ensure it cannot be interpreted as the
6801 end-of-data marker (Bruce)</para></listitem>
6803 <listitem><para>Various fixes for functions returning <literal>RECORD</>s
6804 (Tom) </para></listitem>
6806 <listitem><para>Fix processing of <filename>postgresql.conf</> so a
6807 final line with no newline is processed properly (Tom)
6810 <listitem><para>Fix bug in <filename>/contrib/pgcrypto</> gen_salt,
6811 which caused it not to use all available salt space for MD5 and
6812 XDES algorithms (Marko Kreen, Solar Designer)</para>
6813 <para>Salts for Blowfish and standard DES are unaffected.</para></listitem>
6815 <listitem><para>Fix autovacuum crash when processing expression indexes
6818 <listitem><para>Fix <filename>/contrib/dblink</> to throw an error,
6819 rather than crashing, when the number of columns specified is different from
6820 what's actually returned by the query (Joe)</para></listitem>
6827 <sect1 id="release-8-1-1">
6828 <title>Release 8.1.1</title>
6831 <title>Release date</title>
6832 <simpara>2005-12-12</simpara>
6836 This release contains a variety of fixes from 8.1.0.
6840 <title>Migration to Version 8.1.1</title>
6843 A dump/restore is not required for those running 8.1.X.
6848 <title>Changes</title>
6851 <listitem><para>Fix incorrect optimizations of outer-join conditions
6852 (Tom)</para></listitem>
6854 <listitem><para>Fix problems with wrong reported column names in cases
6855 involving sub-selects flattened by the optimizer (Tom)</para></listitem>
6857 <listitem><para>Fix update failures in scenarios involving CHECK constraints,
6858 toasted columns, <emphasis>and</> indexes (Tom)</para></listitem>
6860 <listitem><para>Fix bgwriter problems after recovering from errors
6863 The background writer was found to leak buffer pins after write errors.
6864 While not fatal in itself, this might lead to mysterious blockages of
6865 later VACUUM commands.
6869 <listitem><para>Prevent failure if client sends Bind protocol message
6870 when current transaction is already aborted</para></listitem>
6872 <listitem><para><filename>/contrib/tsearch2</> and <filename>/contrib/ltree</>
6873 fixes (Teodor)</para></listitem>
6875 <listitem><para>Fix problems with translated error messages in
6876 languages that require word reordering, such as Turkish; also problems with
6877 unexpected truncation of output strings and wrong display of the smallest
6878 possible bigint value (Andrew, Tom)</para>
6880 These problems only appeared on platforms that were using our
6881 <filename>port/snprintf.c</> code, which includes BSD variants if
6882 <literal>--enable-nls</> was given, and perhaps others. In addition,
6883 a different form of the translated-error-message problem could appear
6884 on Windows depending on which version of <filename>libintl</> was used.
6887 <listitem><para>Re-allow <literal>AM</>/<literal>PM</>, <literal>HH</>,
6888 <literal>HH12</>, and <literal>D</> format specifiers for
6889 <function>to_char(time)</> and <function>to_char(interval)</>.
6890 (<function>to_char(interval)</> should probably use
6891 <literal>HH24</>.) (Bruce)</para></listitem>
6893 <listitem><para>AIX, HPUX, and MSVC compile fixes (Tom, Hiroshi
6894 Saito)</para></listitem>
6896 <listitem><para>Optimizer improvements (Tom)</para></listitem>
6898 <listitem><para>Retry file reads and writes after Windows
6899 NO_SYSTEM_RESOURCES error (Qingqing Zhou)</para></listitem>
6901 <listitem><para>Prevent <application>autovacuum</> from crashing during
6902 ANALYZE of expression index (Alvaro)</para></listitem>
6904 <listitem><para>Fix problems with ON COMMIT DELETE ROWS temp
6905 tables</para></listitem>
6907 <listitem><para>Fix problems when a trigger alters the output of a SELECT
6908 DISTINCT query</para></listitem>
6910 <listitem><para>Add 8.1.0 release note item on how to migrate invalid
6911 <literal>UTF-8</> byte sequences (Paul Lindner)</para></listitem>
6917 <sect1 id="release-8-1">
6918 <title>Release 8.1</title>
6921 <title>Release date</title>
6922 <simpara>2005-11-08</simpara>
6926 <title>Overview</title>
6929 Major changes in this release:
6936 Improve concurrent access to the shared buffer cache (Tom)
6941 Access to the shared buffer cache was identified as a
6942 significant scalability problem, particularly on multi-CPU
6943 systems. In this release, the way that locking is done in the
6944 buffer manager has been overhauled to reduce lock contention
6945 and improve scalability. The buffer manager has also been
6946 changed to use a <quote>clock sweep</quote> replacement
6954 Allow index scans to use an intermediate in-memory bitmap (Tom)
6959 In previous releases, only a single index could be used to do
6960 lookups on a table. With this feature, if a query has
6961 <command>WHERE tab.col1 = 4 and tab.col2 = 9</>, and there is
6962 no multicolumn index on <literal>col1</> and <literal>col2</>,
6963 but there is an index on <literal>col1</> and another on
6964 <literal>col2</>, it is possible to search both indexes and
6965 combine the results in memory, then do heap fetches for only
6966 the rows matching both the <literal>col1</> and
6967 <literal>col2</> restrictions. This is very useful in
6968 environments that have a lot of unstructured queries where it
6969 is impossible to create indexes that match all possible access
6970 conditions. Bitmap scans are useful even with a single index,
6971 as they reduce the amount of random access needed; a bitmap
6972 index scan is efficient for retrieving fairly large fractions
6973 of the complete table, whereas plain index scans are not.
6980 Add two-phase commit (Heikki Linnakangas, Alvaro, Tom)
6985 Two-phase commit allows transactions to be "prepared" on several
6986 computers, and once all computers have successfully prepared
6987 their transactions (none failed), all transactions can be
6988 committed. Even if a machine crashes after a prepare, the
6989 prepared transaction can be committed after the machine is
6990 restarted. New syntax includes <command>PREPARE TRANSACTION</> and
6991 <command>COMMIT/ROLLBACK PREPARED</>. A new system view
6992 <literal>pg_prepared_xacts</> has also been added.
6999 Create a new role system that replaces users and groups
7005 Roles are a combination of users and groups. Like users, they
7006 can have login capability, and like groups, a role can have
7007 other roles as members. Roles basically remove the distinction
7008 between users and groups. For example, a role can:
7015 Have login capability (optionally)
7027 Hold access permissions for database objects
7033 Inherit permissions from other roles it is a member of
7039 Once a user logs into a role, she obtains capabilities of
7040 the login role plus any inherited roles, and can use
7041 <command>SET ROLE</> to switch to other roles she is a member of.
7042 This feature is a generalization of the SQL standard's concept of
7044 This change also replaces <structname>pg_shadow</> and
7045 <structname>pg_group</> by new role-capable catalogs
7046 <structname>pg_authid</> and <structname>pg_auth_members</>. The old
7047 tables are redefined as read-only views on the new role tables.
7054 Automatically use indexes for <function>MIN()</> and
7055 <function>MAX()</> (Tom)
7060 In previous releases, the only way to use an index for
7061 <function>MIN()</> or <function>MAX()</> was to rewrite the
7062 query as <command>SELECT col FROM tab ORDER BY col LIMIT 1</>.
7063 Index usage now happens automatically.
7070 Move <filename>/contrib/pg_autovacuum</> into the main server
7076 Integrating autovacuum into the server allows it to be
7077 automatically started and stopped in sync with the database
7078 server, and allows autovacuum to be configured from
7079 <filename>postgresql.conf</>.
7086 Add shared row level locks using <command>SELECT ... FOR SHARE</>
7092 While <productname>PostgreSQL</productname>'s MVCC locking
7093 allows <command>SELECT</> to never be blocked by writers and
7094 therefore does not need shared row locks for typical operations,
7095 shared locks are useful for applications that require shared row
7096 locking. In particular this reduces the locking requirements
7097 imposed by referential integrity checks.
7104 Add dependencies on shared objects, specifically roles
7110 This extension of the dependency mechanism prevents roles from
7111 being dropped while there are still database objects they own.
7112 Formerly it was possible to accidentally <quote>orphan</> objects by
7113 deleting their owner. While this could be recovered from, it
7114 was messy and unpleasant.
7121 Improve performance for partitioned tables (Simon)
7126 The new <varname>constraint_exclusion</varname> configuration
7127 parameter avoids lookups on child tables where constraints indicate
7128 that no matching rows exist in the child table.
7131 This allows for a basic type of table partitioning. If child tables
7132 store separate key ranges and this is enforced using appropriate
7133 <command>CHECK</> constraints, the optimizer will skip child
7134 table accesses when the constraint guarantees no matching rows
7135 exist in the child table.
7144 <title>Migration to Version 8.1</title>
7147 A dump/restore using <application>pg_dump</application> is required
7148 for those wishing to migrate data from any previous release.
7152 The 8.0 release announced that the <function>to_char()</> function
7153 for intervals would be removed in 8.1. However, since no better API
7154 has been suggested, <function>to_char(interval)</> has been enhanced in
7155 8.1 and will remain in the server.
7159 Observe the following incompatibilities:
7166 <varname>add_missing_from</> is now false by default (Neil)
7169 By default, we now generate an error if a table is used in a query
7170 without a <literal>FROM</> reference. The old behavior is still
7171 available, but the parameter must be set to 'true' to obtain it.
7175 It might be necessary to set <varname>add_missing_from</> to true
7176 in order to load an existing dump file, if the dump contains any
7177 views or rules created using the implicit-<literal>FROM</> syntax.
7178 This should be a one-time annoyance, because
7179 <productname>PostgreSQL</productname> 8.1 will convert
7180 such views and rules to standard explicit-<literal>FROM</> syntax.
7181 Subsequent dumps will therefore not have the problem.
7187 Cause input of a zero-length string (<literal>''</literal>) for
7188 <type>float4</type>/<type>float8</type>/<type>oid</type>
7189 to throw an error, rather than treating it as a zero (Neil)
7192 This change is consistent with the current handling of
7193 zero-length strings for integers. The schedule for this change
7194 was announced in 8.0.
7200 <varname>default_with_oids</> is now false by default (Neil)
7203 With this option set to false, user-created tables no longer
7204 have an OID column unless <command>WITH OIDS</> is specified in
7205 <command>CREATE TABLE</>. Though OIDs have existed in all
7206 releases of <productname>PostgreSQL</>, their use is limited
7207 because they are only four bytes long and the counter is shared
7208 across all installed databases. The preferred way of uniquely
7209 identifying rows is via sequences and the <type>SERIAL</> type,
7210 which have been supported since <productname>PostgreSQL</> 6.4.
7216 Add <literal>E''</> syntax so eventually ordinary strings can
7217 treat backslashes literally (Bruce)
7220 Currently <productname>PostgreSQL</productname> processes a
7221 backslash in a string literal as introducing a special escape sequence,
7222 e.g. <literal>\n</> or <literal>\010</>.
7223 While this allows easy entry of special values, it is
7224 nonstandard and makes porting of applications from other
7225 databases more difficult. For this reason, the
7226 <productname>PostgreSQL</productname> project is planning to
7227 remove the special meaning of backslashes in strings. For
7228 backward compatibility and for users who want special backslash
7229 processing, a new string syntax has been created. This new string
7230 syntax is formed by writing an <literal>E</> immediately preceding the
7231 single quote that starts the string, e.g. <literal>E'hi\n'</>. While
7232 this release does not change the handling of backslashes in strings, it
7233 does add new configuration parameters to help users migrate applications
7234 for future releases:
7240 <varname>standard_conforming_strings</> — does this release
7241 treat backslashes literally in ordinary strings?
7247 <varname>escape_string_warning</> — warn about backslashes in
7248 ordinary (non-E) strings
7255 The <varname>standard_conforming_strings</> value is read-only.
7256 Applications can retrieve the value to know how backslashes are
7257 processed. (Presence of the parameter can also be taken as an
7258 indication that <literal>E''</> string syntax is supported.)
7259 In a future release, <varname>standard_conforming_strings</>
7260 will be true, meaning backslashes will be treated literally in
7261 non-E strings. To prepare for this change, use <literal>E''</>
7262 strings in places that need special backslash processing, and
7263 turn on <varname>escape_string_warning</> to find additional
7264 strings that need to be converted to use <literal>E''</>.
7265 Also, use two single-quotes (<literal>''</>) to embed a literal
7266 single-quote in a string, rather than the
7267 <productname>PostgreSQL</productname>-supported syntax of
7268 backslash single-quote (<literal>\'</>). The former is
7269 standards-conforming and does not require the use of the
7270 <literal>E''</> string syntax. You can also use the
7271 <literal>$$</> string syntax, which does not treat backslashes
7278 Make <command>REINDEX DATABASE</> reindex all indexes in the
7282 Formerly, <command>REINDEX DATABASE</> reindexed only
7283 system tables. This new behavior seems more intuitive. A new
7284 command <command>REINDEX SYSTEM</> provides the old functionality
7285 of reindexing just the system tables.
7291 Read-only large object descriptors now obey MVCC snapshot semantics
7294 When a large object is opened with <literal>INV_READ</> (and not
7295 <literal>INV_WRITE</>), the data read from the descriptor will now
7296 reflect a <quote>snapshot</> of the large object's state at the
7297 time of the transaction snapshot in use by the query that called
7298 <function>lo_open()</>. To obtain the old behavior of always
7299 returning the latest committed data, include <literal>INV_WRITE</>
7300 in the mode flags for <function>lo_open()</>.
7306 Add proper dependencies for arguments of sequence functions (Tom)
7309 In previous releases, sequence names passed to <function>nextval()</>,
7310 <function>currval()</>, and <function>setval()</> were stored as
7311 simple text strings, meaning that renaming or dropping a
7312 sequence used in a <literal>DEFAULT</> clause made the clause
7313 invalid. This release stores all newly-created sequence function
7314 arguments as internal OIDs, allowing them to track sequence
7315 renaming, and adding dependency information that prevents
7316 improper sequence removal. It also makes such <literal>DEFAULT</>
7317 clauses immune to schema renaming and search path changes.
7320 Some applications might rely on the old behavior of
7321 run-time lookup for sequence names. This can still be done by
7322 explicitly casting the argument to <type>text</>, for example
7323 <literal>nextval('myseq'::text)</>.
7326 Pre-8.1 database dumps loaded into 8.1 will use the old text-based
7327 representation and therefore will not have the features of
7328 OID-stored arguments. However, it is possible to update a
7329 database containing text-based <literal>DEFAULT</> clauses.
7330 First, save this query into a file, such as <filename>fixseq.sql</>:
7332 SELECT 'ALTER TABLE ' ||
7333 pg_catalog.quote_ident(n.nspname) || '.' ||
7334 pg_catalog.quote_ident(c.relname) ||
7335 ' ALTER COLUMN ' || pg_catalog.quote_ident(a.attname) ||
7337 regexp_replace(d.adsrc,
7338 $$val\(\(('[^']*')::text\)::regclass$$,
7342 FROM pg_namespace n, pg_class c, pg_attribute a, pg_attrdef d
7343 WHERE n.oid = c.relnamespace AND
7344 c.oid = a.attrelid AND
7345 a.attrelid = d.adrelid AND
7346 a.attnum = d.adnum AND
7347 d.adsrc ~ $$val\(\('[^']*'::text\)::regclass$$;
7349 Next, run the query against a database to find what
7350 adjustments are required, like this for database <literal>db1</>:
7352 psql -t -f fixseq.sql db1
7354 This will show the <command>ALTER TABLE</> commands needed to
7355 convert the database to the newer OID-based representation.
7356 If the commands look reasonable, run this to update the database:
7358 psql -t -f fixseq.sql db1 | psql -e db1
7360 This process must be repeated in each database to be updated.
7366 In <application>psql</application>, treat unquoted
7367 <literal>\{digit}+</> sequences as octal (Bruce)
7370 In previous releases, <literal>\{digit}+</> sequences were
7371 treated as decimal, and only <literal>\0{digit}+</> were treated
7372 as octal. This change was made for consistency.
7378 Remove grammar productions for prefix and postfix <literal>%</>
7379 and <literal>^</> operators
7383 These have never been documented and complicated the use of the
7384 modulus operator (<literal>%</>) with negative numbers.
7390 Make <literal>&<</> and <literal>&></> for polygons
7391 consistent with the box "over" operators (Tom)
7397 <command>CREATE LANGUAGE</> can ignore the provided arguments
7398 in favor of information from <structname>pg_pltemplate</>
7402 A new system catalog <structname>pg_pltemplate</> has been defined
7403 to carry information about the preferred definitions of procedural
7404 languages (such as whether they have validator functions). When
7405 an entry exists in this catalog for the language being created,
7406 <command>CREATE LANGUAGE</> will ignore all its parameters except the
7407 language name and instead use the catalog information. This measure
7408 was taken because of increasing problems with obsolete language
7409 definitions being loaded by old dump files. As of 8.1,
7410 <application>pg_dump</> will dump procedural language definitions as
7411 just <command>CREATE LANGUAGE <replaceable>name</></command>, relying
7412 on a template entry to exist at load time. We expect this will be a
7413 more future-proof representation.
7419 Make <function>pg_cancel_backend(int)</function> return a
7420 <type>boolean</type> rather than an <type>integer</type> (Neil)
7426 Some users are having problems loading UTF-8 data into 8.1.X.
7427 This is because previous versions allowed invalid UTF-8 byte
7428 sequences to be entered into the database, and this release
7429 properly accepts only valid UTF-8 sequences. One way to correct a
7430 dumpfile is to run the command <command>iconv -c -f UTF-8 -t
7431 UTF-8 -o cleanfile.sql dumpfile.sql</>. The <literal>-c</> option
7432 removes invalid character sequences. A diff of the two files will
7433 show the sequences that are invalid. <command>iconv</> reads the
7434 entire input file into memory so it might be necessary to use
7435 <application>split</> to break up the dump into multiple smaller
7436 files for processing.
7444 <title>Additional Changes</title>
7447 Below you will find a detailed account of the additional changes
7448 between <productname>PostgreSQL</productname> 8.1 and the
7449 previous major release.
7453 <title>Performance Improvements</title>
7458 Improve GiST and R-tree index performance (Neil)
7464 Improve the optimizer, including auto-resizing of hash joins
7471 Overhaul internal API in several areas
7477 Change WAL record CRCs from 64-bit to 32-bit (Tom)
7480 We determined that the extra cost of computing 64-bit CRCs was
7481 significant, and the gain in reliability too marginal to justify it.
7487 Prevent writing large empty gaps in WAL pages (Tom)
7493 Improve spinlock behavior on SMP machines, particularly Opterons (Tom)
7499 Allow nonconsecutive index columns to be used in a multicolumn
7503 For example, this allows an index on columns a,b,c to be used in
7504 a query with <command>WHERE a = 4 and c = 10</>.
7510 Skip WAL logging for <command>CREATE TABLE AS</> /
7511 <command>SELECT INTO</> (Simon)
7514 Since a crash during <command>CREATE TABLE AS</> would cause the
7515 table to be dropped during recovery, there is no reason to WAL
7516 log as the table is loaded. (Logging still happens if WAL
7517 archiving is enabled, however.)
7523 Allow concurrent GiST index access (Teodor, Oleg)
7529 Add configuration parameter <varname>full_page_writes</> to
7530 control writing full pages to WAL (Bruce)
7533 To prevent partial disk writes from corrupting the database,
7534 <productname>PostgreSQL</productname> writes a complete copy of
7535 each database disk page to WAL the first time it is modified
7536 after a checkpoint. This option turns off that functionality for more
7537 speed. This is safe to use with battery-backed disk caches where
7538 partial page writes cannot happen.
7544 Use <literal>O_DIRECT</> if available when using
7545 <literal>O_SYNC</> for <varname>wal_sync_method</varname>
7549 <literal>O_DIRECT</> causes disk writes to bypass the kernel
7550 cache, and for WAL writes, this improves performance.
7556 Improve <command>COPY FROM</> performance (Alon Goldshuv)
7559 This was accomplished by reading <command>COPY</> input in
7560 larger chunks, rather than character by character.
7566 Improve the performance of <function>COUNT()</function>,
7567 <function>SUM</function>, <function>AVG()</function>,
7568 <function>STDDEV()</function>, and
7569 <function>VARIANCE()</function> (Neil, Tom)
7576 <title>Server Changes</title>
7581 Prevent problems due to transaction ID (XID) wraparound (Tom)
7584 The server will now warn when the transaction counter approaches
7585 the wraparound point. If the counter becomes too close to wraparound,
7586 the server will stop accepting queries. This ensures that data is
7587 not lost before needed vacuuming is performed.
7593 Fix problems with object IDs (OIDs) conflicting with existing system
7594 objects after the OID counter has wrapped around (Tom)
7600 Add warning about the need to increase
7601 <varname>max_fsm_relations</> and <varname>max_fsm_pages</>
7602 during <command>VACUUM</> (Ron Mayer)
7608 Add <varname>temp_buffers</> configuration parameter to allow
7609 users to determine the size of the local buffer area for
7610 temporary table access (Tom)
7616 Add session start time and client IP address to
7617 <literal>pg_stat_activity</> (Magnus)
7623 Adjust <literal>pg_stat</> views for bitmap scans (Tom)
7626 The meanings of some of the fields have changed slightly.
7632 Enhance <literal>pg_locks</> view (Tom)
7638 Log queries for client-side <command>PREPARE</> and
7639 <command>EXECUTE</> (Simon)
7645 Allow Kerberos name and user name case sensitivity to be
7646 specified in <filename>postgresql.conf</> (Magnus)
7652 Add configuration parameter <varname>krb_server_hostname</> so
7653 that the server host name can be specified as part of service
7654 principal (Todd Kover)
7657 If not set, any service principal matching an entry in the
7658 keytab can be used. This is new Kerberos matching behavior in
7665 Add <varname>log_line_prefix</> options for millisecond
7666 timestamps (<literal>%m</>) and remote host (<literal>%h</>) (Ed
7673 Add WAL logging for GiST indexes (Teodor, Oleg)
7676 GiST indexes are now safe for crash and point-in-time recovery.
7682 Remove old <filename>*.backup</> files when we do
7683 <function>pg_stop_backup()</> (Bruce)
7686 This prevents a large number of <filename>*.backup</> files from
7687 existing in <filename>pg_xlog/</>.
7693 Add configuration parameters to control TCP/IP keep-alive
7694 times for idle, interval, and count (Oliver Jowett)
7698 These values can be changed to allow more rapid detection of
7699 lost client connections.
7705 Add per-user and per-database connection limits (Petr Jelinek)
7708 Using <command>ALTER USER</> and <command>ALTER DATABASE</>,
7709 limits can now be enforced on the maximum number of sessions that
7710 can concurrently connect as a specific user or to a specific database.
7711 Setting the limit to zero disables user or database connections.
7717 Allow more than two gigabytes of shared memory and per-backend
7718 work memory on 64-bit machines (Koichi Suzuki)
7724 New system catalog <structname>pg_pltemplate</> allows overriding
7725 obsolete procedural-language definitions in dump files (Tom)
7734 <title>Query Changes</title>
7739 Add temporary views (Koju Iijima, Neil)
7745 Fix <command>HAVING</> without any aggregate functions or
7746 <command>GROUP BY</> so that the query returns a single group (Tom)
7749 Previously, such a case would treat the <command>HAVING</>
7750 clause the same as a <command>WHERE</> clause. This was not per spec.
7756 Add <command>USING</> clause to allow additional tables to be
7757 specified to <command>DELETE</> (Euler Taveira de Oliveira, Neil)
7760 In prior releases, there was no clear method for specifying
7761 additional tables to be used for joins in a <command>DELETE</>
7762 statement. <command>UPDATE</> already has a <literal>FROM</>
7763 clause for this purpose.
7769 Add support for <literal>\x</> hex escapes in backend and ecpg
7773 This is just like the standard C <literal>\x</> escape syntax.
7774 Octal escapes were already supported.
7780 Add <command>BETWEEN SYMMETRIC</> query syntax (Pavel Stehule)
7783 This feature allows <command>BETWEEN</> comparisons without
7784 requiring the first value to be less than the second. For
7785 example, <command>2 BETWEEN [ASYMMETRIC] 3 AND 1</> returns
7786 false, while <command>2 BETWEEN SYMMETRIC 3 AND 1</> returns
7787 true. <command>BETWEEN ASYMMETRIC</> was already supported.
7793 Add <command>NOWAIT</> option to <command>SELECT ... FOR
7794 UPDATE/SHARE</> (Hans-Juergen Schoenig)
7797 While the <varname>statement_timeout</> configuration
7798 parameter allows a query taking more than a certain amount of
7799 time to be cancelled, the <command>NOWAIT</> option allows a
7800 query to be canceled as soon as a <command>SELECT ... FOR
7801 UPDATE/SHARE</> command cannot immediately acquire a row lock.
7809 <title>Object Manipulation Changes</title>
7814 Track dependencies of shared objects (Alvaro)
7817 <productname>PostgreSQL</productname> allows global tables
7818 (users, databases, tablespaces) to reference information in
7819 multiple databases. This addition adds dependency information
7820 for global tables, so, for example, user ownership can be
7821 tracked across databases, so a user who owns something in any
7822 database can no longer be removed. Dependency tracking already
7823 existed for database-local objects.
7829 Allow limited <command>ALTER OWNER</> commands to be performed
7830 by the object owner (Stephen Frost)
7833 Prior releases allowed only superusers to change object owners.
7834 Now, ownership can be transferred if the user executing the command
7835 owns the object and would be able to create it as the new owner
7836 (that is, the user is a member of the new owning role and that role
7837 has the CREATE permission that would be needed to create the object
7844 Add <command>ALTER</> object <command>SET SCHEMA</> capability
7845 for some object types (tables, functions, types) (Bernd Helmle)
7848 This allows objects to be moved to different schemas.
7854 Add <command>ALTER TABLE ENABLE/DISABLE TRIGGER</command> to
7855 disable triggers (Satoshi Nagayasu)
7864 <title>Utility Command Changes</title>
7869 Allow <command>TRUNCATE</> to truncate multiple tables in a
7870 single command (Alvaro)
7873 Because of referential integrity checks, it is not allowed to
7874 truncate a table that is part of a referential integrity
7875 constraint. Using this new functionality, <command>TRUNCATE</>
7876 can be used to truncate such tables, if both tables involved in
7877 a referential integrity constraint are truncated in a single
7878 <command>TRUNCATE</> command.
7884 Properly process carriage returns and line feeds in
7885 <command>COPY CSV</> mode (Andrew)
7888 In release 8.0, carriage returns and line feeds in <command>CSV
7889 COPY TO</> were processed in an inconsistent manner. (This was
7890 documented on the TODO list.)
7896 Add <command>COPY WITH CSV HEADER</> to allow a header line as
7897 the first line in <command>COPY</> (Andrew)
7900 This allows handling of the common <command>CSV</> usage of
7901 placing the column names on the first line of the data file. For
7902 <command>COPY TO</>, the first line contains the column names,
7903 and for <command>COPY FROM</>, the first line is ignored.
7909 On Windows, display better sub-second precision in
7910 <command>EXPLAIN ANALYZE</> (Magnus)
7916 Add trigger duration display to <command>EXPLAIN ANALYZE</>
7920 Prior releases included trigger execution time as part of the
7921 total execution time, but did not show it separately. It is now
7922 possible to see how much time is spent in each trigger.
7928 Add support for <literal>\x</> hex escapes in <command>COPY</>
7932 Previous releases only supported octal escapes.
7938 Make <command>SHOW ALL</> include variable descriptions
7942 <command>SHOW</> varname still only displays the variable's
7943 value and does not include the description.
7949 Make <application>initdb</application> create a new standard
7950 database called <literal>postgres</>, and convert utilities to
7951 use <literal>postgres</> rather than <literal>template1</> for
7952 standard lookups (Dave)
7955 In prior releases, <literal>template1</> was used both as a
7956 default connection for utilities like
7957 <application>createuser</application>, and as a template for
7958 new databases. This caused <command>CREATE DATABASE</> to
7959 sometimes fail, because a new database cannot be created if
7960 anyone else is in the template database. With this change, the
7961 default connection database is now <literal>postgres</>,
7962 meaning it is much less likely someone will be using
7963 <literal>template1</> during <command>CREATE DATABASE</>.
7969 Create new <application>reindexdb</application> command-line
7970 utility by moving <filename>/contrib/reindexdb</> into the
7971 server (Euler Taveira de Oliveira)
7980 <title>Data Type and Function Changes</title>
7985 Add <function>MAX()</> and <function>MIN()</> aggregates for
7986 array types (Koju Iijima)
7992 Fix <function>to_date()</> and <function>to_timestamp()</> to
7993 behave reasonably when <literal>CC</> and <literal>YY</> fields
7994 are both used (Karel Zak)
7997 If the format specification contains <literal>CC</> and a year
7998 specification is <literal>YYY</> or longer, ignore the
7999 <literal>CC</>. If the year specification is <literal>YY</> or
8000 shorter, interpret <literal>CC</> as the previous century.
8006 Add <function>md5(bytea)</> (Abhijit Menon-Sen)
8009 <function>md5(text)</> already existed.
8015 Add support for <command>numeric ^ numeric</> based on
8016 <function>power(numeric, numeric)</>
8019 The function already existed, but there was no operator assigned
8026 Fix <type>NUMERIC</> modulus by properly truncating the quotient
8027 during computation (Bruce)
8030 In previous releases, modulus for large values sometimes
8031 returned negative results due to rounding of the quotient.
8037 Add a function <function>lastval()</> (Dennis Björklund)
8040 <function>lastval()</> is a simplified version of
8041 <function>currval()</>. It automatically determines the proper
8042 sequence name based on the most recent <function>nextval()</> or
8043 <function>setval()</> call performed by the current session.
8049 Add <function>to_timestamp(DOUBLE PRECISION) (Michael Glaesemann)</>
8052 Converts Unix seconds since 1970 to a <type>TIMESTAMP WITH
8059 Add <function>pg_postmaster_start_time()</> function (Euler
8060 Taveira de Oliveira, Matthias Schmidt)
8066 Allow the full use of time zone names in <command>AT TIME
8067 ZONE</>, not just the short list previously available (Magnus)
8070 Previously, only a predefined list of time zone names were
8071 supported by <command>AT TIME ZONE</>. Now any supported time
8072 zone name can be used, e.g.:
8074 SELECT CURRENT_TIMESTAMP AT TIME ZONE 'Europe/London';
8076 In the above query, the time zone used is adjusted based on the
8077 daylight saving time rules that were in effect on the supplied
8084 Add <function>GREATEST()</> and <function>LEAST()</> variadic
8085 functions (Pavel Stehule)
8088 These functions take a variable number of arguments and return
8089 the greatest or least value among the arguments.
8095 Add <function>pg_column_size()</> (Mark Kirkwood)
8098 This returns storage size of a column, which might be compressed.
8104 Add <function>regexp_replace()</> (Atsushi Ogawa)
8107 This allows regular expression replacement, like sed. An optional
8108 flag argument allows selection of global (replace all) and
8109 case-insensitive modes.
8115 Fix interval division and multiplication (Bruce)
8118 Previous versions sometimes returned unjustified results, like
8119 <command>'4 months'::interval / 5</> returning <command>'1 mon
8126 Fix roundoff behavior in timestamp, time, and interval output (Tom)
8129 This fixes some cases in which the seconds field would be shown as
8130 <literal>60</> instead of incrementing the higher-order fields.
8136 Add a separate day field to type <type>interval</> so a one day
8137 interval can be distinguished from a 24 hour interval (Michael
8141 Days that contain a daylight saving time adjustment are not 24
8142 hours long, but typically 23 or 25 hours. This change creates a
8143 conceptual distinction between intervals of <quote>so many days</>
8144 and intervals of <quote>so many hours</>. Adding
8145 <literal>1 day</> to a timestamp now gives the same local time on
8146 the next day even if a daylight saving time adjustment occurs
8147 between, whereas adding <literal>24 hours</> will give a different
8148 local time when this happens. For example, under US DST rules:
8150 '2005-04-03 00:00:00-05' + '1 day' = '2005-04-04 00:00:00-04'
8151 '2005-04-03 00:00:00-05' + '24 hours' = '2005-04-04 01:00:00-04'
8158 Add <function>justify_days()</> and <function>justify_hours()</>
8159 (Michael Glaesemann)
8162 These functions, respectively, adjust days to an appropriate
8163 number of full months and days, and adjust hours to an
8164 appropriate number of full days and hours.
8170 Move <filename>/contrib/dbsize</> into the backend, and rename
8171 some of the functions (Dave Page, Andreas Pflug)
8178 <function>pg_tablespace_size()</>
8184 <function>pg_database_size()</>
8190 <function>pg_relation_size()</>
8196 <function>pg_total_relation_size()</>
8202 <function>pg_size_pretty()</>
8209 <function>pg_total_relation_size()</> includes indexes and TOAST
8216 Add functions for read-only file access to the cluster directory
8217 (Dave Page, Andreas Pflug)
8224 <function>pg_stat_file()</>
8230 <function>pg_read_file()</>
8236 <function>pg_ls_dir()</>
8246 Add <function>pg_reload_conf()</> to force reloading of the
8247 configuration files (Dave Page, Andreas Pflug)
8253 Add <function>pg_rotate_logfile()</> to force rotation of the
8254 server log file (Dave Page, Andreas Pflug)
8260 Change <literal>pg_stat_*</> views to include TOAST tables (Tom)
8269 <title>Encoding and Locale Changes</title>
8274 Rename some encodings to be more consistent and to follow
8275 international standards (Bruce)
8282 <literal>UNICODE</> is now <literal>UTF8</>
8288 <literal>ALT</> is now <literal>WIN866</>
8294 <literal>WIN</> is now <literal>WIN1251</>
8300 <literal>TCVN</> is now <literal>WIN1258</>
8308 The original names still work.
8314 Add support for <literal>WIN1252</> encoding (Roland Volkmann)
8320 Add support for four-byte <literal>UTF8</> characters (John
8324 Previously only one, two, and three-byte <literal>UTF8</> characters
8325 were supported. This is particularly important for support for
8326 some Chinese character sets.
8332 Allow direct conversion between <literal>EUC_JP</> and
8333 <literal>SJIS</> to improve performance (Atsushi Ogawa)
8339 Allow the UTF8 encoding to work on Windows (Magnus)
8342 This is done by mapping UTF8 to the Windows-native UTF16
8352 <title>General Server-Side Language Changes</title>
8357 Fix <command>ALTER LANGUAGE RENAME</> (Sergey Yatskevich)
8363 Allow function characteristics, like strictness and volatility,
8364 to be modified via <command>ALTER FUNCTION</> (Neil)
8370 Increase the maximum number of function arguments to 100 (Tom)
8376 Allow SQL and PL/PgSQL functions to use <command>OUT</> and
8377 <command>INOUT</> parameters (Tom)
8380 <command>OUT</> is an alternate way for a function to return
8381 values. Instead of using <command>RETURN</>, values can be
8382 returned by assigning to parameters declared as <command>OUT</> or
8383 <command>INOUT</>. This is notationally simpler in some cases,
8384 particularly so when multiple values need to be returned.
8385 While returning multiple values from a function
8386 was possible in previous releases, this greatly simplifies the
8387 process. (The feature will be extended to other server-side
8388 languages in future releases.)
8394 Move language handler functions into the <literal>pg_catalog</> schema
8397 This makes it easier to drop the public schema if desired.
8403 Add <function>SPI_getnspname()</function> to SPI (Neil)
8411 <title>PL/PgSQL Server-Side Language Changes</title>
8416 Overhaul the memory management of PL/PgSQL functions (Neil)
8419 The parsetree of each function is now stored in a separate
8420 memory context. This allows this memory to be easily reclaimed
8421 when it is no longer needed.
8427 Check function syntax at <command>CREATE FUNCTION</> time,
8428 rather than at runtime (Neil)
8431 Previously, most syntax errors were reported only when the
8432 function was executed.
8438 Allow <command>OPEN</> to open non-<command>SELECT</> queries
8439 like <command>EXPLAIN</> and <command>SHOW</> (Tom)
8445 No longer require functions to issue a <command>RETURN</>
8449 This is a byproduct of the newly added <command>OUT</> and
8450 <command>INOUT</> functionality. <command>RETURN</> can
8451 be omitted when it is not needed to provide the function's
8458 Add support for an optional <command>INTO</> clause to
8459 PL/PgSQL's <command>EXECUTE</> statement (Pavel Stehule, Neil)
8465 Make <command>CREATE TABLE AS</> set <command>ROW_COUNT</> (Tom)
8471 Define <literal>SQLSTATE</> and <literal>SQLERRM</> to return
8472 the <literal>SQLSTATE</> and error message of the current
8473 exception (Pavel Stehule, Neil)
8476 These variables are only defined inside exception blocks.
8482 Allow the parameters to the <command>RAISE</> statement to be
8483 expressions (Pavel Stehule, Neil)
8489 Add a loop <command>CONTINUE</> statement (Pavel Stehule, Neil)
8495 Allow block and loop labels (Pavel Stehule)
8504 <title>PL/Perl Server-Side Language Changes</title>
8509 Allow large result sets to be returned efficiently (Abhijit
8513 This allows functions to use <function>return_next()</> to avoid
8514 building the entire result set in memory.
8520 Allow one-row-at-a-time retrieval of query results (Abhijit Menon-Sen)
8523 This allows functions to use <function>spi_query()</> and
8524 <function>spi_fetchrow()</> to avoid accumulating the entire
8525 result set in memory.
8531 Force PL/Perl to handle strings as <literal>UTF8</> if the
8532 server encoding is <literal>UTF8</> (David Kamholz)
8538 Add a validator function for PL/Perl (Andrew)
8541 This allows syntax errors to be reported at definition time,
8542 rather than execution time.
8548 Allow PL/Perl to return a Perl array when the function returns
8549 an array type (Andrew)
8552 This basically maps <productname>PostgreSQL</productname> arrays
8559 Allow Perl nonfatal warnings to generate <command>NOTICE</>
8566 Allow Perl's <literal>strict</> mode to be enabled (Andrew)
8575 <title><application>psql</> Changes</title>
8580 Add <command>\set ON_ERROR_ROLLBACK</> to allow statements in
8581 a transaction to error without affecting the rest of the
8582 transaction (Greg Sabino Mullane)
8585 This is basically implemented by wrapping every statement in a
8592 Add support for <literal>\x</> hex strings in
8593 <application>psql</> variables (Bruce)
8596 Octal escapes were already supported.
8602 Add support for <command>troff -ms</> output format (Roger
8609 Allow the history file location to be controlled by
8610 <envar>HISTFILE</> (Andreas Seltenreich)
8613 This allows configuration of per-database history storage.
8619 Prevent <command>\x</> (expanded mode) from affecting
8620 the output of <command>\d tablename</> (Neil)
8626 Add <option>-L</> option to <application>psql</application> to
8627 log sessions (Lorne Sunley)
8630 This option was added because some operating systems do not have
8631 simple command-line activity logging functionality.
8637 Make <command>\d</> show the tablespaces of indexes (Qingqing
8644 Allow <application>psql</application> help (<command>\h</>) to
8645 make a best guess on the proper help information (Greg Sabino
8649 This allows the user to just add <command>\h</> to the front of
8650 the syntax error query and get help on the supported syntax.
8651 Previously any additional query text beyond the command name
8652 had to be removed to use <command>\h</>.
8658 Add <command>\pset numericlocale</> to allow numbers to be
8659 output in a locale-aware format (Eugen Nedelcu)
8662 For example, using <literal>C</> locale <literal>100000</> would
8663 be output as <literal>100,000.0</> while a European locale might
8664 output this value as <literal>100.000,0</>.
8670 Make startup banner show both server version number and
8671 <application>psql</>'s version number, when they are different (Bruce)
8674 Also, a warning will be shown if the server and <application>psql</>
8675 are from different major releases.
8684 <title><application>pg_dump</> Changes</title>
8689 Add <option>-n</> / <option>--schema</> switch to
8690 <application>pg_restore</> (Richard van den Berg)
8693 This allows just the objects in a specified schema to be restored.
8699 Allow <application>pg_dump</> to dump large objects even in
8703 With this change, large objects are now always dumped; the former
8704 <option>-b</> switch is a no-op.
8710 Allow <application>pg_dump</> to dump a consistent snapshot of
8717 Dump comments for large objects (Tom)
8723 Add <option>--encoding</> to <application>pg_dump</>
8727 This allows a database to be dumped in an encoding that is
8728 different from the server's encoding. This is valuable when
8729 transferring the dump to a machine with a different encoding.
8735 Rely on <structname>pg_pltemplate</> for procedural languages (Tom)
8738 If the call handler for a procedural language is in the
8739 <literal>pg_catalog</> schema, <application>pg_dump</> does not
8740 dump the handler. Instead, it dumps the language using just
8741 <command>CREATE LANGUAGE <replaceable>name</></command>,
8742 relying on the <structname>pg_pltemplate</> catalog to provide
8743 the language's creation parameters at load time.
8752 <title><application>libpq</application> Changes</title>
8757 Add a <envar>PGPASSFILE</> environment variable to specify the
8758 password file's filename (Andrew)
8764 Add <function>lo_create()</>, that is similar to
8765 <function>lo_creat()</> but allows the OID of the large object
8766 to be specified (Tom)
8772 Make <application>libpq</application> consistently return an error
8773 to the client application on <function>malloc()</function>
8782 <title>Source Code Changes</title>
8787 Fix <application>pgxs</> to support building against a relocated
8794 Add spinlock support for the Itanium processor using Intel
8795 compiler (Vikram Kalsi)
8801 Add Kerberos 5 support for Windows (Magnus)
8807 Add Chinese FAQ (laser@pgsqldb.com)
8813 Rename Rendezvous to Bonjour to match OS/X feature renaming
8820 Add support for <literal>fsync_writethrough</literal> on
8821 Darwin (Chris Campbell)
8827 Streamline the passing of information within the server, the
8828 optimizer, and the lock system (Tom)
8834 Allow <application>pg_config</> to be compiled using MSVC (Andrew)
8837 This is required to build DBD::Pg using <application>MSVC</>.
8843 Remove support for Kerberos V4 (Magnus)
8846 Kerberos 4 had security vulnerabilities and is no longer
8853 Code cleanups (Coverity static analysis performed by
8860 Modify <filename>postgresql.conf</> to use documentation defaults
8861 <literal>on</>/<literal>off</> rather than
8862 <literal>true</>/<literal>false</> (Bruce)
8868 Enhance <application>pg_config</> to be able to report more
8869 build-time values (Tom)
8875 Allow <application>libpq</application> to be built thread-safe
8876 on Windows (Dave Page)
8882 Allow IPv6 connections to be used on Windows (Andrew)
8888 Add Server Administration documentation about I/O subsystem
8895 Move private declarations from <filename>gist.h</filename> to
8896 <filename>gist_private.h</filename> (Neil)
8900 In previous releases, <filename>gist.h</> contained both the
8901 public GiST API (intended for use by authors of GiST index
8902 implementations) as well as some private declarations used by
8903 the implementation of GiST itself. The latter have been moved
8904 to a separate file, <filename>gist_private.h</>. Most GiST
8905 index implementations should be unaffected.
8911 Overhaul GiST memory management (Neil)
8915 GiST methods are now always invoked in a short-lived memory
8916 context. Therefore, memory allocated via <function>palloc()</>
8917 will be reclaimed automatically, so GiST index implementations
8918 do not need to manually release allocated memory via
8919 <function>pfree()</>.
8927 <title>Contrib Changes</title>
8932 Add <filename>/contrib/pg_buffercache</> contrib module (Mark
8936 This displays the contents of the buffer cache, for debugging and
8937 performance tuning purposes.
8943 Remove <filename>/contrib/array</> because it is obsolete (Tom)
8949 Clean up the <filename>/contrib/lo</> module (Tom)
8955 Move <filename>/contrib/findoidjoins</> to
8956 <filename>/src/tools</> (Tom)
8962 Remove the <literal><<</>, <literal>>></>,
8963 <literal>&<</>, and <literal>&></> operators from
8964 <filename>/contrib/cube</>
8967 These operators were not useful.
8973 Improve <filename>/contrib/btree_gist</> (Janko Richter)
8979 Improve <filename>/contrib/pgbench</> (Tomoaki Sato, Tatsuo)
8982 There is now a facility for testing with SQL command scripts given
8983 by the user, instead of only a hard-wired command sequence.
8989 Improve <filename>/contrib/pgcrypto</> (Marko Kreen)
8996 Implementation of OpenPGP symmetric-key and public-key encryption
8999 Both RSA and Elgamal public-key algorithms are supported.
9005 Stand alone build: include SHA256/384/512 hashes, Fortuna PRNG
9011 OpenSSL build: support 3DES, use internal AES with OpenSSL < 0.9.7
9017 Take build parameters (OpenSSL, zlib) from <filename>configure</> result
9020 There is no need to edit the <filename>Makefile</> anymore.
9026 Remove support for <filename>libmhash</> and <filename>libmcrypt</>
9039 <sect1 id="release-8-0-14">
9040 <title>Release 8.0.14</title>
9043 <title>Release date</title>
9044 <simpara>2007-09-17</simpara>
9048 This release contains a variety of fixes from 8.0.13.
9052 <title>Migration to Version 8.0.14</title>
9055 A dump/restore is not required for those running 8.0.X. However,
9056 if you are upgrading from a version earlier than 8.0.6, see the release
9063 <title>Changes</title>
9069 Prevent index corruption when a transaction inserts rows and
9070 then aborts close to the end of a concurrent <command>VACUUM</>
9071 on the same table (Tom)
9077 Make <command>CREATE DOMAIN ... DEFAULT NULL</> work properly (Tom)
9083 Fix excessive logging of <acronym>SSL</> error messages (Tom)
9089 Fix logging so that log messages are never interleaved when using
9090 the syslogger process (Andrew)
9096 Fix crash when <varname>log_min_error_statement</> logging runs out
9103 Fix incorrect handling of some foreign-key corner cases (Tom)
9109 Prevent <command>CLUSTER</> from failing
9110 due to attempting to process temporary tables of other sessions (Alvaro)
9116 Update the time zone database rules, particularly New Zealand's upcoming changes (Tom)
9122 Windows socket improvements (Magnus)
9128 Suppress timezone name (<literal>%Z</>) in log timestamps on Windows
9129 because of possible encoding mismatches (Tom)
9135 Require non-superusers who use <filename>/contrib/dblink</> to use only
9136 password authentication, as a security measure (Joe)
9145 <sect1 id="release-8-0-13">
9146 <title>Release 8.0.13</title>
9149 <title>Release date</title>
9150 <simpara>2007-04-23</simpara>
9154 This release contains a variety of fixes from 8.0.12,
9155 including a security fix.
9159 <title>Migration to Version 8.0.13</title>
9162 A dump/restore is not required for those running 8.0.X. However,
9163 if you are upgrading from a version earlier than 8.0.6, see the release
9170 <title>Changes</title>
9176 Support explicit placement of the temporary-table schema within
9177 <varname>search_path</>, and disable searching it for functions
9181 This is needed to allow a security-definer function to set a
9182 truly secure value of <varname>search_path</>. Without it,
9183 an unprivileged SQL user can use temporary objects to execute code
9184 with the privileges of the security-definer function (CVE-2007-2138).
9185 See <command>CREATE FUNCTION</> for more information.
9191 <filename>/contrib/tsearch2</> crash fixes (Teodor)
9197 Fix potential-data-corruption bug in how <command>VACUUM FULL</> handles
9198 <command>UPDATE</> chains (Tom, Pavan Deolasee)
9204 Fix PANIC during enlargement of a hash index (bug introduced in 8.0.10)
9211 Fix POSIX-style timezone specs to follow new USA DST rules (Tom)
9220 <sect1 id="release-8-0-12">
9221 <title>Release 8.0.12</title>
9224 <title>Release date</title>
9225 <simpara>2007-02-07</simpara>
9229 This release contains one fix from 8.0.11.
9233 <title>Migration to Version 8.0.12</title>
9236 A dump/restore is not required for those running 8.0.X. However,
9237 if you are upgrading from a version earlier than 8.0.6, see the release
9244 <title>Changes</title>
9250 Remove overly-restrictive check for type length in constraints and
9251 functional indexes(Tom)
9260 <sect1 id="release-8-0-11">
9261 <title>Release 8.0.11</title>
9264 <title>Release date</title>
9265 <simpara>2007-02-05</simpara>
9269 This release contains a variety of fixes from 8.0.10, including
9274 <title>Migration to Version 8.0.11</title>
9277 A dump/restore is not required for those running 8.0.X. However,
9278 if you are upgrading from a version earlier than 8.0.6, see the release
9285 <title>Changes</title>
9291 Remove security vulnerabilities that allowed connected users
9292 to read backend memory (Tom)
9295 The vulnerabilities involve suppressing the normal check that a SQL
9296 function returns the data type it's declared to, and changing the
9297 data type of a table column (CVE-2007-0555, CVE-2007-0556). These
9298 errors can easily be exploited to cause a backend crash, and in
9299 principle might be used to read database content that the user
9300 should not be able to access.
9306 Fix rare bug wherein btree index page splits could fail
9307 due to choosing an infeasible split point (Heikki Linnakangas)
9313 Fix for rare Assert() crash triggered by <literal>UNION</> (Tom)
9319 Tighten security of multi-byte character processing for UTF8 sequences
9320 over three bytes long (Tom)
9329 <sect1 id="release-8-0-10">
9330 <title>Release 8.0.10</title>
9333 <title>Release date</title>
9334 <simpara>2007-01-08</simpara>
9338 This release contains a variety of fixes from 8.0.9.
9342 <title>Migration to Version 8.0.10</title>
9345 A dump/restore is not required for those running 8.0.X. However,
9346 if you are upgrading from a version earlier than 8.0.6, see the release
9353 <title>Changes</title>
9359 Improve handling of <function>getaddrinfo()</> on AIX (Tom)
9363 This fixes a problem with starting the statistics collector,
9370 Fix <quote>failed to re-find parent key</> errors in
9371 <command>VACUUM</> (Tom)
9377 Fix race condition for truncation of a large relation across a
9378 gigabyte boundary by <command>VACUUM</> (Tom)
9384 Fix bugs affecting multi-gigabyte hash indexes (Tom)
9390 Fix possible deadlock in Windows signal handling (Teodor)
9396 Fix error when constructing an <literal>ARRAY[]</> made up of multiple
9397 empty elements (Tom)
9403 Fix ecpg memory leak during connection (Michael)
9409 <function>to_number()</> and <function>to_char(numeric)</>
9410 are now <literal>STABLE</>, not <literal>IMMUTABLE</>, for
9411 new <application>initdb</> installs (Tom)
9415 This is because <varname>lc_numeric</> can potentially
9416 change the output of these functions.
9422 Improve index usage of regular expressions that use parentheses (Tom)
9426 This improves <application>psql</> <literal>\d</> performance also.
9432 Update timezone database
9436 This affects Australian and Canadian daylight-savings rules in
9446 <sect1 id="release-8-0-9">
9447 <title>Release 8.0.9</title>
9450 <title>Release date</title>
9451 <simpara>2006-10-16</simpara>
9455 This release contains a variety of fixes from 8.0.8.
9459 <title>Migration to Version 8.0.9</title>
9462 A dump/restore is not required for those running 8.0.X. However,
9463 if you are upgrading from a version earlier than 8.0.6, see the release
9470 <title>Changes</title>
9473 <listitem><para>Fix crash when referencing <literal>NEW</> row
9474 values in rule WHERE expressions (Tom)</para></listitem>
9475 <listitem><para>Fix core dump when an untyped literal is taken as
9476 ANYARRAY</para></listitem>
9477 <listitem><para>Fix mishandling of AFTER triggers when query contains a SQL
9478 function returning multiple rows (Tom)</para></listitem>
9479 <listitem><para>Fix <command>ALTER TABLE ... TYPE</> to recheck
9480 <literal>NOT NULL</> for <literal>USING</> clause (Tom)</para></listitem>
9481 <listitem><para>Fix <function>string_to_array()</> to handle overlapping
9482 matches for the separator string</para>
9483 <para>For example, <literal>string_to_array('123xx456xxx789', 'xx')</>.
9485 <listitem><para>Fix corner cases in pattern matching for
9486 <application>psql</>'s <literal>\d</> commands</para></listitem>
9487 <listitem><para>Fix index-corrupting bugs in /contrib/ltree
9488 (Teodor)</para></listitem>
9489 <listitem><para>Numerous robustness fixes in <application>ecpg</> (Joachim
9490 Wieland)</para></listitem>
9491 <listitem><para>Fix backslash escaping in /contrib/dbmirror</para></listitem>
9492 <listitem><para>Fix instability of statistics collection on Win32 (Tom, Andrew)</para></listitem>
9493 <listitem><para>Fixes for <systemitem class="osname">AIX</> and
9494 <productname>Intel</> compilers (Tom)</para></listitem>
9500 <sect1 id="release-8-0-8">
9501 <title>Release 8.0.8</title>
9504 <title>Release date</title>
9505 <simpara>2006-05-23</simpara>
9509 This release contains a variety of fixes from 8.0.7,
9510 including patches for extremely serious security issues.
9514 <title>Migration to Version 8.0.8</title>
9517 A dump/restore is not required for those running 8.0.X. However,
9518 if you are upgrading from a version earlier than 8.0.6, see the release
9523 Full security against the SQL-injection attacks described in
9524 CVE-2006-2313 and CVE-2006-2314 might require changes in application
9525 code. If you have applications that embed untrustworthy strings
9526 into SQL commands, you should examine them as soon as possible to
9527 ensure that they are using recommended escaping techniques. In
9528 most cases, applications should be using subroutines provided by
9529 libraries or drivers (such as <application>libpq</>'s
9530 <function>PQescapeStringConn()</>) to perform string escaping,
9531 rather than relying on <foreignphrase>ad hoc</> code to do it.
9536 <title>Changes</title>
9539 <listitem><para>Change the server to reject invalidly-encoded multibyte
9540 characters in all cases (Tatsuo, Tom)</para>
9541 <para>While <productname>PostgreSQL</> has been moving in this direction for
9542 some time, the checks are now applied uniformly to all encodings and all
9543 textual input, and are now always errors not merely warnings. This change
9544 defends against SQL-injection attacks of the type described in CVE-2006-2313.
9547 <listitem><para>Reject unsafe uses of <literal>\'</> in string literals</para>
9548 <para>As a server-side defense against SQL-injection attacks of the type
9549 described in CVE-2006-2314, the server now only accepts <literal>''</> and not
9550 <literal>\'</> as a representation of ASCII single quote in SQL string
9551 literals. By default, <literal>\'</> is rejected only when
9552 <varname>client_encoding</> is set to a client-only encoding (SJIS, BIG5, GBK,
9553 GB18030, or UHC), which is the scenario in which SQL injection is possible.
9554 A new configuration parameter <varname>backslash_quote</> is available to
9555 adjust this behavior when needed. Note that full security against
9556 CVE-2006-2314 might require client-side changes; the purpose of
9557 <varname>backslash_quote</> is in part to make it obvious that insecure
9558 clients are insecure.
9561 <listitem><para>Modify <application>libpq</>'s string-escaping routines to be
9562 aware of encoding considerations and
9563 <varname>standard_conforming_strings</></para>
9564 <para>This fixes <application>libpq</>-using applications for the security
9565 issues described in CVE-2006-2313 and CVE-2006-2314, and also future-proofs
9566 them against the planned changeover to SQL-standard string literal syntax.
9567 Applications that use multiple <productname>PostgreSQL</> connections
9568 concurrently should migrate to <function>PQescapeStringConn()</> and
9569 <function>PQescapeByteaConn()</> to ensure that escaping is done correctly
9570 for the settings in use in each database connection. Applications that
9571 do string escaping <quote>by hand</> should be modified to rely on library
9575 <listitem><para>Fix some incorrect encoding conversion functions</para>
9576 <para><function>win1251_to_iso</>, <function>alt_to_iso</>,
9577 <function>euc_tw_to_big5</>, <function>euc_tw_to_mic</>,
9578 <function>mic_to_euc_tw</> were all broken to varying
9582 <listitem><para>Clean up stray remaining uses of <literal>\'</> in strings
9583 (Bruce, Jan)</para></listitem>
9585 <listitem><para>Fix bug that sometimes caused OR'd index scans to
9586 miss rows they should have returned</para></listitem>
9588 <listitem><para>Fix WAL replay for case where a btree index has been
9589 truncated</para></listitem>
9591 <listitem><para>Fix <literal>SIMILAR TO</> for patterns involving
9592 <literal>|</> (Tom)</para></listitem>
9594 <listitem><para>Fix <command>SELECT INTO</> and <command>CREATE TABLE AS</> to
9595 create tables in the default tablespace, not the base directory (Kris
9596 Jurka)</para></listitem>
9598 <listitem><para>Fix server to use custom DH SSL parameters correctly (Michael
9599 Fuhr)</para></listitem>
9601 <listitem><para>Fix for Bonjour on Intel Macs (Ashley Clark)</para></listitem>
9603 <listitem><para>Fix various minor memory leaks</para></listitem>
9605 <listitem><para>Fix problem with password prompting on some Win32 systems
9606 (Robert Kinberg)</para></listitem>
9612 <sect1 id="release-8-0-7">
9613 <title>Release 8.0.7</title>
9616 <title>Release date</title>
9617 <simpara>2006-02-14</simpara>
9621 This release contains a variety of fixes from 8.0.6.
9625 <title>Migration to Version 8.0.7</title>
9628 A dump/restore is not required for those running 8.0.X. However,
9629 if you are upgrading from a version earlier than 8.0.6, see the release
9635 <title>Changes</title>
9639 <listitem><para>Fix potential crash in <command>SET
9640 SESSION AUTHORIZATION</> (CVE-2006-0553)</para>
9641 <para>An unprivileged user could crash the server process, resulting in
9642 momentary denial of service to other users, if the server has been compiled
9643 with Asserts enabled (which is not the default).
9644 Thanks to Akio Ishida for reporting this problem.
9647 <listitem><para>Fix bug with row visibility logic in self-inserted
9649 <para>Under rare circumstances a row inserted by the current command
9650 could be seen as already valid, when it should not be. Repairs bug
9651 created in 8.0.4, 7.4.9, and 7.3.11 releases.
9654 <listitem><para>Fix race condition that could lead to <quote>file already
9655 exists</> errors during pg_clog and pg_subtrans file creation
9656 (Tom)</para></listitem>
9658 <listitem><para>Fix cases that could lead to crashes if a cache-invalidation
9659 message arrives at just the wrong time (Tom)</para></listitem>
9661 <listitem><para>Properly check <literal>DOMAIN</> constraints for
9662 <literal>UNKNOWN</> parameters in prepared statements
9663 (Neil)</para></listitem>
9665 <listitem><para>Ensure <command>ALTER COLUMN TYPE</> will process
9666 <literal>FOREIGN KEY</>, <literal>UNIQUE</>, and <literal>PRIMARY KEY</>
9667 constraints in the proper order (Nakano Yoshihisa)</para></listitem>
9669 <listitem><para>Fixes to allow restoring dumps that have cross-schema
9670 references to custom operators or operator classes (Tom)</para></listitem>
9672 <listitem><para>Allow <application>pg_restore</> to continue properly after a
9673 <command>COPY</> failure; formerly it tried to treat the remaining
9674 <command>COPY</> data as SQL commands (Stephen Frost)</para></listitem>
9676 <listitem><para>Fix <application>pg_ctl</> <literal>unregister</> crash
9677 when the data directory is not specified (Magnus)</para></listitem>
9679 <listitem><para>Fix <application>ecpg</> crash on AMD64 and PPC
9680 (Neil)</para></listitem>
9682 <listitem><para>Recover properly if error occurs during argument passing
9683 in <application>PL/python</> (Neil)</para></listitem>
9685 <listitem><para>Fix <application>PL/perl</>'s handling of locales on
9686 Win32 to match the backend (Andrew)</para></listitem>
9688 <listitem><para>Fix crash when <literal>log_min_messages</> is set to
9689 <literal>DEBUG3</> or above in <filename>postgresql.conf</> on Win32
9690 (Bruce)</para></listitem>
9692 <listitem><para>Fix <application>pgxs</> <literal>-L</> library path
9693 specification for Win32, Cygwin, OS X, AIX (Bruce)</para></listitem>
9695 <listitem><para>Check that SID is enabled while checking for Win32 admin
9696 privileges (Magnus)</para></listitem>
9698 <listitem><para>Properly reject out-of-range date inputs (Kris
9699 Jurka)</para></listitem>
9701 <listitem><para>Portability fix for testing presence of <function>finite</>
9702 and <function>isinf</> during configure (Tom)</para></listitem>
9709 <sect1 id="release-8-0-6">
9710 <title>Release 8.0.6</title>
9713 <title>Release date</title>
9714 <simpara>2006-01-09</simpara>
9718 This release contains a variety of fixes from 8.0.5.
9722 <title>Migration to Version 8.0.6</title>
9725 A dump/restore is not required for those running 8.0.X. However,
9726 if you are upgrading from a version earlier than 8.0.3, see the release
9728 Also, you might need to <command>REINDEX</> indexes on textual
9729 columns after updating, if you are affected by the locale or
9730 <application>plperl</> issues described below.
9735 <title>Changes</title>
9739 <listitem><para>Fix Windows code so that postmaster will continue rather
9740 than exit if there is no more room in ShmemBackendArray (Magnus)</para>
9741 <para>The previous behavior could lead to a denial-of-service situation if too
9742 many connection requests arrive close together. This applies
9743 <emphasis>only</> to the Windows port.</para></listitem>
9745 <listitem><para>Fix bug introduced in 8.0 that could allow ReadBuffer
9746 to return an already-used page as new, potentially causing loss of
9747 recently-committed data (Tom)</para></listitem>
9749 <listitem><para>Fix for protocol-level Describe messages issued
9750 outside a transaction or in a failed transaction (Tom)</para></listitem>
9752 <listitem><para>Fix character string comparison for locales that consider
9753 different character combinations as equal, such as Hungarian (Tom)</para>
9754 <para>This might require <command>REINDEX</> to fix existing indexes on
9755 textual columns.</para></listitem>
9757 <listitem><para>Set locale environment variables during postmaster startup
9758 to ensure that <application>plperl</> won't change the locale later</para>
9759 <para>This fixes a problem that occurred if the <application>postmaster</> was
9760 started with environment variables specifying a different locale than what
9761 <application>initdb</> had been told. Under these conditions, any use of
9762 <application>plperl</> was likely to lead to corrupt indexes. You might need
9763 <command>REINDEX</> to fix existing indexes on
9764 textual columns if this has happened to you.</para></listitem>
9766 <listitem><para>Allow more flexible relocation of installation
9767 directories (Tom)</para>
9768 <para>Previous releases supported relocation only if all installation
9769 directory paths were the same except for the last component.</para></listitem>
9771 <listitem><para>Fix longstanding bug in strpos() and regular expression
9772 handling in certain rarely used Asian multi-byte character sets (Tatsuo)
9775 <listitem><para>Various fixes for functions returning <literal>RECORD</>s
9776 (Tom) </para></listitem>
9778 <listitem><para>Fix bug in <filename>/contrib/pgcrypto</> gen_salt,
9779 which caused it not to use all available salt space for MD5 and
9780 XDES algorithms (Marko Kreen, Solar Designer)</para>
9781 <para>Salts for Blowfish and standard DES are unaffected.</para></listitem>
9783 <listitem><para>Fix <filename>/contrib/dblink</> to throw an error,
9784 rather than crashing, when the number of columns specified is different from
9785 what's actually returned by the query (Joe)</para></listitem>
9792 <sect1 id="release-8-0-5">
9793 <title>Release 8.0.5</title>
9796 <title>Release date</title>
9797 <simpara>2005-12-12</simpara>
9801 This release contains a variety of fixes from 8.0.4.
9805 <title>Migration to Version 8.0.5</title>
9808 A dump/restore is not required for those running 8.0.X. However,
9809 if you are upgrading from a version earlier than 8.0.3, see the release
9815 <title>Changes</title>
9819 <listitem><para>Fix race condition in transaction log management</para>
9820 <para>There was a narrow window in which an I/O operation could be initiated
9821 for the wrong page, leading to an Assert failure or data
9825 <listitem><para>Fix bgwriter problems after recovering from errors
9828 The background writer was found to leak buffer pins after write errors.
9829 While not fatal in itself, this might lead to mysterious blockages of
9830 later VACUUM commands.
9834 <listitem><para>Prevent failure if client sends Bind protocol message
9835 when current transaction is already aborted</para></listitem>
9837 <listitem><para><filename>/contrib/ltree</> fixes (Teodor)</para></listitem>
9839 <listitem><para>AIX and HPUX compile fixes (Tom)</para></listitem>
9841 <listitem><para>Retry file reads and writes after Windows
9842 NO_SYSTEM_RESOURCES error (Qingqing Zhou)</para></listitem>
9844 <listitem><para>Fix intermittent failure when <varname>log_line_prefix</>
9845 includes <literal>%i</></para></listitem>
9847 <listitem><para>Fix <application>psql</> performance issue with long scripts
9848 on Windows (Merlin Moncure)</para></listitem>
9850 <listitem><para>Fix missing updates of <filename>pg_group</> flat
9851 file</para></listitem>
9853 <listitem><para>Fix longstanding planning error for outer joins</para>
9854 <para>This bug sometimes caused a bogus error <quote>RIGHT JOIN is
9855 only supported with merge-joinable join conditions</>.</para></listitem>
9857 <listitem><para>Postpone timezone initialization until after
9858 <filename>postmaster.pid</> is created</para>
9859 <para>This avoids confusing startup scripts that expect the pid file to appear
9860 quickly.</para></listitem>
9862 <listitem><para>Prevent core dump in <application>pg_autovacuum</> when a
9863 table has been dropped</para></listitem>
9865 <listitem><para>Fix problems with whole-row references (<literal>foo.*</>)
9866 to subquery results</para></listitem>
9872 <sect1 id="release-8-0-4">
9873 <title>Release 8.0.4</title>
9876 <title>Release date</title>
9877 <simpara>2005-10-04</simpara>
9881 This release contains a variety of fixes from 8.0.3.
9885 <title>Migration to Version 8.0.4</title>
9888 A dump/restore is not required for those running 8.0.X. However,
9889 if you are upgrading from a version earlier than 8.0.3, see the release
9895 <title>Changes</title>
9898 <listitem><para>Fix error that allowed <command>VACUUM</> to remove
9899 <literal>ctid</> chains too soon, and add more checking in code that follows
9900 <literal>ctid</> links</para>
9901 <para>This fixes a long-standing problem that could cause crashes in very rare
9902 circumstances.</para></listitem>
9903 <listitem><para>Fix <type>CHAR()</> to properly pad spaces to the specified
9904 length when using a multiple-byte character set (Yoshiyuki Asaba)</para>
9905 <para>In prior releases, the padding of <type>CHAR()</> was incorrect
9906 because it only padded to the specified number of bytes without
9907 considering how many characters were stored.</para></listitem>
9908 <listitem><para>Force a checkpoint before committing <command>CREATE
9910 <para>This should fix recent reports of <quote>index is not a btree</>
9911 failures when a crash occurs shortly after <command>CREATE
9912 DATABASE</>.</para></listitem>
9913 <listitem><para>Fix the sense of the test for read-only transaction
9914 in <command>COPY</></para>
9915 <para>The code formerly prohibited <command>COPY TO</>, where it should
9916 prohibit <command>COPY FROM</>.
9918 <listitem><para>Handle consecutive embedded newlines in <command>COPY</>
9919 CSV-mode input</para></listitem>
9920 <listitem><para>Fix <function>date_trunc(week)</> for dates near year
9921 end</para></listitem>
9922 <listitem><para>Fix planning problem with outer-join ON clauses that reference
9923 only the inner-side relation</para></listitem>
9924 <listitem><para>Further fixes for <literal>x FULL JOIN y ON true</> corner
9925 cases</para></listitem>
9926 <listitem><para>Fix overenthusiastic optimization of <literal>x IN (SELECT
9927 DISTINCT ...)</> and related cases</para></listitem>
9928 <listitem><para>Fix mis-planning of queries with small <literal>LIMIT</>
9929 values due to poorly thought out <quote>fuzzy</> cost
9930 comparison</para></listitem>
9931 <listitem><para>Make <function>array_in</> and <function>array_recv</> more
9932 paranoid about validating their OID parameter</para></listitem>
9933 <listitem><para>Fix missing rows in queries like <literal>UPDATE a=... WHERE
9934 a...</> with GiST index on column <literal>a</></para></listitem>
9935 <listitem><para>Improve robustness of datetime parsing</para></listitem>
9936 <listitem><para>Improve checking for partially-written WAL
9937 pages</para></listitem>
9938 <listitem><para>Improve robustness of signal handling when SSL is
9939 enabled</para></listitem>
9940 <listitem><para>Improve MIPS and M68K spinlock code</para></listitem>
9941 <listitem><para>Don't try to open more than <literal>max_files_per_process</>
9942 files during postmaster startup</para></listitem>
9943 <listitem><para>Various memory leakage fixes</para></listitem>
9944 <listitem><para>Various portability improvements</para></listitem>
9945 <listitem><para>Update timezone data files</para></listitem>
9946 <listitem><para>Improve handling of DLL load failures on Windows</para></listitem>
9947 <listitem><para>Improve random-number generation on Windows</para></listitem>
9948 <listitem><para>Make <literal>psql -f filename</> return a nonzero exit code
9949 when opening the file fails</para></listitem>
9950 <listitem><para>Change <application>pg_dump</> to handle inherited check
9951 constraints more reliably</para></listitem>
9952 <listitem><para>Fix password prompting in <application>pg_restore</> on
9953 Windows</para></listitem>
9954 <listitem><para>Fix PL/PgSQL to handle <literal>var := var</> correctly when
9955 the variable is of pass-by-reference type</para></listitem>
9956 <listitem><para>Fix PL/Perl <literal>%_SHARED</> so it's actually
9957 shared</para></listitem>
9958 <listitem><para>Fix <filename>contrib/pg_autovacuum</> to allow sleep
9959 intervals over 2000 sec</para></listitem>
9960 <listitem><para>Update <filename>contrib/tsearch2</> to use current Snowball
9961 code</para></listitem>
9967 <sect1 id="release-8-0-3">
9968 <title>Release 8.0.3</title>
9971 <title>Release date</title>
9972 <simpara>2005-05-09</simpara>
9976 This release contains a variety of fixes from 8.0.2, including several
9977 security-related issues.
9981 <title>Migration to Version 8.0.3</title>
9984 A dump/restore is not required for those running 8.0.X. However,
9985 it is one possible way of handling two significant security problems
9986 that have been found in the initial contents of 8.0.X system
9987 catalogs. A dump/initdb/reload sequence using 8.0.3's initdb will
9988 automatically correct these problems.
9992 The larger security problem is that the built-in character set encoding
9993 conversion functions can be invoked from SQL commands by unprivileged
9994 users, but the functions were not designed for such use and are not
9995 secure against malicious choices of arguments. The fix involves changing
9996 the declared parameter list of these functions so that they can no longer
9997 be invoked from SQL commands. (This does not affect their normal use
9998 by the encoding conversion machinery.)
10002 The lesser problem is that the <filename>contrib/tsearch2</> module
10003 creates several functions that are improperly declared to return
10004 <type>internal</> when they do not accept <type>internal</> arguments.
10005 This breaks type safety for all functions using <type>internal</>
10010 It is strongly recommended that all installations repair these errors,
10011 either by initdb or by following the manual repair procedure given
10012 below. The errors at least allow unprivileged database users to crash
10013 their server process, and might allow unprivileged users to gain the
10014 privileges of a database superuser.
10018 If you wish not to do an initdb, perform the same manual repair
10019 procedures shown in the <link linkend="release-7-4-8">7.4.8 release
10025 <title>Changes</title>
10028 <listitem><para>Change encoding function signature to prevent
10029 misuse</para></listitem>
10030 <listitem><para>Change <filename>contrib/tsearch2</> to avoid unsafe use of
10031 <type>INTERNAL</> function results</para></listitem>
10032 <listitem><para>Guard against incorrect second parameter to
10033 <function>record_out</></para></listitem>
10034 <listitem><para>Repair ancient race condition that allowed a transaction to be
10035 seen as committed for some purposes (eg SELECT FOR UPDATE) slightly sooner
10036 than for other purposes</para>
10037 <para>This is an extremely serious bug since it could lead to apparent
10038 data inconsistencies being briefly visible to applications.</para></listitem>
10039 <listitem><para>Repair race condition between relation extension and
10041 <para>This could theoretically have caused loss of a page's worth of
10042 freshly-inserted data, although the scenario seems of very low probability.
10043 There are no known cases of it having caused more than an Assert failure.
10045 <listitem><para>Fix comparisons of <type>TIME WITH TIME ZONE</> values</para>
10047 The comparison code was wrong in the case where the
10048 <literal>--enable-integer-datetimes</> configuration switch had been used.
10049 NOTE: if you have an index on a <type>TIME WITH TIME ZONE</> column,
10050 it will need to be <command>REINDEX</>ed after installing this update, because
10051 the fix corrects the sort order of column values.
10053 <listitem><para>Fix <function>EXTRACT(EPOCH)</> for
10054 <type>TIME WITH TIME ZONE</> values</para></listitem>
10055 <listitem><para>Fix mis-display of negative fractional seconds in
10056 <type>INTERVAL</> values</para>
10058 This error only occurred when the
10059 <literal>--enable-integer-datetimes</> configuration switch had been used.
10061 <listitem><para>Fix pg_dump to dump trigger names containing <literal>%</>
10062 correctly (Neil)</para></listitem>
10063 <listitem><para>Still more 64-bit fixes for
10064 <filename>contrib/intagg</></para></listitem>
10065 <listitem><para>Prevent incorrect optimization of functions returning
10066 <type>RECORD</></para></listitem>
10067 <listitem><para>Prevent crash on <literal>COALESCE(NULL,NULL)</></para></listitem>
10068 <listitem><para>Fix Borland makefile for libpq</para></listitem>
10069 <listitem><para>Fix <filename>contrib/btree_gist</> for <type>timetz</> type
10070 (Teodor)</para></listitem>
10071 <listitem><para>Make <command>pg_ctl</> check the PID found in
10072 <filename>postmaster.pid</> to see if it is still a live
10073 process</para></listitem>
10074 <listitem><para>Fix <command>pg_dump</>/<command>pg_restore</> problems caused
10075 by addition of dump timestamps</para></listitem>
10076 <listitem><para>Fix interaction between materializing holdable cursors and
10077 firing deferred triggers during transaction commit</para></listitem>
10078 <listitem><para>Fix memory leak in SQL functions returning pass-by-reference
10079 data types</para></listitem>
10085 <sect1 id="release-8-0-2">
10086 <title>Release 8.0.2</title>
10089 <title>Release date</title>
10090 <simpara>2005-04-07</simpara>
10094 This release contains a variety of fixes from 8.0.1.
10098 <title>Migration to Version 8.0.2</title>
10101 A dump/restore is not required for those running 8.0.*.
10102 This release updates the major version number of the
10103 <productname>PostgreSQL</productname> libraries, so it might be
10104 necessary to re-link some user applications if they cannot
10105 find the properly-numbered shared library.
10110 <title>Changes</title>
10113 <listitem><para>Increment the major version number of all interface
10114 libraries (Bruce)</para>
10116 This should have been done in 8.0.0. It is required so 7.4.X versions
10117 of PostgreSQL client applications, like <application>psql</>,
10118 can be used on the same machine as 8.0.X applications. This might require
10119 re-linking user applications that use these libraries.
10121 <listitem><para>Add Windows-only <varname>wal_sync_method</> setting of
10122 <option>fsync_writethrough</> (Magnus, Bruce)</para>
10124 This setting causes <productname>PostgreSQL</productname> to write through
10125 any disk-drive write cache when writing to WAL.
10126 This behavior was formerly called <option>fsync</>, but was
10127 renamed because it acts quite differently from <option>fsync</> on other
10131 <listitem><para>Enable the <varname>wal_sync_method</> setting of
10132 <option>open_datasync</> on Windows, and make it the default for that
10133 platform (Magnus, Bruce)</para>
10135 Because the default is no longer <option>fsync_writethrough</>,
10136 data loss is possible during a power failure if the disk drive has
10137 write caching enabled. To turn off the write cache on Windows,
10138 from the <application>Device Manager</>, choose the drive properties,
10139 then <literal>Policies</>.
10142 <listitem><para>New cache management algorithm <acronym>2Q</> replaces
10143 <acronym>ARC</> (Tom)</para>
10145 This was done to avoid a pending US patent on <acronym>ARC</>. The
10146 <acronym>2Q</> code might be a few percentage points slower than
10147 <acronym>ARC</> for some work loads. A better cache management algorithm
10148 will appear in 8.1.
10150 <listitem><para>Planner adjustments to improve behavior on freshly-created
10151 tables (Tom)</para></listitem>
10152 <listitem><para>Allow plpgsql to assign to an element of an array that is
10153 initially <literal>NULL</> (Tom)</para>
10155 Formerly the array would remain <literal>NULL</>, but now it becomes a
10156 single-element array. The main SQL engine was changed to handle
10157 <command>UPDATE</> of a null array value this way in 8.0, but the similar
10158 case in plpgsql was overlooked.
10161 <listitem><para>Convert <literal>\r\n</> and <literal>\r</> to <literal>\n</>
10162 in plpython function bodies (Michael Fuhr)</para>
10164 This prevents syntax errors when plpython code is written on a Windows or
10168 <listitem><para>Allow SPI cursors to handle utility commands that return rows,
10169 such as <command>EXPLAIN</> (Tom)</para></listitem>
10170 <listitem><para>Fix <command>CLUSTER</> failure after <command>ALTER TABLE
10171 SET WITHOUT OIDS</> (Tom)</para></listitem>
10172 <listitem><para>Reduce memory usage of <command>ALTER TABLE ADD COLUMN</>
10173 (Neil)</para></listitem>
10174 <listitem><para>Fix <command>ALTER LANGUAGE RENAME</> (Tom)</para></listitem>
10175 <listitem><para>Document the Windows-only <literal>register</> and
10176 <literal>unregister</> options of <application>pg_ctl</> (Magnus)</para></listitem>
10177 <listitem><para>Ensure operations done during backend shutdown are counted by
10178 statistics collector</para>
10180 This is expected to resolve reports of <application>pg_autovacuum</>
10181 not vacuuming the system catalogs often enough — it was not being
10182 told about catalog deletions caused by temporary table removal during
10185 <listitem><para>Change the Windows default for configuration parameter
10186 <varname>log_destination</> to <option>eventlog</> (Magnus)</para>
10188 By default, a server running on Windows will now send log output to the
10189 Windows event logger rather than standard error.
10191 <listitem><para>Make Kerberos authentication work on Windows (Magnus)</para></listitem>
10192 <listitem><para>Allow <command>ALTER DATABASE RENAME</> by superusers
10193 who aren't flagged as having CREATEDB privilege (Tom)</para></listitem>
10194 <listitem><para>Modify WAL log entries for <command>CREATE</> and
10195 <command>DROP DATABASE</> to not specify absolute paths (Tom)</para>
10196 <para>This allows point-in-time recovery on a different machine with possibly
10197 different database location. Note that <command>CREATE TABLESPACE</> still
10198 poses a hazard in such situations.
10200 <listitem><para>Fix crash from a backend exiting with an open transaction
10201 that created a table and opened a cursor on it (Tom)</para></listitem>
10202 <listitem><para>Fix <function>array_map()</> so it can call PL functions
10203 (Tom)</para></listitem>
10204 <listitem><para>Several <filename>contrib/tsearch2</> and
10205 <filename>contrib/btree_gist</> fixes (Teodor)
10207 <listitem><para>Fix crash of some <filename>contrib/pgcrypto</>
10208 functions on some platforms (Marko Kreen)</para></listitem>
10209 <listitem><para>Fix <filename>contrib/intagg</> for 64-bit platforms
10210 (Tom)</para></listitem>
10211 <listitem><para>Fix ecpg bugs in parsing of <command>CREATE</> statement
10212 (Michael)</para></listitem>
10213 <listitem><para>Work around gcc bug on powerpc and amd64 causing problems in
10214 ecpg (Christof Petig)</para></listitem>
10215 <listitem><para>Do not use locale-aware versions of <function>upper()</>,
10216 <function>lower()</>, and <function>initcap()</> when the locale is
10217 <literal>C</> (Bruce)</para>
10219 This allows these functions to work on platforms that generate errors
10220 for non-7-bit data when the locale is <literal>C</>.
10222 <listitem><para>Fix <function>quote_ident()</> to quote names that match keywords (Tom)</para></listitem>
10223 <listitem><para>Fix <function>to_date()</> to behave reasonably when
10224 <literal>CC</> and <literal>YY</> fields are both used (Karel)</para></listitem>
10225 <listitem><para>Prevent <function>to_char(interval)</> from failing
10226 when given a zero-month interval (Tom)</para></listitem>
10227 <listitem><para>Fix wrong week returned by <function>date_trunc('week')</>
10230 <function>date_trunc('week')</>
10231 returned the wrong year for the first few days of January in some years.
10233 <listitem><para>Use the correct default mask length for class <literal>D</>
10234 addresses in <type>INET</> data types (Tom)</para></listitem>
10240 <sect1 id="release-8-0-1">
10241 <title>Release 8.0.1</title>
10244 <title>Release date</title>
10245 <simpara>2005-01-31</simpara>
10249 This release contains a variety of fixes from 8.0.0, including several
10250 security-related issues.
10254 <title>Migration to Version 8.0.1</title>
10257 A dump/restore is not required for those running 8.0.0.
10262 <title>Changes</title>
10265 <listitem><para>Disallow <command>LOAD</> to non-superusers</para>
10267 On platforms that will automatically execute initialization functions of a
10268 shared library (this includes at least Windows and ELF-based Unixen),
10269 <command>LOAD</> can be used to make the server execute arbitrary code.
10270 Thanks to NGS Software for reporting this.</para></listitem>
10271 <listitem><para>Check that creator of an aggregate function has the right to
10272 execute the specified transition functions</para>
10274 This oversight made it possible to bypass denial of EXECUTE
10275 permission on a function.</para></listitem>
10276 <listitem><para>Fix security and 64-bit issues in
10277 contrib/intagg</para></listitem>
10278 <listitem><para>Add needed STRICT marking to some contrib functions (Kris
10279 Jurka)</para></listitem>
10280 <listitem><para>Avoid buffer overrun when plpgsql cursor declaration has too
10281 many parameters (Neil)</para></listitem>
10282 <listitem><para>Make <command>ALTER TABLE ADD COLUMN</> enforce domain
10283 constraints in all cases</para></listitem>
10284 <listitem><para>Fix planning error for FULL and RIGHT outer joins</para>
10286 The result of the join was mistakenly supposed to be sorted the same as the
10287 left input. This could not only deliver mis-sorted output to the user, but
10288 in case of nested merge joins could give outright wrong answers.
10290 <listitem><para>Improve planning of grouped aggregate queries</para></listitem>
10291 <listitem><para><command>ROLLBACK TO <replaceable>savepoint</></command>
10292 closes cursors created since the savepoint</para></listitem>
10293 <listitem><para>Fix inadequate backend stack size on Windows</para></listitem>
10294 <listitem><para>Avoid SHGetSpecialFolderPath() on Windows
10295 (Magnus)</para></listitem>
10296 <listitem><para>Fix some problems in running pg_autovacuum as a Windows
10297 service (Dave Page)</para></listitem>
10298 <listitem><para>Multiple minor bug fixes in
10299 pg_dump/pg_restore</para></listitem>
10300 <listitem><para>Fix ecpg segfault with named structs used in
10301 typedefs (Michael)</para></listitem>
10307 <sect1 id="release-8-0">
10308 <title>Release 8.0</title>
10311 <title>Release date</title>
10312 <simpara>2005-01-19</simpara>
10316 <title>Overview</title>
10319 Major changes in this release:
10325 Microsoft Windows Native Server
10330 This is the first <productname>PostgreSQL</productname> release
10331 to run natively on <trademark class="registered">Microsoft Windows</> as
10332 a server. It can run as a <productname>Windows</> service. This
10333 release supports NT-based Windows releases like
10334 <productname>Windows 2000 SP4</>, <productname>Windows XP</>, and
10335 <productname>Windows 2003</>. Older releases like
10336 <productname>Windows 95</>, <productname>Windows 98</>, and
10337 <productname>Windows ME</> are not supported because these operating
10338 systems do not have the infrastructure to support
10339 <productname>PostgreSQL</productname>. A separate installer
10340 project has been created to ease installation on
10341 <productname>Windows</> — see <ulink
10342 url="http://www.postgresql.org/ftp/win32/"></ulink>.
10346 Although tested throughout our release cycle, the Windows port
10347 does not have the benefit of years of use in production
10348 environments that <productname>PostgreSQL</productname> has on
10349 Unix platforms. Therefore it should be treated with the same
10350 level of caution as you would a new product.
10354 Previous releases required the Unix emulation toolkit
10355 <productname>Cygwin</> in order to run the server on Windows
10356 operating systems. <productname>PostgreSQL</productname> has
10357 supported native clients on Windows for many years.
10369 Savepoints allow specific parts of a transaction to be aborted
10370 without affecting the remainder of the transaction. Prior
10371 releases had no such capability; there was no way to recover
10372 from a statement failure within a transaction except by
10373 aborting the whole transaction. This feature is valuable for
10374 application writers who require error recovery within a
10375 complex transaction.
10382 Point-In-Time Recovery
10387 In previous releases there was no way to recover from disk
10388 drive failure except to restore from a previous backup or use
10389 a standby replication server. Point-in-time recovery allows
10390 continuous backup of the server. You can recover either to
10391 the point of failure or to some transaction in the past.
10403 Tablespaces allow administrators to select different file systems
10404 for storage of individual tables, indexes, and databases.
10405 This improves performance and control over disk space
10406 usage. Prior releases used <application>initlocation</> and
10407 manual symlink management for such tasks.
10414 Improved Buffer Management, <command>CHECKPOINT</command>,
10415 <command>VACUUM</command>
10420 This release has a more intelligent buffer replacement strategy,
10421 which will make better use of available shared buffers and
10422 improve performance. The performance impact of vacuum and
10423 checkpoints is also lessened.
10430 Change Column Types
10435 A column's data type can now be changed with <command>ALTER
10443 New Perl Server-Side Language
10448 A new version of the <application>plperl</> server-side language now
10449 supports a persistent shared storage area, triggers, returning records
10450 and arrays of records, and SPI calls to access the database.
10457 Comma-separated-value (CSV) support in <command>COPY</command>
10462 <command>COPY</command> can now read and write
10463 comma-separated-value files. It has the flexibility to
10464 interpret nonstandard quoting and separation characters too.
10473 <title>Migration to Version 8.0</title>
10476 A dump/restore using <application>pg_dump</application> is
10477 required for those wishing to migrate data from any previous
10482 Observe the following incompatibilities:
10489 In <option>READ COMMITTED</> serialization mode, volatile functions
10490 now see the results of concurrent transactions committed up to the
10491 beginning of each statement within the function, rather than up to the
10492 beginning of the interactive command that called the function.
10498 Functions declared <option>STABLE</> or <option>IMMUTABLE</> always
10499 use the snapshot of the calling query, and therefore do not see the
10500 effects of actions taken after the calling query starts, whether in
10501 their own transaction or other transactions. Such a function must be
10502 read-only, too, meaning that it cannot use any SQL commands other than
10503 <command>SELECT</>.
10509 Nondeferred <option>AFTER</> triggers are now fired immediately
10510 after completion of the triggering query, rather than upon
10511 finishing the current interactive command. This makes a
10512 difference when the triggering query occurred within a function:
10513 the trigger is invoked before the function proceeds to its next
10520 Server configuration parameters <varname>virtual_host</> and
10521 <varname>tcpip_socket</> have been replaced with a more general
10522 parameter <varname>listen_addresses</>. Also, the server now listens on
10523 <literal>localhost</> by default, which eliminates the need for the
10524 <literal>-i</> postmaster switch in many scenarios.
10530 Server configuration parameters <varname>SortMem</> and
10531 <varname>VacuumMem</> have been renamed to <varname>work_mem</>
10532 and <varname>maintenance_work_mem</> to better reflect their
10533 use. The original names are still supported in
10534 <command>SET</command> and <command>SHOW</command>.
10540 Server configuration parameters <varname>log_pid</>,
10541 <varname>log_timestamp</>, and <varname>log_source_port</> have been
10542 replaced with a more general parameter <varname>log_line_prefix</>.
10548 Server configuration parameter <varname>syslog</> has been
10549 replaced with a more logical <varname>log_destination</> variable to
10550 control the log output destination.
10556 Server configuration parameter <varname>log_statement</> has been
10557 changed so it can selectively log just database modification or
10558 data definition statements. Server configuration parameter
10559 <varname>log_duration</> now prints only when <varname>log_statement</>
10566 Server configuration parameter <varname>max_expr_depth</> parameter has
10567 been replaced with <varname>max_stack_depth</> which measures the
10568 physical stack size rather than the expression nesting depth. This
10569 helps prevent session termination due to stack overflow caused by
10570 recursive functions.
10576 The <function>length()</> function no longer counts trailing spaces in
10577 <type>CHAR(n)</> values.
10583 Casting an integer to <type>BIT(N)</> selects the rightmost N bits of the
10584 integer, not the leftmost N bits as before.
10590 Updating an element or slice of a NULL array value now produces
10591 a nonnull array result, namely an array containing
10592 just the assigned-to positions.
10598 Syntax checking of array input values has been tightened up
10599 considerably. Junk that was previously allowed in odd places with
10600 odd results now causes an error. Empty-string element values
10601 must now be written as <literal>""</>, rather than writing nothing.
10602 Also changed behavior with respect to whitespace surrounding
10603 array elements: trailing whitespace is now ignored, for symmetry
10604 with leading whitespace (which has always been ignored).
10610 Overflow in integer arithmetic operations is now detected and
10611 reported as an error.
10617 The arithmetic operators associated with the single-byte
10618 <type>"char"</> data type have been removed.
10624 The <function>extract()</> function (also called
10625 <function>date_part</>) now returns the proper year for BC dates.
10626 It previously returned one less than the correct year. The
10627 function now also returns the proper values for millennium and
10634 <type>CIDR</> values now must have their nonmasked bits be zero.
10635 For example, we no longer allow
10636 <literal>204.248.199.1/31</literal> as a <type>CIDR</> value. Such
10637 values should never have been accepted by
10638 <productname>PostgreSQL</productname> and will now be rejected.
10644 <command>EXECUTE</command> now returns a completion tag that
10645 matches the executed statement.
10651 <application>psql</>'s <command>\copy</> command now reads or
10652 writes to the query's <literal>stdin/stdout</>, rather than
10653 <application>psql</>'s <literal>stdin/stdout</>. The previous
10654 behavior can be accessed via new
10655 <option>pstdin</>/<option>pstdout</> parameters.
10661 The JDBC client interface has been removed from the core
10662 distribution, and is now hosted at <ulink url=
10663 "http://jdbc.postgresql.org"></ulink>.
10669 The Tcl client interface has also been removed. There are several
10670 Tcl interfaces now hosted at <ulink url=
10671 "http://gborg.postgresql.org"></ulink>.
10677 The server now uses its own time zone database, rather than the
10678 one supplied by the operating system. This will provide consistent
10679 behavior across all platforms. In most cases, there should be
10680 little noticeable difference in time zone behavior, except that
10681 the time zone names used by <command>SET</>/<command>SHOW</>
10682 <varname>TimeZone</> might be different from what your platform provides.
10688 <application>Configure</>'s threading option no longer requires
10689 users to run tests or edit configuration files; threading options
10690 are now detected automatically.
10696 Now that tablespaces have been implemented,
10697 <application>initlocation</> has been removed.
10703 The API for user-defined GiST indexes has been changed. The
10704 Union and PickSplit methods are now passed a pointer to a
10705 special <structname>GistEntryVector</structname> structure,
10706 rather than a <type>bytea</type>.
10714 <title>Deprecated Features</title>
10717 Some aspects of <productname>PostgreSQL</productname>'s behavior
10718 have been determined to be suboptimal. For the sake of backward
10719 compatibility these have not been removed in 8.0, but they are
10720 considered deprecated and will be removed in the next major
10727 The 8.1 release will remove the <function>to_char()</> function
10734 The server now warns of empty strings passed to
10735 <type>oid</type>/<type>float4</type>/<type>float8</type> data
10736 types, but continues to interpret them as zeroes as before.
10737 In the next major release, empty strings will be considered
10738 invalid input for these data types.
10744 By default, tables in <productname>PostgreSQL</productname> 8.0
10745 and earlier are created with <type>OID</>s. In the next release,
10746 this will <emphasis>not</emphasis> be the case: to create a table
10747 that contains <type>OID</>s, the <option>WITH OIDS</> clause must
10748 be specified or the <varname>default_with_oids</varname>
10749 configuration parameter must be set. Users are encouraged to
10750 explicitly specify <option>WITH OIDS</> if their tables
10751 require OIDs for compatibility with future releases of
10752 <productname>PostgreSQL</productname>.
10760 <title>Changes</title>
10763 Below you will find a detailed account of the changes between
10764 release 8.0 and the previous major release.
10768 <title>Performance Improvements</title>
10773 Support cross-data-type index usage (Tom)
10776 Before this change, many queries would not use an index if the data
10777 types did not match exactly. This improvement makes index usage more
10778 intuitive and consistent.
10784 New buffer replacement strategy that improves caching (Jan)
10787 Prior releases used a least-recently-used (LRU) cache to keep
10788 recently referenced pages in memory. The LRU algorithm
10789 did not consider the number of times a specific cache entry was
10790 accessed, so large table scans could force out useful cache pages.
10791 The new cache algorithm uses four separate lists to track most
10792 recently used and most frequently used cache pages and dynamically
10793 optimize their replacement based on the work load. This should
10794 lead to much more efficient use of the shared buffer cache.
10795 Administrators who have tested shared buffer sizes in the past
10796 should retest with this new cache replacement policy.
10802 Add subprocess to write dirty buffers periodically to reduce
10803 checkpoint writes (Jan)
10806 In previous releases, the checkpoint process, which runs every few
10807 minutes, would write all dirty buffers to the operating system's
10808 buffer cache then flush all dirty operating system buffers to
10809 disk. This resulted in a periodic spike in disk usage that often
10810 hurt performance. The new code uses a background writer to trickle
10811 disk writes at a steady pace so checkpoints have far fewer dirty
10812 pages to write to disk. Also, the new code does not issue a global
10813 <function>sync()</> call, but instead <function>fsync()</>s just
10814 the files written since the last checkpoint. This should improve
10815 performance and minimize degradation during checkpoints.
10821 Add ability to prolong vacuum to reduce performance impact (Jan)
10824 On busy systems, <command>VACUUM</command> performs many I/O
10825 requests which can hurt performance for other users. This
10826 release allows you to slow down <command>VACUUM</command> to
10827 reduce its impact on other users, though this increases the
10828 total duration of <command>VACUUM</command>.
10834 Improve B-tree index performance for duplicate keys (Dmitry Tkach, Tom)
10837 This improves the way indexes are scanned when many duplicate
10838 values exist in the index.
10844 Use dynamically-generated table size estimates while planning (Tom)
10847 Formerly the planner estimated table sizes using the values seen
10848 by the last <command>VACUUM</command> or <command>ANALYZE</command>,
10849 both as to physical table size (number of pages) and number of rows.
10850 Now, the current physical table size is obtained from the kernel,
10851 and the number of rows is estimated by multiplying the table size
10852 by the row density (rows per page) seen by the last
10853 <command>VACUUM</command> or <command>ANALYZE</command>. This should
10854 produce more reliable estimates in cases where the table size has
10855 changed significantly since the last housekeeping command.
10861 Improved index usage with <literal>OR</> clauses (Tom)
10864 This allows the optimizer to use indexes in statements with many OR
10865 clauses that would not have been indexed in the past. It can also use
10866 multi-column indexes where the first column is specified and the second
10867 column is part of an <literal>OR</> clause.
10873 Improve matching of partial index clauses (Tom)
10876 The server is now smarter about using partial indexes in queries
10877 involving complex <option>WHERE</> clauses.
10883 Improve performance of the GEQO optimizer (Tom)
10886 The GEQO optimizer is used to plan queries involving many tables (by
10887 default, twelve or more). This release speeds up the way queries are
10888 analyzed to decrease time spent in optimization.
10894 Miscellaneous optimizer improvements
10897 There is not room here to list all the minor improvements made, but
10898 numerous special cases work better than in prior releases.
10904 Improve lookup speed for C functions (Tom)
10907 This release uses a hash table to lookup information for dynamically
10908 loaded C functions. This improves their speed so they perform nearly as
10909 quickly as functions that are built into the server executable.
10915 Add type-specific <command>ANALYZE</command> statistics
10916 capability (Mark Cave-Ayland)
10919 This feature allows more flexibility in generating statistics
10920 for nonstandard data types.
10926 <command>ANALYZE</command> now collects statistics for
10927 expression indexes (Tom)
10930 Expression indexes (also called functional indexes) allow users to
10931 index not just columns but the results of expressions and function
10932 calls. With this release, the optimizer can gather and use statistics
10933 about the contents of expression indexes. This will greatly improve
10934 the quality of planning for queries in which an expression index is
10941 New two-stage sampling method for <command>ANALYZE</command>
10945 This gives better statistics when the density of valid rows is very
10946 different in different regions of a table.
10952 Speed up <command>TRUNCATE</command> (Tom)
10955 This buys back some of the performance loss observed in 7.4, while still
10956 keeping <command>TRUNCATE</command> transaction-safe.
10965 <title>Server Changes</title>
10970 Add WAL file archiving and point-in-time recovery (Simon Riggs)
10976 Add tablespaces so admins can control disk layout (Gavin)
10982 Add a built-in log rotation program (Andreas Pflug)
10985 It is now possible to log server messages conveniently without
10986 relying on either <application>syslog</> or an external log
10993 Add new read-only server configuration parameters to show server
10994 compile-time settings: <varname>block_size</>,
10995 <varname>integer_datetimes</>, <varname>max_function_args</>,
10996 <varname>max_identifier_length</>, <varname>max_index_keys</> (Joe)
11002 Make quoting of <literal>sameuser</>, <literal>samegroup</>, and
11003 <literal>all</> remove special meaning of these terms in
11004 <filename>pg_hba.conf</> (Andrew)
11010 Use clearer IPv6 name <literal>::1/128</> for
11011 <literal>localhost</> in default <filename>pg_hba.conf</> (Andrew)
11017 Use CIDR format in <filename>pg_hba.conf</> examples (Andrew)
11023 Rename server configuration parameters <varname>SortMem</> and
11024 <varname>VacuumMem</> to <varname>work_mem</> and
11025 <varname>maintenance_work_mem</> (Old names still supported) (Tom)
11028 This change was made to clarify that bulk operations such as index and
11029 foreign key creation use <varname>maintenance_work_mem</>, while
11030 <varname>work_mem</> is for workspaces used during query execution.
11036 Allow logging of session disconnections using server configuration
11037 <varname>log_disconnections</> (Andrew)
11043 Add new server configuration parameter <varname>log_line_prefix</> to
11044 allow control of information emitted in each log line (Andrew)
11047 Available information includes user name, database name, remote IP
11048 address, and session start time.
11054 Remove server configuration parameters <varname>log_pid</>,
11055 <varname>log_timestamp</>, <varname>log_source_port</>; functionality
11056 superseded by <varname>log_line_prefix</> (Andrew)
11062 Replace the <varname>virtual_host</> and <varname>tcpip_socket</>
11063 parameters with a unified <varname>listen_addresses</> parameter
11067 <varname>virtual_host</> could only specify a single IP address to
11068 listen on. <varname>listen_addresses</> allows multiple addresses
11075 Listen on localhost by default, which eliminates the need for the
11076 <option>-i</> postmaster switch in many scenarios (Andrew)
11079 Listening on localhost (<literal>127.0.0.1</>) opens no new
11080 security holes but allows configurations like Windows and JDBC,
11081 which do not support local sockets, to work without special
11088 Remove <varname>syslog</> server configuration parameter, and add more
11089 logical <varname>log_destination</> variable to control log output
11096 Change server configuration parameter <varname>log_statement</> to take
11097 values <varname>all</>, <varname>mod</>, <varname>ddl</>, or
11098 <varname>none</> to select which queries are logged (Bruce)
11101 This allows administrators to log only data definition changes or
11102 only data modification statements.
11108 Some logging-related configuration parameters could formerly be adjusted
11109 by ordinary users, but only in the <quote>more verbose</> direction.
11110 They are now treated more strictly: only superusers can set them.
11111 However, a superuser can use <command>ALTER USER</> to provide per-user
11112 settings of these values for non-superusers. Also, it is now possible
11113 for superusers to set values of superuser-only configuration parameters
11114 via <literal>PGOPTIONS</>.
11120 Allow configuration files to be placed outside the data directory (mlw)
11123 By default, configuration files are kept in the cluster's top directory.
11124 With this addition, configuration files can be placed outside the
11125 data directory, easing administration.
11131 Plan prepared queries only when first executed so constants can be
11132 used for statistics (Oliver Jowett)
11135 Prepared statements plan queries once and execute them many
11136 times. While prepared queries avoid the overhead of re-planning
11137 on each use, the quality of the plan suffers from not knowing the exact
11138 parameters to be used in the query. In this release, planning of
11139 unnamed prepared statements is delayed until the first execution,
11140 and the actual parameter values of that execution are used as
11141 optimization hints. This allows use of out-of-line parameter passing
11142 without incurring a performance penalty.
11148 Allow <command>DECLARE CURSOR</command> to take parameters
11152 It is now useful to issue <command>DECLARE CURSOR</command> in a
11153 <function>Parse</> message with parameters. The parameter values
11154 sent at <function>Bind</> time will be substituted into the
11155 execution of the cursor's query.
11161 Fix hash joins and aggregates of <type>inet</type> and
11162 <type>cidr</type> data types (Tom)
11165 Release 7.4 handled hashing of mixed <type>inet</type> and
11166 <type>cidr</type> values incorrectly. (This bug did not exist
11167 in prior releases because they wouldn't try to hash either
11174 Make <varname>log_duration</> print only when <varname>log_statement</>
11175 prints the query (Ed L.)
11184 <title>Query Changes</title>
11189 Add savepoints (nested transactions) (Alvaro)
11195 Unsupported isolation levels are now accepted and promoted to the
11196 nearest supported level (Peter)
11199 The SQL specification states that if a database doesn't support a
11200 specific isolation level, it should use the next more restrictive level.
11201 This change complies with that recommendation.
11207 Allow <command>BEGIN WORK</command> to specify transaction
11208 isolation levels like <command>START TRANSACTION</command> does
11215 Fix table permission checking for cases in which rules generate
11216 a query type different from the originally submitted query (Tom)
11222 Implement dollar quoting to simplify single-quote usage (Andrew, Tom,
11226 In previous releases, because single quotes had to be used to
11227 quote a function's body, the use of single quotes inside the
11228 function text required use of two single quotes or other error-prone
11229 notations. With this release we add the ability to use "dollar
11230 quoting" to quote a block of text. The ability to use different
11231 quoting delimiters at different nesting levels greatly simplifies
11232 the task of quoting correctly, especially in complex functions.
11233 Dollar quoting can be used anywhere quoted text is needed.
11239 Make <literal>CASE val WHEN compval1 THEN ...</> evaluate <literal>val</> only once (Tom)
11242 <option>CASE</> no longer evaluates the tested expression multiple
11243 times. This has benefits when the expression is complex or is
11250 Test <option>HAVING</> before computing target list of an
11251 aggregate query (Tom)
11254 Fixes improper failure of cases such as <literal>SELECT SUM(win)/SUM(lose)
11255 ... GROUP BY ... HAVING SUM(lose) > 0</>. This should work but formerly
11256 could fail with divide-by-zero.
11262 Replace <varname>max_expr_depth</> parameter with
11263 <varname>max_stack_depth</> parameter, measured in kilobytes of stack
11267 This gives us a fairly bulletproof defense against crashing due to
11268 runaway recursive functions. Instead of measuring the depth of expression
11269 nesting, we now directly measure the size of the execution stack.
11275 Allow arbitrary row expressions (Tom)
11278 This release allows SQL expressions to contain arbitrary composite
11279 types, that is, row values. It also allows functions to more easily
11280 take rows as arguments and return row values.
11286 Allow <option>LIKE</>/<option>ILIKE</> to be used as the operator
11287 in row and subselect comparisons (Fabien Coelho)
11293 Avoid locale-specific case conversion of basic ASCII letters in
11294 identifiers and keywords (Tom)
11297 This solves the <quote>Turkish problem</> with mangling of words
11298 containing <literal>I</> and <literal>i</>. Folding of characters
11299 outside the 7-bit-ASCII set is still locale-aware.
11305 Improve syntax error reporting (Fabien, Tom)
11308 Syntax error reports are more useful than before.
11314 Change <command>EXECUTE</command> to return a completion tag
11315 matching the executed statement (Kris Jurka)
11318 Previous releases return an <command>EXECUTE</command> tag for
11319 any <command>EXECUTE</command> call. In this release, the tag
11320 returned will reflect the command executed.
11326 Avoid emitting <option>NATURAL CROSS JOIN</> in rule listings (Tom)
11329 Such a clause makes no logical sense, but in some cases the rule
11330 decompiler formerly produced this syntax.
11339 <title>Object Manipulation Changes</title>
11344 Add <command>COMMENT ON</> for casts, conversions, languages,
11345 operator classes, and large objects (Christopher)
11351 Add new server configuration parameter <varname>default_with_oids</> to
11352 control whether tables are created with <type>OID</>s by default (Neil)
11355 This allows administrators to control whether <command>CREATE
11356 TABLE</command> commands create tables with or without <type>OID</>
11357 columns by default. (Note: the current factory default setting for
11358 <varname>default_with_oids</> is <literal>TRUE</>, but the default
11359 will become <literal>FALSE</> in future releases.)
11365 Add <option>WITH</> / <option>WITHOUT OIDS</> clause to
11366 <command>CREATE TABLE AS</command> (Neil)
11372 Allow <command>ALTER TABLE DROP COLUMN</> to drop an <type>OID</>
11373 column (<command>ALTER TABLE SET WITHOUT OIDS</> still works)
11380 Allow composite types as table columns (Tom)
11386 Allow <command>ALTER ... ADD COLUMN</> with defaults and
11387 <option>NOT NULL</> constraints; works per SQL spec (Rod)
11390 It is now possible for <option>ADD COLUMN</> to create a column
11391 that is not initially filled with NULLs, but with a specified
11398 Add <command>ALTER COLUMN TYPE</> to change column's type (Rod)
11401 It is now possible to alter a column's data type without dropping
11402 and re-adding the column.
11408 Allow multiple <command>ALTER</> actions in a single <command>ALTER
11409 TABLE</command> command (Rod)
11412 This is particularly useful for <command>ALTER</> commands that
11413 rewrite the table (which include <option>ALTER COLUMN TYPE</> and
11414 <option>ADD COLUMN</> with a default). By grouping
11415 <command>ALTER</> commands together, the table need be rewritten
11422 Allow <command>ALTER TABLE</command> to add <type>SERIAL</type>
11426 This falls out from the new capability of specifying defaults for new
11433 Allow changing the owners of aggregates, conversions, databases,
11434 functions, operators, operator classes, schemas, types, and tablespaces
11435 (Christopher, Euler Taveira de Oliveira)
11438 Previously this required modifying the system tables directly.
11444 Allow temporary object creation to be limited to <option>SECURITY
11445 DEFINER</> functions (Sean Chittenden)
11451 Add <option>ALTER TABLE ... SET WITHOUT CLUSTER</> (Christopher)
11454 Prior to this release, there was no way to clear an auto-cluster
11455 specification except to modify the system tables.
11461 Constraint/Index/<type>SERIAL</> names are now
11462 <replaceable>table_column_type</>
11463 with numbers appended to guarantee uniqueness within the schema
11467 The SQL specification states that such names should be unique
11474 Add <function>pg_get_serial_sequence()</> to return a
11475 <type>SERIAL</> column's sequence name (Christopher)
11478 This allows automated scripts to reliably find the <type>SERIAL</>
11485 Warn when primary/foreign key data type mismatch requires costly lookup
11491 New <command>ALTER INDEX</> command to allow moving of indexes
11492 between tablespaces (Gavin)
11498 Make <command>ALTER TABLE OWNER</> change dependent sequence
11499 ownership too (Alvaro)
11509 <title>Utility Command Changes</title>
11514 Allow <command>CREATE SCHEMA</command> to create triggers,
11515 indexes, and sequences (Neil)
11521 Add <option>ALSO</> keyword to <command>CREATE RULE</command> (Fabien
11525 This allows <option>ALSO</> to be added to rule creation to contrast it with
11526 <option>INSTEAD</> rules.
11532 Add <option>NOWAIT</> option to <command>LOCK</command> (Tatsuo)
11535 This allows the <command>LOCK</command> command to fail if it
11536 would have to wait for the requested lock.
11542 Allow <command>COPY</command> to read and write
11543 comma-separated-value (CSV) files (Andrew, Bruce)
11549 Generate error if the <command>COPY</command> delimiter and NULL
11550 string conflict (Bruce)
11556 <command>GRANT</command>/<command>REVOKE</command> behavior
11557 follows the SQL spec more closely
11563 Avoid locking conflict between <command>CREATE INDEX</command>
11564 and <command>CHECKPOINT</command> (Tom)
11567 In 7.3 and 7.4, a long-running B-tree index build could block concurrent
11568 <command>CHECKPOINT</>s from completing, thereby causing WAL bloat because the
11569 WAL log could not be recycled.
11575 Database-wide <command>ANALYZE</command> does not hold locks
11576 across tables (Tom)
11579 This reduces the potential for deadlocks against other backends
11580 that want exclusive locks on tables. To get the benefit of this
11581 change, do not execute database-wide <command>ANALYZE</command>
11582 inside a transaction block (<command>BEGIN</command> block); it
11583 must be able to commit and start a new transaction for each
11590 <command>REINDEX</command> does not exclusively lock the index's
11591 parent table anymore
11594 The index itself is still exclusively locked, but readers of the
11595 table can continue if they are not using the particular index
11602 Erase MD5 user passwords when a user is renamed (Bruce)
11605 <productname>PostgreSQL</productname> uses the user name as salt
11606 when encrypting passwords via MD5. When a user's name is changed,
11607 the salt will no longer match the stored MD5 password, so the
11608 stored password becomes useless. In this release a notice is
11609 generated and the password is cleared. A new password must then
11610 be assigned if the user is to be able to log in with a password.
11616 New <application>pg_ctl</> <option>kill</> option for Windows (Andrew)
11619 Windows does not have a <literal>kill</> command to send signals to
11620 backends so this capability was added to <application>pg_ctl</>.
11626 Information schema improvements
11632 Add <option>--pwfile</> option to
11633 <application>initdb</application> so the initial password can be
11634 set by GUI tools (Magnus)
11640 Detect locale/encoding mismatch in
11641 <application>initdb</application> (Peter)
11647 Add <option>register</> command to <application>pg_ctl</> to
11648 register Windows operating system service (Dave Page)
11657 <title>Data Type and Function Changes</title>
11662 More complete support for composite types (row types) (Tom)
11665 Composite values can be used in many places where only scalar values
11672 Reject nonrectangular array values as erroneous (Joe)
11675 Formerly, <function>array_in</> would silently build a
11682 Overflow in integer arithmetic operations is now detected (Tom)
11688 The arithmetic operators associated with the single-byte
11689 <type>"char"</> data type have been removed.
11692 Formerly, the parser would select these operators in many situations
11693 where an <quote>unable to select an operator</> error would be more
11694 appropriate, such as <literal>null * null</>. If you actually want
11695 to do arithmetic on a <type>"char"</> column, you can cast it to
11696 integer explicitly.
11702 Syntax checking of array input values considerably tightened up (Joe)
11705 Junk that was previously allowed in odd places with odd results
11706 now causes an <literal>ERROR</>, for example, non-whitespace
11707 after the closing right brace.
11713 Empty-string array element values must now be written as
11714 <literal>""</>, rather than writing nothing (Joe)
11717 Formerly, both ways of writing an empty-string element value were
11718 allowed, but now a quoted empty string is required. The case where
11719 nothing at all appears will probably be considered to be a NULL
11720 element value in some future release.
11726 Array element trailing whitespace is now ignored (Joe)
11729 Formerly leading whitespace was ignored, but trailing whitespace
11730 between an element value and the delimiter or right brace was
11731 significant. Now trailing whitespace is also ignored.
11737 Emit array values with explicit array bounds when lower bound is not one
11744 Accept <literal>YYYY-monthname-DD</> as a date string (Tom)
11750 Make <function>netmask</> and <function>hostmask</> functions
11751 return maximum-length mask length (Tom)
11757 Change factorial function to return <type>numeric</type> (Gavin)
11760 Returning <type>numeric</type> allows the factorial function to
11761 work for a wider range of input values.
11767 <function>to_char</>/<function>to_date()</> date conversion
11768 improvements (Kurt Roeckx, Fabien Coelho)
11774 Make <function>length()</> disregard trailing spaces in
11775 <type>CHAR(n)</> (Gavin)
11778 This change was made to improve consistency: trailing spaces are
11779 semantically insignificant in <type>CHAR(n)</> data, so they
11780 should not be counted by <function>length()</>.
11786 Warn about empty string being passed to
11787 <type>OID</>/<type>float4</>/<type>float8</> data types (Neil)
11790 8.1 will throw an error instead.
11796 Allow leading or trailing whitespace in
11797 <type>int2</>/<type>int4</>/<type>int8</>/<type>float4</>/<type>float8</>
11805 Better support for IEEE <literal>Infinity</> and <literal>NaN</>
11806 values in <type>float4</type>/<type>float8</type> (Neil)
11809 These should now work on all platforms that support IEEE-compliant
11810 floating point arithmetic.
11816 Add <option>week</> option to <function>date_trunc()</> (Robert Creager)
11822 Fix <function>to_char</function> for <literal>1 BC</>
11823 (previously it returned <literal>1 AD</>) (Bruce)
11829 Fix <function>date_part(year)</> for BC dates (previously it
11830 returned one less than the correct year) (Bruce)
11836 Fix <function>date_part()</> to return the proper millennium and
11837 century (Fabien Coelho)
11840 In previous versions, the century and millennium results had a wrong
11841 number and started in the wrong year, as compared to standard
11842 reckoning of such things.
11848 Add <function>ceiling()</> as an alias for <function>ceil()</>,
11849 and <function>power()</> as an alias for <function>pow()</> for
11850 standards compliance (Neil)
11856 Change <function>ln()</>, <function>log()</>,
11857 <function>power()</>, and <function>sqrt()</> to emit the correct
11858 <literal>SQLSTATE</> error codes for certain error conditions, as
11859 specified by SQL:2003 (Neil)
11865 Add <function>width_bucket()</> function as defined by SQL:2003 (Neil)
11871 Add <function>generate_series()</> functions to simplify working
11872 with numeric sets (Joe)
11878 Fix <function>upper/lower/initcap()</> functions to work with
11879 multibyte encodings (Tom)
11885 Add boolean and bitwise integer <option>AND</>/<option>OR</>
11886 aggregates (Fabien Coelho)
11892 New session information functions to return network addresses for client
11893 and server (Sean Chittenden)
11899 Add function to determine the area of a closed path (Sean Chittenden)
11905 Add function to send cancel request to other backends (Magnus)
11911 Add <type>interval</> plus <type>datetime</> operators (Tom)
11914 The reverse ordering, <type>datetime</> plus <type>interval</>,
11915 was already supported, but both are required by the SQL standard.
11921 Casting an integer to <type>BIT(N)</> selects the rightmost N bits
11926 In prior releases, the leftmost N bits were selected, but this was
11927 deemed unhelpful, not to mention inconsistent with casting from bit
11934 Require <type>CIDR</> values to have all nonmasked bits be zero
11944 <title>Server-Side Language Changes</title>
11949 In <literal>READ COMMITTED</> serialization mode, volatile functions
11950 now see the results of concurrent transactions committed up to the
11951 beginning of each statement within the function, rather than up to the
11952 beginning of the interactive command that called the function.
11958 Functions declared <literal>STABLE</> or <literal>IMMUTABLE</> always
11959 use the snapshot of the calling query, and therefore do not see the
11960 effects of actions taken after the calling query starts, whether in
11961 their own transaction or other transactions. Such a function must be
11962 read-only, too, meaning that it cannot use any SQL commands other than
11963 <command>SELECT</>. There is a considerable performance gain from
11964 declaring a function <literal>STABLE</> or <literal>IMMUTABLE</>
11965 rather than <literal>VOLATILE</>.
11971 Nondeferred <option>AFTER</> triggers are now fired immediately
11972 after completion of the triggering query, rather than upon
11973 finishing the current interactive command. This makes a difference
11974 when the triggering query occurred within a function: the trigger
11975 is invoked before the function proceeds to its next operation. For
11976 example, if a function inserts a new row into a table, any
11977 nondeferred foreign key checks occur before proceeding with the
11984 Allow function parameters to be declared with names (Dennis Björklund)
11987 This allows better documentation of functions. Whether the names
11988 actually do anything depends on the specific function language
11995 Allow PL/pgSQL parameter names to be referenced in the function (Dennis Björklund)
11998 This basically creates an automatic alias for each named parameter.
12004 Do minimal syntax checking of PL/pgSQL functions at creation time (Tom)
12007 This allows us to catch simple syntax errors sooner.
12013 More support for composite types (row and record variables) in PL/pgSQL
12016 For example, it now works to pass a rowtype variable to another function
12017 as a single variable.
12023 Default values for PL/pgSQL variables can now reference previously
12030 Improve parsing of PL/pgSQL FOR loops (Tom)
12033 Parsing is now driven by presence of <literal>".."</> rather than
12034 data type of <option>FOR</> variable. This makes no difference for
12035 correct functions, but should result in more understandable error
12036 messages when a mistake is made.
12042 Major overhaul of PL/Perl server-side language (Command Prompt, Andrew Dunstan)
12048 In PL/Tcl, SPI commands are now run in subtransactions. If an error
12049 occurs, the subtransaction is cleaned up and the error is reported
12050 as an ordinary Tcl error, which can be trapped with <literal>catch</>.
12051 Formerly, it was not possible to catch such errors.
12057 Accept <command>ELSEIF</> in PL/pgSQL (Neil)
12060 Previously PL/pgSQL only allowed <command>ELSIF</>, but many people
12061 are accustomed to spelling this keyword <command>ELSEIF</>.
12070 <title><application>psql</> Changes</title>
12075 Improve <application>psql</> information display about database
12076 objects (Christopher)
12082 Allow <application>psql</> to display group membership in
12083 <command>\du</> and <command>\dg</> (Markus Bertheau)
12089 Prevent <application>psql</> <command>\dn</command> from showing
12090 temporary schemas (Bruce)
12096 Allow <application>psql</> to handle tilde user expansion for file
12103 Allow <application>psql</> to display fancy prompts, including
12104 color, via <application>readline</> (Reece Hart, Chet Ramey)
12110 Make <application>psql</> <command>\copy</> match <command>COPY</command> command syntax
12117 Show the location of syntax errors (Fabien Coelho, Tom)
12123 Add <command>CLUSTER</command> information to <application>psql</>
12124 <command>\d</> display
12131 Change <application>psql</> <command>\copy stdin/stdout</> to read
12132 from command input/output (Bruce)
12138 Add <option>pstdin</>/<option>pstdout</> to read from
12139 <application>psql</>'s <literal>stdin</>/<literal>stdout</> (Mark
12146 Add global <application>psql</> configuration file, <filename>psqlrc.sample</filename>
12150 This allows a central file where global <application>psql</> startup commands can
12157 Have <application>psql</> <command>\d+</> indicate if the table
12158 has an <type>OID</> column (Neil)
12164 On Windows, use binary mode in <application>psql</> when reading files so control-Z
12165 is not seen as end-of-file
12171 Have <command>\dn+</> show permissions and description for schemas (Dennis
12178 Improve tab completion support (Stefan Kaltenbrunn, Greg Sabino Mullane)
12184 Allow boolean settings to be set using upper or lower case (Michael Paesold)
12193 <title><application>pg_dump</> Changes</title>
12198 Use dependency information to improve the reliability of
12199 <application>pg_dump</> (Tom)
12202 This should solve the longstanding problems with related objects
12203 sometimes being dumped in the wrong order.
12209 Have <application>pg_dump</> output objects in alphabetical order if possible (Tom)
12212 This should make it easier to identify changes between
12219 Allow <application>pg_restore</> to ignore some SQL errors (Fabien Coelho)
12222 This makes <application>pg_restore</>'s behavior similar to the
12223 results of feeding a <application>pg_dump</> output script to
12224 <application>psql</>. In most cases, ignoring errors and plowing
12225 ahead is the most useful thing to do. Also added was a pg_restore
12226 option to give the old behavior of exiting on an error.
12232 <application>pg_restore</> <option>-l</> display now includes
12233 objects' schema names
12239 New begin/end markers in <application>pg_dump</> text output (Bruce)
12245 Add start/stop times for
12246 <application>pg_dump</>/<application>pg_dumpall</> in verbose mode
12253 Allow most <application>pg_dump</> options in
12254 <application>pg_dumpall</> (Christopher)
12260 Have <application>pg_dump</> use <command>ALTER OWNER</> rather
12261 than <command>SET SESSION AUTHORIZATION</> by default
12271 <title>libpq Changes</title>
12276 Make libpq's <option>SIGPIPE</> handling thread-safe (Bruce)
12282 Add <function>PQmbdsplen()</> which returns the display length
12283 of a character (Tatsuo)
12289 Add thread locking to <application>SSL</> and
12290 <application>Kerberos</> connections (Manfred Spraul)
12296 Allow <function>PQoidValue()</>, <function>PQcmdTuples()</>, and
12297 <function>PQoidStatus()</> to work on <command>EXECUTE</command>
12304 Add <function>PQserverVersion()</> to provide more convenient
12305 access to the server version number (Greg Sabino Mullane)
12311 Add <function>PQprepare/PQsendPrepared()</> functions to support
12312 preparing statements without necessarily specifying the data types
12313 of their parameters (Abhijit Menon-Sen)
12319 Many ECPG improvements, including <command>SET DESCRIPTOR</> (Michael)
12328 <title>Source Code Changes</title>
12333 Allow the database server to run natively on Windows (Claudio, Magnus, Andrew)
12339 Shell script commands converted to C versions for Windows support (Andrew)
12345 Create an extension makefile framework (Fabien Coelho, Peter)
12348 This simplifies the task of building extensions outside the original
12355 Support relocatable installations (Bruce)
12358 Directory paths for installed files (such as the
12359 <filename>/share</> directory) are now computed relative to the
12360 actual location of the executables, so that an installation tree
12361 can be moved to another place without reconfiguring and
12368 Use <option>--with-docdir</> to choose installation location of documentation; also
12369 allow <option>--infodir</> (Peter)
12375 Add <option>--without-docdir</> to prevent installation of documentation (Peter)
12381 Upgrade to <application>DocBook</> V4.2 SGML (Peter)
12387 New <literal>PostgreSQL</> <application>CVS</> tag (Marc)
12390 This was done to make it easier for organizations to manage their
12391 own copies of the <productname>PostgreSQL</productname>
12392 <application>CVS</> repository. File version stamps from the master
12393 repository will not get munged by checking into or out of a copied
12400 Clarify locking code (Manfred Koizar)
12406 Buffer manager cleanup (Neil)
12412 Decouple platform tests from CPU spinlock code (Bruce, Tom)
12418 Add inlined test-and-set code on PA-RISC for <application>gcc</>
12425 Improve i386 spinlock code (Manfred Spraul)
12431 Clean up spinlock assembly code to avoid warnings from newer
12432 <application>gcc</> releases (Tom)
12438 Remove JDBC from source tree; now a separate project
12444 Remove the libpgtcl client interface; now a separate project
12450 More accurately estimate memory and file descriptor usage (Tom)
12456 Improvements to the Mac OS X startup scripts (Ray A.)
12462 New <function>fsync()</> test program (Bruce)
12468 Major documentation improvements (Neil, Peter)
12474 Remove <application>pg_encoding</application>; not needed
12481 Remove <application>pg_id</application>; not needed anymore
12487 Remove <application>initlocation</application>; not needed
12494 Auto-detect thread flags (no more manual testing) (Bruce)
12500 Use Olson's public domain <application>timezone</> library (Magnus)
12506 With threading enabled, use thread flags on Unixware for
12507 backend executables too (Bruce)
12510 Unixware cannot mix threaded and nonthreaded object files in the
12511 same executable, so everything must be compiled as threaded.
12517 <application>psql</> now uses a <application>flex</>-generated
12518 lexical analyzer to process command strings
12524 Reimplement the linked list data structure used throughout the
12528 This improves performance by allowing list append and length
12529 operations to be more efficient.
12535 Allow dynamically loaded modules to create their own server configuration
12536 parameters (Thomas Hallgren)
12542 New Brazilian version of FAQ (Euler Taveira de Oliveira)
12548 Add French FAQ (Guillaume Lelarge)
12554 New <application>pgevent</> for Windows logging
12560 Make libpq and ECPG build as proper shared libraries on OS X (Tom)
12569 <title>Contrib Changes</title>
12574 Overhaul of <filename>contrib/dblink</> (Joe)
12580 <filename>contrib/dbmirror</> improvements (Steven Singer)
12586 New <filename>contrib/xml2</> (John Gray, Torchbox)
12592 Updated <filename>contrib/mysql</filename>
12598 New version of <filename>contrib/btree_gist</> (Teodor)
12604 New <filename>contrib/trgm</>, trigram matching for
12605 <productname>PostgreSQL</productname> (Teodor)
12611 Many <filename>contrib/tsearch2</> improvements (Teodor)
12617 Add double metaphone to <filename>contrib/fuzzystrmatch</> (Andrew)
12623 Allow <filename>contrib/pg_autovacuum</> to run as a Windows service (Dave Page)
12629 Add functions to <filename>contrib/dbsize</> (Andreas Pflug)
12635 Removed <filename>contrib/pg_logger</>: obsoleted by integrated logging
12642 Removed <filename>contrib/rserv</>: obsoleted by various separate projects
12652 <sect1 id="release-7-4-18">
12653 <title>Release 7.4.18</title>
12656 <title>Release date</title>
12657 <simpara>2007-09-17</simpara>
12661 This release contains fixes from 7.4.17.
12665 <title>Migration to Version 7.4.18</title>
12668 A dump/restore is not required for those running 7.4.X. However,
12669 if you are upgrading from a version earlier than 7.4.11, see the release
12676 <title>Changes</title>
12682 Prevent index corruption when a transaction inserts rows and
12683 then aborts close to the end of a concurrent <command>VACUUM</>
12684 on the same table (Tom)
12690 Make <command>CREATE DOMAIN ... DEFAULT NULL</> work properly (Tom)
12696 Fix excessive logging of <acronym>SSL</> error messages (Tom)
12702 Fix crash when <varname>log_min_error_statement</> logging runs out
12709 Prevent <command>CLUSTER</> from failing
12710 due to attempting to process temporary tables of other sessions (Alvaro)
12716 Require non-superusers who use <filename>/contrib/dblink</> to use only
12717 password authentication, as a security measure (Joe)
12726 <sect1 id="release-7-4-17">
12727 <title>Release 7.4.17</title>
12730 <title>Release date</title>
12731 <simpara>2007-04-23</simpara>
12735 This release contains fixes from 7.4.16,
12736 including a security fix.
12740 <title>Migration to Version 7.4.17</title>
12743 A dump/restore is not required for those running 7.4.X. However,
12744 if you are upgrading from a version earlier than 7.4.11, see the release
12751 <title>Changes</title>
12757 Support explicit placement of the temporary-table schema within
12758 <varname>search_path</>, and disable searching it for functions
12759 and operators (Tom)
12762 This is needed to allow a security-definer function to set a
12763 truly secure value of <varname>search_path</>. Without it,
12764 an unprivileged SQL user can use temporary objects to execute code
12765 with the privileges of the security-definer function (CVE-2007-2138).
12766 See <command>CREATE FUNCTION</> for more information.
12772 <filename>/contrib/tsearch2</> crash fixes (Teodor)
12778 Fix potential-data-corruption bug in how <command>VACUUM FULL</> handles
12779 <command>UPDATE</> chains (Tom, Pavan Deolasee)
12785 Fix PANIC during enlargement of a hash index (bug introduced in 7.4.15)
12795 <sect1 id="release-7-4-16">
12796 <title>Release 7.4.16</title>
12799 <title>Release date</title>
12800 <simpara>2007-02-05</simpara>
12804 This release contains a variety of fixes from 7.4.15, including
12809 <title>Migration to Version 7.4.16</title>
12812 A dump/restore is not required for those running 7.4.X. However,
12813 if you are upgrading from a version earlier than 7.4.11, see the release
12820 <title>Changes</title>
12826 Remove security vulnerability that allowed connected users
12827 to read backend memory (Tom)
12830 The vulnerability involves suppressing the normal check that a SQL
12831 function returns the data type it's declared to, or changing the
12832 data type of a table column used in a SQL function (CVE-2007-0555).
12833 This error can easily be exploited to cause a backend crash, and in
12834 principle might be used to read database content that the user
12835 should not be able to access.
12841 Fix rare bug wherein btree index page splits could fail
12842 due to choosing an infeasible split point (Heikki Linnakangas)
12848 Fix for rare Assert() crash triggered by <literal>UNION</> (Tom)
12854 Tighten security of multi-byte character processing for UTF8 sequences
12855 over three bytes long (Tom)
12864 <sect1 id="release-7-4-15">
12865 <title>Release 7.4.15</title>
12868 <title>Release date</title>
12869 <simpara>2007-01-08</simpara>
12873 This release contains a variety of fixes from 7.4.14.
12877 <title>Migration to Version 7.4.15</title>
12880 A dump/restore is not required for those running 7.4.X. However,
12881 if you are upgrading from a version earlier than 7.4.11, see the release
12888 <title>Changes</title>
12894 Improve handling of <function>getaddrinfo()</> on AIX (Tom)
12898 This fixes a problem with starting the statistics collector,
12899 among other things.
12905 Fix <quote>failed to re-find parent key</> errors in
12906 <command>VACUUM</> (Tom)
12912 Fix bugs affecting multi-gigabyte hash indexes (Tom)
12918 Fix error when constructing an <literal>ARRAY[]</> made up of multiple
12919 empty elements (Tom)
12925 <function>to_number()</> and <function>to_char(numeric)</>
12926 are now <literal>STABLE</>, not <literal>IMMUTABLE</>, for
12927 new <application>initdb</> installs (Tom)
12931 This is because <varname>lc_numeric</> can potentially
12932 change the output of these functions.
12938 Improve index usage of regular expressions that use parentheses (Tom)
12942 This improves <application>psql</> <literal>\d</> performance also.
12951 <sect1 id="release-7-4-14">
12952 <title>Release 7.4.14</title>
12955 <title>Release date</title>
12956 <simpara>2006-10-16</simpara>
12960 This release contains a variety of fixes from 7.4.13.
12964 <title>Migration to Version 7.4.14</title>
12967 A dump/restore is not required for those running 7.4.X. However,
12968 if you are upgrading from a version earlier than 7.4.11, see the release
12975 <title>Changes</title>
12978 <listitem><para>Fix core dump when an untyped literal is taken as
12979 ANYARRAY</para></listitem>
12980 <listitem><para>Fix <function>string_to_array()</> to handle overlapping
12981 matches for the separator string</para>
12982 <para>For example, <literal>string_to_array('123xx456xxx789', 'xx')</>.
12984 <listitem><para>Fix corner cases in pattern matching for
12985 <application>psql</>'s <literal>\d</> commands</para></listitem>
12986 <listitem><para>Fix index-corrupting bugs in /contrib/ltree
12987 (Teodor)</para></listitem>
12988 <listitem><para>Fix backslash escaping in /contrib/dbmirror</para></listitem>
12989 <listitem><para>Adjust regression tests for recent changes in US DST laws
12990 </para> </listitem>
12996 <sect1 id="release-7-4-13">
12997 <title>Release 7.4.13</title>
13000 <title>Release date</title>
13001 <simpara>2006-05-23</simpara>
13005 This release contains a variety of fixes from 7.4.12,
13006 including patches for extremely serious security issues.
13010 <title>Migration to Version 7.4.13</title>
13013 A dump/restore is not required for those running 7.4.X. However,
13014 if you are upgrading from a version earlier than 7.4.11, see the release
13019 Full security against the SQL-injection attacks described in
13020 CVE-2006-2313 and CVE-2006-2314 might require changes in application
13021 code. If you have applications that embed untrustworthy strings
13022 into SQL commands, you should examine them as soon as possible to
13023 ensure that they are using recommended escaping techniques. In
13024 most cases, applications should be using subroutines provided by
13025 libraries or drivers (such as <application>libpq</>'s
13026 <function>PQescapeStringConn()</>) to perform string escaping,
13027 rather than relying on <foreignphrase>ad hoc</> code to do it.
13032 <title>Changes</title>
13035 <listitem><para>Change the server to reject invalidly-encoded multibyte
13036 characters in all cases (Tatsuo, Tom)</para>
13037 <para>While <productname>PostgreSQL</> has been moving in this direction for
13038 some time, the checks are now applied uniformly to all encodings and all
13039 textual input, and are now always errors not merely warnings. This change
13040 defends against SQL-injection attacks of the type described in CVE-2006-2313.
13043 <listitem><para>Reject unsafe uses of <literal>\'</> in string literals</para>
13044 <para>As a server-side defense against SQL-injection attacks of the type
13045 described in CVE-2006-2314, the server now only accepts <literal>''</> and not
13046 <literal>\'</> as a representation of ASCII single quote in SQL string
13047 literals. By default, <literal>\'</> is rejected only when
13048 <varname>client_encoding</> is set to a client-only encoding (SJIS, BIG5, GBK,
13049 GB18030, or UHC), which is the scenario in which SQL injection is possible.
13050 A new configuration parameter <varname>backslash_quote</> is available to
13051 adjust this behavior when needed. Note that full security against
13052 CVE-2006-2314 might require client-side changes; the purpose of
13053 <varname>backslash_quote</> is in part to make it obvious that insecure
13054 clients are insecure.
13057 <listitem><para>Modify <application>libpq</>'s string-escaping routines to be
13058 aware of encoding considerations and
13059 <varname>standard_conforming_strings</></para>
13060 <para>This fixes <application>libpq</>-using applications for the security
13061 issues described in CVE-2006-2313 and CVE-2006-2314, and also future-proofs
13062 them against the planned changeover to SQL-standard string literal syntax.
13063 Applications that use multiple <productname>PostgreSQL</> connections
13064 concurrently should migrate to <function>PQescapeStringConn()</> and
13065 <function>PQescapeByteaConn()</> to ensure that escaping is done correctly
13066 for the settings in use in each database connection. Applications that
13067 do string escaping <quote>by hand</> should be modified to rely on library
13071 <listitem><para>Fix some incorrect encoding conversion functions</para>
13072 <para><function>win1251_to_iso</>, <function>alt_to_iso</>,
13073 <function>euc_tw_to_big5</>, <function>euc_tw_to_mic</>,
13074 <function>mic_to_euc_tw</> were all broken to varying
13078 <listitem><para>Clean up stray remaining uses of <literal>\'</> in strings
13079 (Bruce, Jan)</para></listitem>
13081 <listitem><para>Fix bug that sometimes caused OR'd index scans to
13082 miss rows they should have returned</para></listitem>
13084 <listitem><para>Fix WAL replay for case where a btree index has been
13085 truncated</para></listitem>
13087 <listitem><para>Fix <literal>SIMILAR TO</> for patterns involving
13088 <literal>|</> (Tom)</para></listitem>
13090 <listitem><para>Fix server to use custom DH SSL parameters correctly (Michael
13091 Fuhr)</para></listitem>
13093 <listitem><para>Fix for Bonjour on Intel Macs (Ashley Clark)</para></listitem>
13095 <listitem><para>Fix various minor memory leaks</para></listitem>
13101 <sect1 id="release-7-4-12">
13102 <title>Release 7.4.12</title>
13105 <title>Release date</title>
13106 <simpara>2006-02-14</simpara>
13110 This release contains a variety of fixes from 7.4.11.
13114 <title>Migration to Version 7.4.12</title>
13117 A dump/restore is not required for those running 7.4.X. However,
13118 if you are upgrading from a version earlier than 7.4.11, see the release
13124 <title>Changes</title>
13128 <listitem><para>Fix potential crash in <command>SET
13129 SESSION AUTHORIZATION</> (CVE-2006-0553)</para>
13130 <para>An unprivileged user could crash the server process, resulting in
13131 momentary denial of service to other users, if the server has been compiled
13132 with Asserts enabled (which is not the default).
13133 Thanks to Akio Ishida for reporting this problem.
13136 <listitem><para>Fix bug with row visibility logic in self-inserted
13138 <para>Under rare circumstances a row inserted by the current command
13139 could be seen as already valid, when it should not be. Repairs bug
13140 created in 7.4.9 and 7.3.11 releases.
13143 <listitem><para>Fix race condition that could lead to <quote>file already
13144 exists</> errors during pg_clog file creation
13145 (Tom)</para></listitem>
13147 <listitem><para>Properly check <literal>DOMAIN</> constraints for
13148 <literal>UNKNOWN</> parameters in prepared statements
13149 (Neil)</para></listitem>
13151 <listitem><para>Fix to allow restoring dumps that have cross-schema
13152 references to custom operators (Tom)</para></listitem>
13154 <listitem><para>Portability fix for testing presence of <function>finite</>
13155 and <function>isinf</> during configure (Tom)</para></listitem>
13162 <sect1 id="release-7-4-11">
13163 <title>Release 7.4.11</title>
13166 <title>Release date</title>
13167 <simpara>2006-01-09</simpara>
13171 This release contains a variety of fixes from 7.4.10.
13175 <title>Migration to Version 7.4.11</title>
13178 A dump/restore is not required for those running 7.4.X. However,
13179 if you are upgrading from a version earlier than 7.4.8, see the release
13181 Also, you might need to <command>REINDEX</> indexes on textual
13182 columns after updating, if you are affected by the locale or
13183 <application>plperl</> issues described below.
13188 <title>Changes</title>
13192 <listitem><para>Fix for protocol-level Describe messages issued
13193 outside a transaction or in a failed transaction (Tom)</para></listitem>
13195 <listitem><para>Fix character string comparison for locales that consider
13196 different character combinations as equal, such as Hungarian (Tom)</para>
13197 <para>This might require <command>REINDEX</> to fix existing indexes on
13198 textual columns.</para></listitem>
13200 <listitem><para>Set locale environment variables during postmaster startup
13201 to ensure that <application>plperl</> won't change the locale later</para>
13202 <para>This fixes a problem that occurred if the <application>postmaster</> was
13203 started with environment variables specifying a different locale than what
13204 <application>initdb</> had been told. Under these conditions, any use of
13205 <application>plperl</> was likely to lead to corrupt indexes. You might need
13206 <command>REINDEX</> to fix existing indexes on
13207 textual columns if this has happened to you.</para></listitem>
13209 <listitem><para>Fix longstanding bug in strpos() and regular expression
13210 handling in certain rarely used Asian multi-byte character sets (Tatsuo)
13213 <listitem><para>Fix bug in <filename>/contrib/pgcrypto</> gen_salt,
13214 which caused it not to use all available salt space for MD5 and
13215 XDES algorithms (Marko Kreen, Solar Designer)</para>
13216 <para>Salts for Blowfish and standard DES are unaffected.</para></listitem>
13218 <listitem><para>Fix <filename>/contrib/dblink</> to throw an error,
13219 rather than crashing, when the number of columns specified is different from
13220 what's actually returned by the query (Joe)</para></listitem>
13227 <sect1 id="release-7-4-10">
13228 <title>Release 7.4.10</title>
13231 <title>Release date</title>
13232 <simpara>2005-12-12</simpara>
13236 This release contains a variety of fixes from 7.4.9.
13240 <title>Migration to Version 7.4.10</title>
13243 A dump/restore is not required for those running 7.4.X. However,
13244 if you are upgrading from a version earlier than 7.4.8, see the release
13250 <title>Changes</title>
13254 <listitem><para>Fix race condition in transaction log management</para>
13255 <para>There was a narrow window in which an I/O operation could be initiated
13256 for the wrong page, leading to an Assert failure or data
13260 <listitem><para>Prevent failure if client sends Bind protocol message
13261 when current transaction is already aborted</para></listitem>
13263 <listitem><para><filename>/contrib/ltree</> fixes (Teodor)</para></listitem>
13265 <listitem><para>AIX and HPUX compile fixes (Tom)</para></listitem>
13267 <listitem><para>Fix longstanding planning error for outer joins</para>
13268 <para>This bug sometimes caused a bogus error <quote>RIGHT JOIN is
13269 only supported with merge-joinable join conditions</>.</para></listitem>
13271 <listitem><para>Prevent core dump in <application>pg_autovacuum</> when a
13272 table has been dropped</para></listitem>
13278 <sect1 id="release-7-4-9">
13279 <title>Release 7.4.9</title>
13282 <title>Release date</title>
13283 <simpara>2005-10-04</simpara>
13287 This release contains a variety of fixes from 7.4.8.
13291 <title>Migration to Version 7.4.9</title>
13294 A dump/restore is not required for those running 7.4.X. However,
13295 if you are upgrading from a version earlier than 7.4.8, see the release
13301 <title>Changes</title>
13304 <listitem><para>Fix error that allowed <command>VACUUM</> to remove
13305 <literal>ctid</> chains too soon, and add more checking in code that follows
13306 <literal>ctid</> links</para>
13307 <para>This fixes a long-standing problem that could cause crashes in very rare
13308 circumstances.</para></listitem>
13309 <listitem><para>Fix <type>CHAR()</> to properly pad spaces to the specified
13310 length when using a multiple-byte character set (Yoshiyuki Asaba)</para>
13311 <para>In prior releases, the padding of <type>CHAR()</> was incorrect
13312 because it only padded to the specified number of bytes without
13313 considering how many characters were stored.</para></listitem>
13314 <listitem><para>Fix the sense of the test for read-only transaction
13315 in <command>COPY</></para>
13316 <para>The code formerly prohibited <command>COPY TO</>, where it should
13317 prohibit <command>COPY FROM</>.
13319 <listitem><para>Fix planning problem with outer-join ON clauses that reference
13320 only the inner-side relation</para></listitem>
13321 <listitem><para>Further fixes for <literal>x FULL JOIN y ON true</> corner
13322 cases</para></listitem>
13323 <listitem><para>Make <function>array_in</> and <function>array_recv</> more
13324 paranoid about validating their OID parameter</para></listitem>
13325 <listitem><para>Fix missing rows in queries like <literal>UPDATE a=... WHERE
13326 a...</> with GiST index on column <literal>a</></para></listitem>
13327 <listitem><para>Improve robustness of datetime parsing</para></listitem>
13328 <listitem><para>Improve checking for partially-written WAL
13329 pages</para></listitem>
13330 <listitem><para>Improve robustness of signal handling when SSL is
13331 enabled</para></listitem>
13332 <listitem><para>Don't try to open more than <literal>max_files_per_process</>
13333 files during postmaster startup</para></listitem>
13334 <listitem><para>Various memory leakage fixes</para></listitem>
13335 <listitem><para>Various portability improvements</para></listitem>
13336 <listitem><para>Fix PL/PgSQL to handle <literal>var := var</> correctly when
13337 the variable is of pass-by-reference type</para></listitem>
13338 <listitem><para>Update <filename>contrib/tsearch2</> to use current Snowball
13339 code</para></listitem>
13345 <sect1 id="release-7-4-8">
13346 <title>Release 7.4.8</title>
13349 <title>Release date</title>
13350 <simpara>2005-05-09</simpara>
13354 This release contains a variety of fixes from 7.4.7, including several
13355 security-related issues.
13359 <title>Migration to Version 7.4.8</title>
13362 A dump/restore is not required for those running 7.4.X. However,
13363 it is one possible way of handling two significant security problems
13364 that have been found in the initial contents of 7.4.X system
13365 catalogs. A dump/initdb/reload sequence using 7.4.8's initdb will
13366 automatically correct these problems.
13370 The larger security problem is that the built-in character set encoding
13371 conversion functions can be invoked from SQL commands by unprivileged
13372 users, but the functions were not designed for such use and are not
13373 secure against malicious choices of arguments. The fix involves changing
13374 the declared parameter list of these functions so that they can no longer
13375 be invoked from SQL commands. (This does not affect their normal use
13376 by the encoding conversion machinery.)
13380 The lesser problem is that the <filename>contrib/tsearch2</> module
13381 creates several functions that are misdeclared to return
13382 <type>internal</> when they do not accept <type>internal</> arguments.
13383 This breaks type safety for all functions using <type>internal</>
13388 It is strongly recommended that all installations repair these errors,
13389 either by initdb or by following the manual repair procedures given
13390 below. The errors at least allow unprivileged database users to crash
13391 their server process, and might allow unprivileged users to gain the
13392 privileges of a database superuser.
13396 If you wish not to do an initdb, perform the following procedures instead.
13397 As the database superuser, do:
13401 UPDATE pg_proc SET proargtypes[3] = 'internal'::regtype
13402 WHERE pronamespace = 11 AND pronargs = 5
13403 AND proargtypes[2] = 'cstring'::regtype;
13404 -- The command should report having updated 90 rows;
13405 -- if not, rollback and investigate instead of committing!
13409 Next, if you have installed <filename>contrib/tsearch2</>, do:
13413 UPDATE pg_proc SET proargtypes[0] = 'internal'::regtype
13415 'dex_init(text)'::regprocedure,
13416 'snb_en_init(text)'::regprocedure,
13417 'snb_ru_init(text)'::regprocedure,
13418 'spell_init(text)'::regprocedure,
13419 'syn_init(text)'::regprocedure
13421 -- The command should report having updated 5 rows;
13422 -- if not, rollback and investigate instead of committing!
13426 If this command fails with a message like <quote>function
13427 "dex_init(text)" does not exist</>, then either <filename>tsearch2</>
13428 is not installed in this database, or you already did the update.
13432 The above procedures must be carried out in <emphasis>each</> database
13433 of an installation, including <literal>template1</>, and ideally
13434 including <literal>template0</> as well. If you do not fix the
13435 template databases then any subsequently created databases will contain
13436 the same errors. <literal>template1</> can be fixed in the same way
13437 as any other database, but fixing <literal>template0</> requires
13438 additional steps. First, from any database issue:
13440 UPDATE pg_database SET datallowconn = true WHERE datname = 'template0';
13442 Next connect to <literal>template0</> and perform the above repair
13443 procedures. Finally, do:
13445 -- re-freeze template0:
13447 -- and protect it against future alterations:
13448 UPDATE pg_database SET datallowconn = false WHERE datname = 'template0';
13454 <title>Changes</title>
13457 <listitem><para>Change encoding function signature to prevent
13458 misuse</para></listitem>
13459 <listitem><para>Change <filename>contrib/tsearch2</> to avoid unsafe use of
13460 <type>INTERNAL</> function results</para></listitem>
13461 <listitem><para>Repair ancient race condition that allowed a transaction to be
13462 seen as committed for some purposes (eg SELECT FOR UPDATE) slightly sooner
13463 than for other purposes</para>
13464 <para>This is an extremely serious bug since it could lead to apparent
13465 data inconsistencies being briefly visible to applications.</para></listitem>
13466 <listitem><para>Repair race condition between relation extension and
13468 <para>This could theoretically have caused loss of a page's worth of
13469 freshly-inserted data, although the scenario seems of very low probability.
13470 There are no known cases of it having caused more than an Assert failure.
13472 <listitem><para>Fix comparisons of <type>TIME WITH TIME ZONE</> values</para>
13474 The comparison code was wrong in the case where the
13475 <literal>--enable-integer-datetimes</> configuration switch had been used.
13476 NOTE: if you have an index on a <type>TIME WITH TIME ZONE</> column,
13477 it will need to be <command>REINDEX</>ed after installing this update, because
13478 the fix corrects the sort order of column values.
13480 <listitem><para>Fix <function>EXTRACT(EPOCH)</> for
13481 <type>TIME WITH TIME ZONE</> values</para></listitem>
13482 <listitem><para>Fix mis-display of negative fractional seconds in
13483 <type>INTERVAL</> values</para>
13485 This error only occurred when the
13486 <literal>--enable-integer-datetimes</> configuration switch had been used.
13488 <listitem><para>Ensure operations done during backend shutdown are counted by
13489 statistics collector</para>
13491 This is expected to resolve reports of <application>pg_autovacuum</>
13492 not vacuuming the system catalogs often enough — it was not being
13493 told about catalog deletions caused by temporary table removal during
13496 <listitem><para>Additional buffer overrun checks in plpgsql
13497 (Neil)</para></listitem>
13498 <listitem><para>Fix pg_dump to dump trigger names containing <literal>%</>
13499 correctly (Neil)</para></listitem>
13500 <listitem><para>Fix <filename>contrib/pgcrypto</> for newer OpenSSL builds
13501 (Marko Kreen)</para></listitem>
13502 <listitem><para>Still more 64-bit fixes for
13503 <filename>contrib/intagg</></para></listitem>
13504 <listitem><para>Prevent incorrect optimization of functions returning
13505 <type>RECORD</></para></listitem>
13506 <listitem><para>Prevent <function>to_char(interval)</> from dumping core for
13507 month-related formats</para></listitem>
13508 <listitem><para>Prevent crash on <literal>COALESCE(NULL,NULL)</></para></listitem>
13509 <listitem><para>Fix <function>array_map</> to call PL functions correctly</para></listitem>
13510 <listitem><para>Fix permission checking in <command>ALTER DATABASE RENAME</></para></listitem>
13511 <listitem><para>Fix <command>ALTER LANGUAGE RENAME</></para></listitem>
13512 <listitem><para>Make <function>RemoveFromWaitQueue</> clean up after itself</para>
13514 This fixes a lock management error that would only be visible if a transaction
13515 was kicked out of a wait for a lock (typically by query cancel) and then the
13516 holder of the lock released it within a very narrow window.
13518 <listitem><para>Fix problem with untyped parameter appearing in
13519 <command>INSERT ... SELECT</></para></listitem>
13520 <listitem><para>Fix <command>CLUSTER</> failure after
13521 <command>ALTER TABLE SET WITHOUT OIDS</></para></listitem>
13527 <sect1 id="release-7-4-7">
13528 <title>Release 7.4.7</title>
13531 <title>Release date</title>
13532 <simpara>2005-01-31</simpara>
13536 This release contains a variety of fixes from 7.4.6, including several
13537 security-related issues.
13541 <title>Migration to Version 7.4.7</title>
13544 A dump/restore is not required for those running 7.4.X.
13549 <title>Changes</title>
13552 <listitem><para>Disallow <command>LOAD</> to non-superusers</para>
13554 On platforms that will automatically execute initialization functions of a
13555 shared library (this includes at least Windows and ELF-based Unixen),
13556 <command>LOAD</> can be used to make the server execute arbitrary code.
13557 Thanks to NGS Software for reporting this.</para></listitem>
13558 <listitem><para>Check that creator of an aggregate function has the right to
13559 execute the specified transition functions</para>
13561 This oversight made it possible to bypass denial of EXECUTE
13562 permission on a function.</para></listitem>
13563 <listitem><para>Fix security and 64-bit issues in
13564 contrib/intagg</para></listitem>
13565 <listitem><para>Add needed STRICT marking to some contrib functions (Kris
13566 Jurka)</para></listitem>
13567 <listitem><para>Avoid buffer overrun when plpgsql cursor declaration has too
13568 many parameters (Neil)</para></listitem>
13569 <listitem><para>Fix planning error for FULL and RIGHT outer joins</para>
13571 The result of the join was mistakenly supposed to be sorted the same as the
13572 left input. This could not only deliver mis-sorted output to the user, but
13573 in case of nested merge joins could give outright wrong answers.
13575 <listitem><para>Fix plperl for quote marks in tuple fields</para></listitem>
13576 <listitem><para>Fix display of negative intervals in SQL and GERMAN
13577 datestyles</para></listitem>
13578 <listitem><para>Make age(timestamptz) do calculation in local timezone not
13579 GMT</para></listitem>
13585 <sect1 id="release-7-4-6">
13586 <title>Release 7.4.6</title>
13589 <title>Release date</title>
13590 <simpara>2004-10-22</simpara>
13594 This release contains a variety of fixes from 7.4.5.
13599 <title>Migration to Version 7.4.6</title>
13602 A dump/restore is not required for those running 7.4.X.
13607 <title>Changes</title>
13610 <listitem><para>Repair possible failure to update hint bits on disk</para>
13612 Under rare circumstances this oversight could lead to
13613 <quote>could not access transaction status</> failures, which qualifies
13614 it as a potential-data-loss bug.
13616 <listitem><para>Ensure that hashed outer join does not miss tuples</para>
13618 Very large left joins using a hash join plan could fail to output unmatched
13619 left-side rows given just the right data distribution.
13621 <listitem><para>Disallow running <application>pg_ctl</> as root</para>
13623 This is to guard against any possible security issues.
13625 <listitem><para>Avoid using temp files in <filename>/tmp</> in <command>make_oidjoins_check</command></para>
13627 This has been reported as a security issue, though it's hardly worthy of
13628 concern since there is no reason for non-developers to use this script anyway.
13630 <listitem><para>Prevent forced backend shutdown from re-emitting prior command
13633 In rare cases, a client might think that its last command had succeeded when
13634 it really had been aborted by forced database shutdown.
13636 <listitem><para>Repair bug in <function>pg_stat_get_backend_idset</function></para>
13638 This could lead to misbehavior in some of the system-statistics views.
13640 <listitem><para>Fix small memory leak in postmaster</para></listitem>
13641 <listitem><para>Fix <quote>expected both swapped tables to have TOAST
13642 tables</> bug</para>
13644 This could arise in cases such as CLUSTER after ALTER TABLE DROP COLUMN.
13646 <listitem><para>Prevent <literal>pg_ctl restart</> from adding <literal>-D</> multiple times</para></listitem>
13647 <listitem><para>Fix problem with NULL values in GiST indexes</para></listitem>
13648 <listitem><para><literal>::</> is no longer interpreted as a variable in an
13649 ECPG prepare statement</para></listitem>
13655 <sect1 id="release-7-4-5">
13656 <title>Release 7.4.5</title>
13659 <title>Release date</title>
13660 <simpara>2004-08-18</simpara>
13664 This release contains one serious bug fix over 7.4.4.
13669 <title>Migration to Version 7.4.5</title>
13672 A dump/restore is not required for those running 7.4.X.
13677 <title>Changes</title>
13680 <listitem><para>Repair possible crash during concurrent B-tree index insertions</para>
13682 This patch fixes a rare case in which concurrent insertions into a B-tree index
13683 could result in a server panic. No permanent damage would result, but it's
13684 still worth a re-release. The bug does not exist in pre-7.4 releases.
13691 <sect1 id="release-7-4-4">
13692 <title>Release 7.4.4</title>
13695 <title>Release date</title>
13696 <simpara>2004-08-16</simpara>
13700 This release contains a variety of fixes from 7.4.3.
13705 <title>Migration to Version 7.4.4</title>
13708 A dump/restore is not required for those running 7.4.X.
13713 <title>Changes</title>
13716 <listitem><para>Prevent possible loss of committed transactions during crash</para>
13718 Due to insufficient interlocking between transaction commit and checkpointing,
13719 it was possible for transactions committed just before the most recent
13720 checkpoint to be lost, in whole or in part, following a database crash and
13721 restart. This is a serious bug that has existed
13722 since <productname>PostgreSQL</productname> 7.1.
13724 <listitem><para>Check HAVING restriction before evaluating result list of an
13725 aggregate plan</para></listitem>
13726 <listitem><para>Avoid crash when session's current user ID is deleted</para></listitem>
13727 <listitem><para>Fix hashed crosstab for zero-rows case (Joe)</para></listitem>
13728 <listitem><para>Force cache update after renaming a column in a foreign key</para></listitem>
13729 <listitem><para>Pretty-print UNION queries correctly</para></listitem>
13730 <listitem><para>Make psql handle <literal>\r\n</> newlines properly in COPY IN</para></listitem>
13731 <listitem><para><application>pg_dump</> handled ACLs with grant options incorrectly</para></listitem>
13732 <listitem><para>Fix thread support for OS X and Solaris</para></listitem>
13733 <listitem><para>Updated JDBC driver (build 215) with various fixes</para></listitem>
13734 <listitem><para>ECPG fixes</para></listitem>
13735 <listitem><para>Translation updates (various contributors)</para></listitem>
13741 <sect1 id="release-7-4-3">
13742 <title>Release 7.4.3</title>
13745 <title>Release date</title>
13746 <simpara>2004-06-14</simpara>
13750 This release contains a variety of fixes from 7.4.2.
13755 <title>Migration to Version 7.4.3</title>
13758 A dump/restore is not required for those running 7.4.X.
13763 <title>Changes</title>
13766 <listitem><para>Fix temporary memory leak when using non-hashed aggregates (Tom)</para></listitem>
13767 <listitem><para>ECPG fixes, including some for Informix compatibility (Michael)</para></listitem>
13768 <listitem><para>Fixes for compiling with thread-safety, particularly Solaris (Bruce)</para></listitem>
13769 <listitem><para>Fix error in COPY IN termination when using the old network protocol (ljb)</para></listitem>
13770 <listitem><para>Several important fixes in pg_autovacuum, including fixes for
13771 large tables, unsigned oids, stability, temp tables, and debug mode
13772 (Matthew T. O'Connor)</para></listitem>
13773 <listitem><para>Fix problem with reading tar-format dumps on NetBSD and BSD/OS (Bruce)</para></listitem>
13774 <listitem><para>Several JDBC fixes</para></listitem>
13775 <listitem><para>Fix ALTER SEQUENCE RESTART where last_value equals the restart value (Tom)</para></listitem>
13776 <listitem><para>Repair failure to recalculate nested sub-selects (Tom)</para></listitem>
13777 <listitem><para>Fix problems with non-constant expressions in LIMIT/OFFSET</para></listitem>
13778 <listitem><para>Support FULL JOIN with no join clause, such as X FULL JOIN Y ON TRUE (Tom)</para></listitem>
13779 <listitem><para>Fix another zero-column table bug (Tom)</para></listitem>
13780 <listitem><para>Improve handling of non-qualified identifiers in GROUP BY clauses in sub-selects (Tom)</para>
13782 Select-list aliases within the sub-select will now take precedence over
13783 names from outer query levels.
13785 <listitem><para>Do not generate <quote>NATURAL CROSS JOIN</> when decompiling rules (Tom)</para></listitem>
13786 <listitem><para>Add checks for invalid field length in binary COPY (Tom)</para>
13788 This fixes a difficult-to-exploit security hole.
13790 <listitem><para>Avoid locking conflict between <command>ANALYZE</command> and <command>LISTEN</command>/<command>NOTIFY</command></para></listitem>
13791 <listitem><para>Numerous translation updates (various contributors)</para></listitem>
13797 <sect1 id="release-7-4-2">
13798 <title>Release 7.4.2</title>
13801 <title>Release date</title>
13802 <simpara>2004-03-08</simpara>
13806 This release contains a variety of fixes from 7.4.1.
13811 <title>Migration to Version 7.4.2</title>
13814 A dump/restore is not required for those running 7.4.X. However,
13815 it might be advisable as the easiest method of incorporating fixes for
13816 two errors that have been found in the initial contents of 7.4.X system
13817 catalogs. A dump/initdb/reload sequence using 7.4.2's initdb will
13818 automatically correct these problems.
13822 The more severe of the two errors is that data type <type>anyarray</>
13823 has the wrong alignment label; this is a problem because the
13824 <structname>pg_statistic</> system catalog uses <type>anyarray</>
13825 columns. The mislabeling can cause planner misestimations and even
13826 crashes when planning queries that involve <literal>WHERE</> clauses on
13827 double-aligned columns (such as <type>float8</> and <type>timestamp</>).
13828 It is strongly recommended that all installations repair this error,
13829 either by initdb or by following the manual repair procedure given
13834 The lesser error is that the system view <structname>pg_settings</>
13835 ought to be marked as having public update access, to allow
13836 <literal>UPDATE pg_settings</> to be used as a substitute for
13837 <command>SET</>. This can also be fixed either by initdb or manually,
13838 but it is not necessary to fix unless you want to use <literal>UPDATE
13843 If you wish not to do an initdb, the following procedure will work
13844 for fixing <structname>pg_statistic</>. As the database superuser,
13848 -- clear out old data in pg_statistic:
13849 DELETE FROM pg_statistic;
13850 VACUUM pg_statistic;
13851 -- this should update 1 row:
13852 UPDATE pg_type SET typalign = 'd' WHERE oid = 2277;
13853 -- this should update 6 rows:
13854 UPDATE pg_attribute SET attalign = 'd' WHERE atttypid = 2277;
13856 -- At this point you MUST start a fresh backend to avoid a crash!
13858 -- repopulate pg_statistic:
13862 This can be done in a live database, but beware that all backends
13863 running in the altered database must be restarted before it is safe to
13864 repopulate <structname>pg_statistic</>.
13868 To repair the <structname>pg_settings</> error, simply do:
13870 GRANT SELECT, UPDATE ON pg_settings TO PUBLIC;
13875 The above procedures must be carried out in <emphasis>each</> database
13876 of an installation, including <literal>template1</>, and ideally
13877 including <literal>template0</> as well. If you do not fix the
13878 template databases then any subsequently created databases will contain
13879 the same errors. <literal>template1</> can be fixed in the same way
13880 as any other database, but fixing <literal>template0</> requires
13881 additional steps. First, from any database issue:
13883 UPDATE pg_database SET datallowconn = true WHERE datname = 'template0';
13885 Next connect to <literal>template0</> and perform the above repair
13886 procedures. Finally, do:
13888 -- re-freeze template0:
13890 -- and protect it against future alterations:
13891 UPDATE pg_database SET datallowconn = false WHERE datname = 'template0';
13897 <title>Changes</title>
13900 Release 7.4.2 incorporates all the fixes included in release 7.3.6,
13901 plus the following fixes:
13905 <listitem><para>Fix <structname>pg_statistics</> alignment bug that could crash optimizer</para>
13906 <para>See above for details about this problem.</para></listitem>
13907 <listitem><para>Allow non-super users to update <structname>pg_settings</></para></listitem>
13908 <listitem><para>Fix several optimizer bugs, most of which led to
13909 <quote>variable not found in subplan target lists</> errors</para></listitem>
13910 <listitem><para>Avoid out-of-memory failure during startup of large multiple
13911 index scan</para></listitem>
13912 <listitem><para>Fix multibyte problem that could lead to <quote>out of
13913 memory</> error during <command>COPY IN</></para></listitem>
13914 <listitem><para>Fix problems with <command>SELECT INTO</> / <command>CREATE
13915 TABLE AS</> from tables without OIDs</para></listitem>
13916 <listitem><para>Fix problems with <filename>alter_table</> regression test
13917 during parallel testing</para></listitem>
13918 <listitem><para>Fix problems with hitting open file limit, especially on OS X (Tom)</para></listitem>
13919 <listitem><para>Partial fix for Turkish-locale issues</para>
13920 <para>initdb will succeed now in Turkish locale, but there are still some
13921 inconveniences associated with the <literal>i/I</> problem.</para></listitem>
13922 <listitem><para>Make pg_dump set client encoding on restore</para></listitem>
13923 <listitem><para>Other minor pg_dump fixes</para></listitem>
13924 <listitem><para>Allow ecpg to again use C keywords as column names (Michael)</para></listitem>
13925 <listitem><para>Added ecpg <literal>WHENEVER NOT_FOUND</> to
13926 <literal>SELECT/INSERT/UPDATE/DELETE</> (Michael)</para></listitem>
13927 <listitem><para>Fix ecpg crash for queries calling set-returning functions (Michael)</para></listitem>
13928 <listitem><para>Various other ecpg fixes (Michael)</para></listitem>
13929 <listitem><para>Fixes for Borland compiler</para></listitem>
13930 <listitem><para>Thread build improvements (Bruce)</para></listitem>
13931 <listitem><para>Various other build fixes</para></listitem>
13932 <listitem><para>Various JDBC fixes</para></listitem>
13938 <sect1 id="release-7-4-1">
13939 <title>Release 7.4.1</title>
13942 <title>Release date</title>
13943 <simpara>2003-12-22</simpara>
13947 This release contains a variety of fixes from 7.4.
13952 <title>Migration to Version 7.4.1</title>
13955 A dump/restore is <emphasis>not</emphasis> required for those
13960 If you want to install the fixes in the information schema
13961 you need to reload it into the database.
13962 This is either accomplished by initializing a new cluster
13963 by running <command>initdb</command>, or by running the following
13964 sequence of SQL commands in each database (ideally including
13965 <literal>template1</literal>) as a superuser in
13966 <application>psql</application>, after installing the new release:
13968 DROP SCHEMA information_schema CASCADE;
13969 \i /usr/local/pgsql/share/information_schema.sql
13971 Substitute your installation path in the second command.
13977 <title>Changes</title>
13980 <listitem><para>Fixed bug in <command>CREATE SCHEMA</command> parsing in ECPG (Michael)</para></listitem>
13981 <listitem><para>Fix compile error when <option>--enable-thread-safety</option> and <option>--with-perl</option> are used together (Peter)</para></listitem>
13982 <listitem><para>Fix for subqueries that used hash joins (Tom)</para>
13984 Certain subqueries that used hash joins would crash because of
13985 improperly shared structures.
13987 <listitem><para>Fix free space map compaction bug (Tom)</para>
13989 This fixes a bug where compaction of the free space map could lead
13990 to a database server shutdown.
13993 <listitem><para>Fix for Borland compiler build of libpq (Bruce)</para></listitem>
13994 <listitem><para>Fix <function>netmask()</function> and <function>hostmask()</function> to return the maximum-length masklen (Tom)</para>
13996 Fix these functions to return values consistent with pre-7.4
14000 <listitem><para>Several <filename>contrib/pg_autovacuum</filename> fixes</para>
14002 Fixes include improper variable initialization, missing vacuum after
14003 <command>TRUNCATE</command>, and duration computation overflow for long vacuums.
14006 <listitem><para>Allow compile of <filename>contrib/cube</filename> under Cygwin (Jason Tishler)</para></listitem>
14007 <listitem><para>Fix Solaris use of password file when no passwords are defined (Tom)</para>
14009 Fix crash on Solaris caused by use of any type of password
14010 authentication when no passwords were defined.
14013 <listitem><para>JDBC fix for thread problems, other fixes</para></listitem>
14014 <listitem><para>Fix for <type>bytea</type> index lookups (Joe)</para></listitem>
14015 <listitem><para>Fix information schema for bit data types (Peter)</para></listitem>
14016 <listitem><para>Force zero_damaged_pages to be on during recovery from WAL</para></listitem>
14017 <listitem><para>Prevent some obscure cases of <quote>variable not in subplan target lists</quote></para></listitem>
14018 <listitem><para>Make <function>PQescapeBytea</function> and <function>byteaout</function> consistent with each other (Joe)</para></listitem>
14019 <listitem><para>Escape <type>bytea</type> output for bytes > 0x7e(Joe)</para>
14021 If different client encodings are used for <type>bytea</type> output and input, it
14022 is possible for <type>bytea</type> values to be corrupted by the differing
14023 encodings. This fix escapes all bytes that might be affected.
14026 <listitem><para>Added missing <function>SPI_finish()</function> calls to dblink's <function>get_tuple_of_interest()</function> (Joe)</para></listitem>
14027 <listitem><para>New Czech FAQ</para></listitem>
14028 <listitem><para>Fix information schema view <literal>constraint_column_usage</literal> for foreign keys (Peter)</para></listitem>
14029 <listitem><para>ECPG fixes (Michael)</para></listitem>
14030 <listitem><para>Fix bug with multiple <literal>IN</literal> subqueries and joins in the subqueries (Tom)</para></listitem>
14031 <listitem><para>Allow <literal>COUNT('x')</literal> to work (Tom)</para></listitem>
14032 <listitem><para>Install ECPG include files for Informix compatibility into separate directory (Peter)</para>
14034 Some names of ECPG include files for Informix compatibility conflicted with operating system include files.
14035 By installing them in their own directory, name conflicts have been reduced.
14038 <listitem><para>Fix SSL memory leak (Neil)</para>
14040 This release fixes a bug in 7.4 where SSL didn't free all memory it allocated.
14043 <listitem><para>Prevent <filename>pg_service.conf</filename> from using service name as default dbname (Bruce)</para></listitem>
14044 <listitem><para>Fix local ident authentication on FreeBSD (Tom)</para></listitem>
14050 <sect1 id="release-7-4">
14051 <title>Release 7.4</title>
14054 <title>Release date</title>
14055 <simpara>2003-11-17</simpara>
14059 <title>Overview</title>
14062 Major changes in this release:
14068 <literal>IN</literal> / <literal>NOT IN</literal> subqueries are
14069 now much more efficient
14074 In previous releases, <literal>IN</literal>/<literal>NOT
14075 IN</literal> subqueries were joined to the upper query by
14076 sequentially scanning the subquery looking for a match. The
14077 7.4 code uses the same sophisticated techniques used by
14078 ordinary joins and so is much faster. An
14079 <literal>IN</literal> will now usually be as fast as or faster
14080 than an equivalent <literal>EXISTS</literal> subquery; this
14081 reverses the conventional wisdom that applied to previous
14089 Improved <literal>GROUP BY</literal> processing by using hash buckets
14094 In previous releases, rows to be grouped had to be sorted
14095 first. The 7.4 code can do <literal>GROUP BY</literal>
14096 without sorting, by accumulating results into a hash table
14097 with one entry per group. It will still use the sort
14098 technique, however, if the hash table is estimated to be too
14099 large to fit in <varname>sort_mem</>.
14106 New multikey hash join capability
14111 In previous releases, hash joins could only occur on single
14112 keys. This release allows multicolumn hash joins.
14119 Queries using the explicit <literal>JOIN</literal> syntax are
14120 now better optimized
14125 Prior releases evaluated queries using the explicit
14126 <literal>JOIN</literal> syntax only in the order implied by
14127 the syntax. 7.4 allows full optimization of these queries,
14128 meaning the optimizer considers all possible join orderings
14129 and chooses the most efficient. Outer joins, however, must
14130 still follow the declared ordering.
14137 Faster and more powerful regular expression code
14142 The entire regular expression module has been replaced with a
14143 new version by Henry Spencer, originally written for Tcl. The
14144 code greatly improves performance and supports several flavors
14145 of regular expressions.
14152 Function-inlining for simple SQL functions
14157 Simple SQL functions can now be inlined by including their SQL
14158 in the main query. This improves performance by eliminating
14159 per-call overhead. That means simple SQL functions now
14160 behave like macros.
14167 Full support for IPv6 connections and IPv6 address data types
14172 Previous releases allowed only IPv4 connections, and the IP
14173 data types only supported IPv4 addresses. This release adds
14174 full IPv6 support in both of these areas.
14181 Major improvements in SSL performance and reliability
14186 Several people very familiar with the SSL API have overhauled
14187 our SSL code to improve SSL key negotiation and error
14195 Make free space map efficiently reuse empty index pages,
14196 and other free space management improvements
14201 In previous releases, B-tree index pages that were left empty
14202 because of deleted rows could only be reused by rows with
14203 index values similar to the rows originally indexed on that
14204 page. In 7.4, <command>VACUUM</command> records empty index
14205 pages and allows them to be reused for any future index rows.
14212 SQL-standard information schema
14217 The information schema provides a standardized and stable way
14218 to access information about the schema objects defined in a
14226 Cursors conform more closely to the SQL standard
14231 The commands <command>FETCH</command> and
14232 <command>MOVE</command> have been overhauled to conform more
14233 closely to the SQL standard.
14240 Cursors can exist outside transactions
14245 These cursors are also called holdable cursors.
14252 New client-to-server protocol
14257 The new protocol adds error codes, more status information,
14258 faster startup, better support for binary data transmission,
14259 parameter values separated from SQL commands, prepared
14260 statements available at the protocol level, and cleaner
14261 recovery from <command>COPY</command> failures. The older
14262 protocol is still supported by both server and clients.
14269 <application>libpq</application> and
14270 <application>ECPG</application> applications are now fully
14276 While previous <application>libpq</application> releases
14277 already supported threads, this release improves thread safety
14278 by fixing some non-thread-safe code that was used during
14279 database connection startup. The <command>configure</command>
14280 option <option>--enable-thread-safety</option> must be used to
14281 enable this feature.
14288 New version of full-text indexing
14293 A new full-text indexing suite is available in
14294 <filename>contrib/tsearch2</filename>.
14301 New autovacuum tool
14306 The new autovacuum tool in
14307 <filename>contrib/autovacuum</filename> monitors the database
14308 statistics tables for
14309 <command>INSERT</command>/<command>UPDATE</command>/<command>DELETE</command>
14310 activity and automatically vacuums tables when needed.
14317 Array handling has been improved and moved into the server core
14322 Many array limitations have been removed, and arrays behave
14323 more like fully-supported data types.
14332 <title>Migration to Version 7.4</title>
14335 A dump/restore using <application>pg_dump</application> is
14336 required for those wishing to migrate data from any previous
14341 Observe the following incompatibilities:
14347 The server-side autocommit setting was removed and
14348 reimplemented in client applications and languages.
14349 Server-side autocommit was causing too many problems with
14350 languages and applications that wanted to control their own
14351 autocommit behavior, so autocommit was removed from the server
14352 and added to individual client APIs as appropriate.
14358 Error message wording has changed substantially in this
14359 release. Significant effort was invested to make the messages
14360 more consistent and user-oriented. If your applications try to
14361 detect different error conditions by parsing the error message,
14362 you are strongly encouraged to use the new error code facility instead.
14368 Inner joins using the explicit <literal>JOIN</literal> syntax
14369 might behave differently because they are now better
14376 A number of server configuration parameters have been renamed
14377 for clarity, primarily those related to
14384 <literal>FETCH 0</literal> or <literal>MOVE 0</literal> now
14385 does nothing. In prior releases, <literal>FETCH 0</literal>
14386 would fetch all remaining rows, and <literal>MOVE 0</literal>
14387 would move to the end of the cursor.
14393 <command>FETCH</command> and <command>MOVE</command> now return
14394 the actual number of rows fetched/moved, or zero if at the
14395 beginning/end of the cursor. Prior releases would return the
14396 row count passed to the command, not the number of rows
14397 actually fetched or moved.
14403 <command>COPY</command> now can process files that use
14404 carriage-return or carriage-return/line-feed end-of-line
14405 sequences. Literal carriage-returns and line-feeds are no
14406 longer accepted in data values; use <literal>\r</literal> and
14407 <literal>\n</literal> instead.
14413 Trailing spaces are now trimmed when converting from type
14414 <type>char(<replaceable>n</>)</type> to
14415 <type>varchar(<replaceable>n</>)</type> or <type>text</type>.
14416 This is what most people always expected to happen anyway.
14422 The data type <type>float(<replaceable>p</>)</type> now
14423 measures <replaceable>p</> in binary digits, not decimal
14424 digits. The new behavior follows the SQL standard.
14430 Ambiguous date values now must match the ordering specified by
14431 the <varname>datestyle</varname> setting. In prior releases, a
14432 date specification of <literal>10/20/03</> was interpreted as a
14433 date in October even if <varname>datestyle</> specified that
14434 the day should be first. 7.4 will throw an error if a date
14435 specification is invalid for the current setting of
14436 <varname>datestyle</>.
14442 The functions <function>oidrand</function>,
14443 <function>oidsrand</function>, and
14444 <function>userfntest</function> have been removed. These
14445 functions were determined to be no longer useful.
14451 String literals specifying time-varying date/time values, such
14452 as <literal>'now'</literal> or <literal>'today'</literal> will
14453 no longer work as expected in column default expressions; they
14454 now cause the time of the table creation to be the default, not
14455 the time of the insertion. Functions such as
14456 <function>now()</>, <function>current_timestamp</>, or
14457 <function>current_date</function> should be used instead.
14461 In previous releases, there was special code so that strings
14462 such as <literal>'now'</literal> were interpreted at
14463 <command>INSERT</> time and not at table creation time, but
14464 this work around didn't cover all cases. Release 7.4 now
14465 requires that defaults be defined properly using functions such
14466 as <function>now()</> or <function>current_timestamp</>. These
14467 will work in all situations.
14473 The dollar sign (<literal>$</>) is no longer allowed in
14474 operator names. It can instead be a non-first character in
14475 identifiers. This was done to improve compatibility with other
14476 database systems, and to avoid syntax problems when parameter
14477 placeholders (<literal>$<replaceable>n</></>) are written
14478 adjacent to operators.
14486 <title>Changes</title>
14489 Below you will find a detailed account of the changes between
14490 release 7.4 and the previous major release.
14494 <title>Server Operation Changes</title>
14499 Allow IPv6 server connections (Nigel Kukard, Johan Jordaan,
14500 Bruce, Tom, Kurt Roeckx, Andrew Dunstan)
14506 Fix SSL to handle errors cleanly (Nathan Mueller)
14509 In prior releases, certain SSL API error reports were not
14510 handled correctly. This release fixes those problems.
14516 SSL protocol security and performance improvements (Sean Chittenden)
14519 SSL key renegotiation was happening too frequently, causing poor
14520 SSL performance. Also, initial key handling was improved.
14526 Print lock information when a deadlock is detected (Tom)
14529 This allows easier debugging of deadlock situations.
14535 Update <filename>/tmp</filename> socket modification times
14536 regularly to avoid their removal (Tom)
14539 This should help prevent <filename>/tmp</filename> directory
14540 cleaner administration scripts from removing server socket
14545 <listitem><para>Enable PAM for Mac OS X (Aaron Hillegass)</para></listitem>
14548 <para>Make B-tree indexes fully WAL-safe (Tom)</para>
14550 In prior releases, under certain rare cases, a server crash
14551 could cause B-tree indexes to become corrupt. This release
14552 removes those last few rare cases.
14556 <listitem><para>Allow B-tree index compaction and empty page reuse (Tom)</para></listitem>
14560 Fix inconsistent index lookups during split of first root page (Tom)
14563 In prior releases, when a single-page index split into two
14564 pages, there was a brief period when another database session
14565 could miss seeing an index entry. This release fixes that rare
14570 <listitem><para>Improve free space map allocation logic (Tom)</para></listitem>
14573 <para>Preserve free space information between server restarts (Tom)</para>
14575 In prior releases, the free space map was not saved when the
14576 postmaster was stopped, so newly started servers had no free
14577 space information. This release saves the free space map, and
14578 reloads it when the server is restarted.
14582 <listitem><para>Add start time to <literal>pg_stat_activity</literal> (Neil)</para></listitem>
14583 <listitem><para>New code to detect corrupt disk pages; erase with <varname>zero_damaged_pages</varname> (Tom)</para></listitem>
14584 <listitem><para>New client/server protocol: faster, no username length limit, allow clean exit from <command>COPY</command> (Tom)</para></listitem>
14585 <listitem><para>Add transaction status, table ID, column ID to client/server protocol (Tom)</para></listitem>
14586 <listitem><para>Add binary I/O to client/server protocol (Tom)</para></listitem>
14587 <listitem><para>Remove autocommit server setting; move to client applications (Tom)</para></listitem>
14588 <listitem><para>New error message wording, error codes, and three levels of error detail (Tom, Joe, Peter)</para></listitem>
14593 <title>Performance Improvements</title>
14596 <listitem><para>Add hashing for <literal>GROUP BY</literal> aggregates (Tom)</para></listitem>
14597 <listitem><para>Make nested-loop joins be smarter about multicolumn indexes (Tom)</para></listitem>
14598 <listitem><para>Allow multikey hash joins (Tom)</para></listitem>
14599 <listitem><para>Improve constant folding (Tom)</para></listitem>
14600 <listitem><para>Add ability to inline simple SQL functions (Tom)</para></listitem>
14603 <para>Reduce memory usage for queries using complex functions (Tom)</para>
14605 In prior releases, functions returning allocated memory would
14606 not free it until the query completed. This release allows the
14607 freeing of function-allocated memory when the function call
14608 completes, reducing the total memory used by functions.
14613 <para>Improve GEQO optimizer performance (Tom)</para>
14615 This release fixes several inefficiencies in the way the GEQO optimizer
14616 manages potential query paths.
14622 Allow <literal>IN</>/<literal>NOT IN</> to be handled via hash
14629 Improve <literal>NOT IN (<replaceable>subquery</>)</literal>
14636 Allow most <literal>IN</literal> subqueries to be processed as
14643 Pattern matching operations can use indexes regardless of
14647 There is no way for non-ASCII locales to use the standard
14648 indexes for <literal>LIKE</literal> comparisons. This release
14649 adds a way to create a special index for
14650 <literal>LIKE</literal>.
14655 <para>Allow the postmaster to preload libraries using <varname>preload_libraries</varname> (Joe)</para>
14657 For shared libraries that require a long time to load, this
14658 option is available so the library can be preloaded in the
14659 postmaster and inherited by all database sessions.
14665 Improve optimizer cost computations, particularly for subqueries (Tom)
14671 Avoid sort when subquery <literal>ORDER BY</literal> matches upper query (Tom)
14677 Deduce that <literal>WHERE a.x = b.y AND b.y = 42</literal> also
14678 means <literal>a.x = 42</literal> (Tom)
14684 Allow hash/merge joins on complex joins (Tom)
14690 Allow hash joins for more data types (Tom)
14696 Allow join optimization of explicit inner joins, disable with
14697 <varname>join_collapse_limit</varname> (Tom)
14703 Add parameter <varname>from_collapse_limit</varname> to control
14704 conversion of subqueries to joins (Tom)
14710 Use faster and more powerful regular expression code from Tcl
14711 (Henry Spencer, Tom)
14717 Use bit-mapped relation sets in the optimizer (Tom)
14722 <para>Improve connection startup time (Tom)</para>
14724 The new client/server protocol requires fewer network packets to
14725 start a database session.
14731 Improve trigger/constraint performance (Stephan)
14737 Improve speed of <literal>col IN (const, const, const, ...)</literal> (Tom)
14743 Fix hash indexes which were broken in rare cases (Tom)
14747 <listitem><para>Improve hash index concurrency and speed (Tom)</para>
14749 Prior releases suffered from poor hash index performance,
14750 particularly for high concurrency situations. This release fixes
14751 that, and the development group is interested in reports
14752 comparing B-tree and hash index performance.
14757 <para>Align shared buffers on 32-byte boundary for copy speed improvement (Manfred Spraul)</para>
14759 Certain CPU's perform faster data copies when addresses are
14765 <para>Data type <type>numeric</type> reimplemented for better performance (Tom)</para>
14767 <type>numeric</type> used to be stored in base 100. The new code
14768 uses base 10000, for significantly better performance.
14775 <title>Server Configuration Changes</title>
14779 <para>Rename server parameter <varname>server_min_messages</> to <varname>log_min_messages</> (Bruce)</para>
14781 This was done so most parameters that control the server logs
14782 begin with <literal>log_</>.
14786 <listitem><para>Rename <varname>show_*_stats</> to <varname>log_*_stats</> (Bruce)</para></listitem>
14787 <listitem><para>Rename <varname>show_source_port</> to <varname>log_source_port</> (Bruce)</para></listitem>
14788 <listitem><para>Rename <varname>hostname_lookup</> to <varname>log_hostname</> (Bruce)</para></listitem>
14791 <para>Add <varname>checkpoint_warning</> to warn of excessive checkpointing (Bruce)</para>
14793 In prior releases, it was difficult to determine if checkpoint
14794 was happening too frequently. This feature adds a warning to the
14795 server logs when excessive checkpointing happens.
14799 <listitem><para>New read-only server parameters for localization (Tom)</para></listitem>
14803 Change debug server log messages to output as <literal>DEBUG</>
14804 rather than <literal>LOG</> (Bruce)
14809 <para>Prevent server log variables from being turned off by non-superusers (Bruce)</para>
14811 This is a security feature so non-superusers cannot disable
14812 logging that was enabled by the administrator.
14818 <varname>log_min_messages</>/<varname>client_min_messages</> now
14819 controls <varname>debug_*</> output (Bruce)
14822 This centralizes client debug information so all debug output
14823 can be sent to either the client or server logs.
14828 <para>Add Mac OS X Rendezvous server support (Chris Campbell)</para>
14830 This allows Mac OS X hosts to query the network for available
14831 <productname>PostgreSQL</productname> servers.
14837 Add ability to print only slow statements using
14838 <varname>log_min_duration_statement</varname>
14842 This is an often requested debugging feature that allows
14843 administrators to see only slow queries in their server logs.
14848 <para>Allow <filename>pg_hba.conf</filename> to accept netmasks in CIDR format (Andrew Dunstan)</para>
14850 This allows administrators to merge the host IP address and
14851 netmask fields into a single CIDR field in <filename>pg_hba.conf</filename>.
14855 <listitem><para>New read-only parameter <varname>is_superuser</varname> (Tom)</para></listitem>
14858 <para>New parameter <varname>log_error_verbosity</varname> to control error detail (Tom)</para>
14860 This works with the new error reporting feature to supply
14861 additional error information like hints, file names and line
14867 <para><literal>postgres --describe-config</literal> now dumps server config variables (Aizaz Ahmed, Peter)</para>
14869 This option is useful for administration tools that need to know
14870 the configuration variable names and their minimums, maximums,
14871 defaults, and descriptions.
14877 Add new columns in <literal>pg_settings</literal>:
14878 <literal>context</>, <literal>type</>, <literal>source</>,
14879 <literal>min_val</>, <literal>max_val</> (Joe)
14885 Make default <varname>shared_buffers</> 1000 and
14886 <varname>max_connections</> 100, if possible (Tom)
14889 Prior versions defaulted to 64 shared buffers so <productname>PostgreSQL</productname>
14890 would start on even very old systems. This release tests the
14891 amount of shared memory allowed by the platform and selects more
14892 reasonable default values if possible. Of course, users are
14893 still encouraged to evaluate their resource load and size
14894 <varname>shared_buffers</varname> accordingly.
14900 New <filename>pg_hba.conf</filename> record type
14901 <literal>hostnossl</> to prevent SSL connections (Jon
14905 In prior releases, there was no way to prevent SSL connections
14906 if both the client and server supported SSL. This option allows
14913 Remove parameter <varname>geqo_random_seed</varname>
14920 Add server parameter <varname>regex_flavor</varname> to control regular expression processing (Tom)
14926 Make <command>pg_ctl</command> better handle nonstandard ports (Greg)
14933 <title>Query Changes</title>
14936 <listitem><para>New SQL-standard information schema (Peter)</para></listitem>
14937 <listitem><para>Add read-only transactions (Peter)</para></listitem>
14938 <listitem><para>Print key name and value in foreign-key violation messages (Dmitry Tkach)</para></listitem>
14941 <para>Allow users to see their own queries in <literal>pg_stat_activity</literal> (Kevin Brown)</para>
14943 In prior releases, only the superuser could see query strings
14944 using <literal>pg_stat_activity</literal>. Now ordinary users
14945 can see their own query strings.
14950 <para>Fix aggregates in subqueries to match SQL standard (Tom)</para>
14952 The SQL standard says that an aggregate function appearing
14953 within a nested subquery belongs to the outer query if its
14954 argument contains only outer-query variables. Prior
14955 <productname>PostgreSQL</productname> releases did not handle
14956 this fine point correctly.
14961 <para>Add option to prevent auto-addition of tables referenced in query (Nigel J. Andrews)</para>
14963 By default, tables mentioned in the query are automatically
14964 added to the <literal>FROM</> clause if they are not already
14965 there. This is compatible with historic
14966 <productname>POSTGRES</productname> behavior but is contrary to
14967 the SQL standard. This option allows selecting
14968 standard-compatible behavior.
14973 <para>Allow <literal>UPDATE ... SET col = DEFAULT</literal> (Rod)</para>
14975 This allows <command>UPDATE</command> to set a column to its
14976 declared default value.
14981 <para>Allow expressions to be used in <literal>LIMIT</>/<literal>OFFSET</> (Tom)</para>
14983 In prior releases, <literal>LIMIT</>/<literal>OFFSET</> could
14984 only use constants, not expressions.
14989 <para>Implement <literal>CREATE TABLE AS EXECUTE</literal> (Neil, Peter)</para>
14995 <title>Object Manipulation Changes</title>
14999 <para>Make <command>CREATE SEQUENCE</command> grammar more conforming to SQL:2003 (Neil)</para>
15003 <para>Add statement-level triggers (Neil)</para>
15005 While this allows a trigger to fire at the end of a statement,
15006 it does not allow the trigger to access all rows modified by the
15007 statement. This capability is planned for a future release.
15012 <para>Add check constraints for domains (Rod)</para>
15014 This greatly increases the usefulness of domains by allowing
15015 them to use check constraints.
15020 <para>Add <command>ALTER DOMAIN</command> (Rod)</para>
15022 This allows manipulation of existing domains.
15027 <para>Fix several zero-column table bugs (Tom)</para>
15029 <productname>PostgreSQL</productname> supports zero-column tables. This fixes various bugs
15030 that occur when using such tables.
15035 <para>Have <literal>ALTER TABLE ... ADD PRIMARY KEY</literal> add not-null constraint (Rod)</para>
15037 In prior releases, <literal>ALTER TABLE ... ADD
15038 PRIMARY</literal> would add a unique index, but not a not-null
15039 constraint. That is fixed in this release.
15043 <listitem><para>Add <literal>ALTER TABLE ... WITHOUT OIDS</literal> (Rod)</para>
15045 This allows control over whether new and updated rows will have
15046 an OID column. This is most useful for saving storage space.
15052 Add <literal>ALTER SEQUENCE</literal> to modify minimum, maximum,
15053 increment, cache, cycle values (Rod)
15058 <para>Add <literal>ALTER TABLE ... CLUSTER ON</literal> (Alvaro Herrera)</para>
15060 This command is used by <command>pg_dump</command> to record the
15061 cluster column for each table previously clustered. This
15062 information is used by database-wide cluster to cluster all
15063 previously clustered tables.
15067 <listitem><para>Improve automatic type casting for domains (Rod, Tom)</para></listitem>
15068 <listitem><para>Allow dollar signs in identifiers, except as first character (Tom)</para></listitem>
15069 <listitem><para>Disallow dollar signs in operator names, so <literal>x=$1</> works (Tom)</para></listitem>
15073 Allow copying table schema using <literal>LIKE
15074 <replaceable>subtable</replaceable></literal>, also SQL:2003
15075 feature <literal>INCLUDING DEFAULTS</literal> (Rod)
15081 Add <literal>WITH GRANT OPTION</literal> clause to
15082 <command>GRANT</command> (Peter)
15085 This enabled <command>GRANT</command> to give other users the
15086 ability to grant privileges on a object.
15093 <title>Utility Command Changes</title>
15097 <para>Add <literal>ON COMMIT</literal> clause to <command>CREATE TABLE</command> for temporary tables (Gavin)</para>
15099 This adds the ability for a table to be dropped or all rows
15100 deleted on transaction commit.
15105 <para>Allow cursors outside transactions using <literal>WITH HOLD</literal> (Neil)</para>
15107 In previous releases, cursors were removed at the end of the
15108 transaction that created them. Cursors can now be created with
15109 the <literal>WITH HOLD</literal> option, which allows them to
15110 continue to be accessed after the creating transaction has
15116 <para><literal>FETCH 0</literal> and <literal>MOVE 0 </literal> now do nothing (Bruce)</para>
15118 In previous releases, <literal>FETCH 0</literal> fetched all
15119 remaining rows, and <literal>MOVE 0</literal> moved to the end
15126 Cause <command>FETCH</command> and <command>MOVE</command> to
15127 return the number of rows fetched/moved, or zero if at the
15128 beginning/end of cursor, per SQL standard (Bruce)
15131 In prior releases, the row count returned by
15132 <command>FETCH</command> and <command>MOVE</command> did not
15133 accurately reflect the number of rows processed.
15138 <para>Properly handle <literal>SCROLL</literal> with cursors, or
15139 report an error (Neil)</para>
15141 Allowing random access (both forward and backward scrolling) to
15142 some kinds of queries cannot be done without some additional
15143 work. If <literal>SCROLL</literal> is specified when the cursor
15144 is created, this additional work will be performed. Furthermore,
15145 if the cursor has been created with <literal>NO SCROLL</literal>,
15146 no random access is allowed.
15152 Implement SQL-compatible options <literal>FIRST</>,
15153 <literal>LAST</>, <literal>ABSOLUTE <replaceable>n</></>,
15154 <literal>RELATIVE <replaceable>n</></> for
15155 <command>FETCH</command> and <command>MOVE</command> (Tom)
15160 <para>Allow <command>EXPLAIN</command> on <command>DECLARE CURSOR</command> (Tom)</para>
15164 <para>Allow <command>CLUSTER</command> to use index marked as pre-clustered by default (Alvaro Herrera)</para>
15168 <para>Allow <command>CLUSTER</command> to cluster all tables (Alvaro Herrera)</para>
15170 This allows all previously clustered tables in a database to be
15171 reclustered with a single command.
15175 <listitem><para>Prevent <command>CLUSTER</command> on partial indexes (Tom)</para></listitem>
15177 <listitem><para>Allow DOS and Mac line-endings in <command>COPY</> files (Bruce)</para></listitem>
15181 Disallow literal carriage return as a data value,
15182 backslash-carriage-return and <literal>\r</> are still allowed
15188 <para><command>COPY</> changes (binary, <literal>\.</>) (Tom)</para>
15192 <para>Recover from <command>COPY</command> failure cleanly (Tom)</para>
15196 <para>Prevent possible memory leaks in <command>COPY</command> (Tom)</para>
15200 <para>Make <command>TRUNCATE</command> transaction-safe (Rod)</para>
15202 <command>TRUNCATE</command> can now be used inside a
15203 transaction. If the transaction aborts, the changes made by the
15204 <command>TRUNCATE</command> are automatically rolled back.
15210 Allow prepare/bind of utility commands like
15211 <command>FETCH</command> and <command>EXPLAIN</command> (Tom)
15216 <para>Add <command>EXPLAIN EXECUTE</command> (Neil)</para>
15220 <para>Improve <command>VACUUM</command> performance on indexes by reducing WAL traffic (Tom)</para>
15224 <para>Functional indexes have been generalized into indexes on expressions (Tom)</para>
15226 In prior releases, functional indexes only supported a simple
15227 function applied to one or more column names. This release
15228 allows any type of scalar expression.
15234 Have <command>SHOW TRANSACTION ISOLATION</command> match input
15235 to <command>SET TRANSACTION ISOLATION</command>
15242 Have <command>COMMENT ON DATABASE</command> on nonlocal
15243 database generate a warning, rather than an error (Rod)
15247 Database comments are stored in database-local tables so
15248 comments on a database have to be stored in each database.
15254 Improve reliability of <command>LISTEN</>/<command>NOTIFY</> (Tom)
15259 <para>Allow <command>REINDEX</command> to reliably reindex nonshared system catalog indexes (Tom)</para>
15261 This allows system tables to be reindexed without the
15262 requirement of a standalone session, which was necessary in
15263 previous releases. The only tables that now require a standalone
15264 session for reindexing are the global system tables
15265 <literal>pg_database</>, <literal>pg_shadow</>, and
15266 <literal>pg_group</>.
15273 <title>Data Type and Function Changes</title>
15278 New server parameter <varname>extra_float_digits</varname> to
15279 control precision display of floating-point numbers (Pedro
15283 This controls output precision which was causing regression
15288 <listitem><para>Allow <literal>+1300</literal> as a numeric time-zone specifier, for FJST (Tom)</para></listitem>
15292 Remove rarely used functions <function>oidrand</>,
15293 <function>oidsrand</>, and <function>userfntest</> functions
15299 <para>Add <function>md5()</> function to main server, already in <filename>contrib/pgcrypto</filename> (Joe)</para>
15301 An MD5 function was frequently requested. For more complex
15302 encryption capabilities, use
15303 <filename>contrib/pgcrypto</filename>.
15307 <listitem><para>Increase date range of <type>timestamp</type> (John Cochran)</para></listitem>
15311 Change <literal>EXTRACT(EPOCH FROM timestamp)</literal> so
15312 <type>timestamp without time zone</type> is assumed to be in
15313 local time, not GMT (Tom)
15317 <listitem><para>Trap division by zero in case the operating system doesn't prevent it (Tom)</para></listitem>
15318 <listitem><para>Change the <type>numeric</type> data type internally to base 10000 (Tom)</para></listitem>
15319 <listitem><para>New <function>hostmask()</function> function (Greg Wickham)</para></listitem>
15320 <listitem><para>Fixes for <function>to_char()</function> and <function>to_timestamp()</function> (Karel)</para></listitem>
15324 Allow functions that can take any argument data type and return
15325 any data type, using <type>anyelement</type> and
15326 <type>anyarray</type> (Joe)
15329 This allows the creation of functions that can work with any
15336 Arrays can now be specified as <literal>ARRAY[1,2,3]</literal>,
15337 <literal>ARRAY[['a','b'],['c','d']]</literal>, or
15338 <literal>ARRAY[ARRAY[ARRAY[2]]]</literal> (Joe)
15344 Allow proper comparisons for arrays, including <literal>ORDER
15345 BY</literal> and <literal>DISTINCT</literal> support
15350 <listitem><para>Allow indexes on array columns (Joe)</para></listitem>
15351 <listitem><para>Allow array concatenation with <literal>||</literal> (Joe)</para></listitem>
15355 Allow <literal>WHERE</literal> qualification
15356 <literal><replaceable>expr</> <replaceable>op</> ANY/SOME/ALL
15357 (<replaceable>array_expr</>)</literal> (Joe)
15360 This allows arrays to behave like a list of values, for purposes
15361 like <literal>SELECT * FROM tab WHERE col IN
15362 (array_val)</literal>.
15368 New array functions <function>array_append</>,
15369 <function>array_cat</>, <function>array_lower</>,
15370 <function>array_prepend</>, <function>array_to_string</>,
15371 <function>array_upper</>, <function>string_to_array</> (Joe)
15375 <listitem><para>Allow user defined aggregates to use polymorphic functions (Joe)</para></listitem>
15376 <listitem><para>Allow assignments to empty arrays (Joe)</para></listitem>
15380 Allow 60 in seconds fields of <type>time</type>,
15381 <type>timestamp</type>, and <type>interval</type> input values
15385 Sixty-second values are needed for leap seconds.
15389 <listitem><para>Allow <type>cidr</type> data type to be cast to <type>text</type> (Tom)</para></listitem>
15391 <listitem><para>Disallow invalid time zone names in SET TIMEZONE</para></listitem>
15395 Trim trailing spaces when <type>char</type> is cast to
15396 <type>varchar</> or <type>text</> (Tom)
15402 Make <type>float(<replaceable>p</>)</> measure the precision
15403 <replaceable>p</> in binary digits, not decimal digits
15409 <para>Add IPv6 support to the <type>inet</type> and <type>cidr</type> data types (Michael Graff)</para>
15413 <para>Add <function>family()</function> function to report whether address is IPv4 or IPv6 (Michael Graff)</para>
15418 Have <literal>SHOW datestyle</literal> generate output similar
15419 to that used by <literal>SET datestyle</literal> (Tom)
15425 Make <literal>EXTRACT(TIMEZONE)</literal> and <literal>SET/SHOW
15426 TIME ZONE</literal> follow the SQL convention for the sign of
15427 time zone offsets, i.e., positive is east from UTC (Tom)
15432 <para>Fix <literal>date_trunc('quarter', ...)</literal> (Böjthe Zoltán)</para>
15434 Prior releases returned an incorrect value for this function call.
15439 <para>Make <function>initcap()</function> more compatible with Oracle (Mike Nolan)</para>
15441 <function>initcap()</function> now uppercases a letter appearing
15442 after any non-alphanumeric character, rather than only after
15448 <para>Allow only <varname>datestyle</varname> field order for date values not in ISO-8601 format (Greg)</para>
15453 Add new <varname>datestyle</varname> values <literal>MDY</>,
15454 <literal>DMY</>, and <literal>YMD</> to set input field order;
15455 honor <literal>US</> and <literal>European</> for backward
15456 compatibility (Tom)
15462 String literals like <literal>'now'</literal> or
15463 <literal>'today'</literal> will no longer work as a column
15464 default. Use functions such as <function>now()</function>,
15465 <function>current_timestamp</function> instead. (change
15466 required for prepared statements) (Tom)
15471 <para>Treat NaN as larger than any other value in <function>min()</>/<function>max()</> (Tom)</para>
15473 NaN was already sorted after ordinary numeric values for most
15474 purposes, but <function>min()</> and <function>max()</> didn't
15480 <para>Prevent interval from suppressing <literal>:00</literal>
15481 seconds display</para>
15486 New functions <function>pg_get_triggerdef(prettyprint)</function>
15487 and <function>pg_conversion_is_visible()</function> (Christopher)
15492 <para>Allow time to be specified as <literal>040506</> or <literal>0405</> (Tom)</para>
15497 Input date order must now be <literal>YYYY-MM-DD</literal> (with 4-digit year) or
15498 match <varname>datestyle</varname>
15504 Make <function>pg_get_constraintdef</function> support
15505 unique, primary-key, and check constraints (Christopher)
15512 <title>Server-Side Language Changes</title>
15517 Prevent PL/pgSQL crash when <literal>RETURN NEXT</literal> is
15518 used on a zero-row record variable (Tom)
15524 Make PL/Python's <function>spi_execute</function> interface
15525 handle null values properly (Andrew Bosma)
15530 <para>Allow PL/pgSQL to declare variables of composite types without <literal>%ROWTYPE</literal> (Tom)</para>
15534 <para>Fix PL/Python's <function>_quote()</function> function to handle big integers</para>
15538 <para>Make PL/Python an untrusted language, now called <literal>plpythonu</literal> (Kevin Jacobs, Tom)</para>
15540 The Python language no longer supports a restricted execution
15541 environment, so the trusted version of PL/Python was removed. If
15542 this situation changes, a version of PL/Python that can be used
15543 by non-superusers will be readded.
15548 <para>Allow polymorphic PL/pgSQL functions (Joe, Tom)</para>
15552 <para>Allow polymorphic SQL functions (Joe)</para>
15557 Improved compiled function caching mechanism in PL/pgSQL with
15558 full support for polymorphism (Joe)
15564 Add new parameter <literal>$0</> in PL/pgSQL representing the
15565 function's actual return type (Joe)
15571 Allow PL/Tcl and PL/Python to use the same trigger on multiple tables (Tom)
15577 Fixed PL/Tcl's <function>spi_prepare</function> to accept fully
15578 qualified type names in the parameter type list
15586 <title>psql Changes</title>
15590 <para>Add <literal>\pset pager always</literal> to always use pager (Greg)</para>
15592 This forces the pager to be used even if the number of rows is
15593 less than the screen height. This is valuable for rows that
15594 wrap across several screen rows.
15598 <listitem><para>Improve tab completion (Rod, Ross Reedstrom, Ian Barwick)</para></listitem>
15599 <listitem><para>Reorder <literal>\?</> help into groupings (Harald Armin Massa, Bruce)</para></listitem>
15600 <listitem><para>Add backslash commands for listing schemas, casts, and conversions (Christopher)</para></listitem>
15604 <command>\encoding</> now changes based on the server parameter
15605 <varname>client_encoding</varname> (Tom)
15608 In previous versions, <command>\encoding</command> was not aware
15609 of encoding changes made using <literal>SET
15610 client_encoding</literal>.
15615 <para>Save editor buffer into readline history (Ross)</para>
15617 When <command>\e</> is used to edit a query, the result is saved
15618 in the readline history for retrieval using the up arrow.
15622 <listitem><para>Improve <command>\d</command> display (Christopher)</para></listitem>
15623 <listitem><para>Enhance HTML mode to be more standards-conforming (Greg)</para></listitem>
15626 <para>New <command>\set AUTOCOMMIT off</command> capability (Tom)</para>
15628 This takes the place of the removed server parameter <varname>autocommit</varname>.
15633 <para>New <command>\set VERBOSITY</command> to control error detail (Tom)</para>
15635 This controls the new error reporting details.
15639 <listitem><para>New prompt escape sequence <literal>%x</literal> to show transaction status (Tom)</para></listitem>
15640 <listitem><para>Long options for <application>psql</application> are now available on all platforms</para></listitem>
15645 <title>pg_dump Changes</title>
15648 <listitem><para>Multiple pg_dump fixes, including tar format and large objects</para></listitem>
15649 <listitem><para>Allow pg_dump to dump specific schemas (Neil)</para></listitem>
15652 <para>Make pg_dump preserve column storage characteristics (Christopher)</para>
15654 This preserves <literal>ALTER TABLE ... SET STORAGE</literal> information.
15658 <listitem><para>Make pg_dump preserve <command>CLUSTER</command> characteristics (Christopher)</para></listitem>
15662 Have pg_dumpall use <command>GRANT</>/<command>REVOKE</> to dump database-level privileges (Tom)
15668 Allow pg_dumpall to support the options <option>-a</>,
15669 <option>-s</>, <option>-x</> of pg_dump (Tom)
15673 <listitem><para>Prevent pg_dump from lowercasing identifiers specified on the command line (Tom)</para></listitem>
15677 pg_dump options <option>--use-set-session-authorization</option>
15678 and <option>--no-reconnect</option> now do nothing, all dumps
15679 use <command>SET SESSION AUTHORIZATION</command>
15682 pg_dump no longer reconnects to switch users, but instead always
15683 uses <command>SET SESSION AUTHORIZATION</command>. This will
15684 reduce password prompting during restores.
15689 <para>Long options for <application>pg_dump</application> are now available on all platforms</para>
15691 <productname>PostgreSQL</productname> now includes its own
15692 long-option processing routines.
15699 <title>libpq Changes</title>
15704 Add function <function>PQfreemem</function> for freeing memory on
15705 Windows, suggested for <command>NOTIFY</command> (Bruce)
15708 Windows requires that memory allocated in a library be freed by
15709 a function in the same library, hence
15710 <function>free()</function> doesn't work for freeing memory
15711 allocated by libpq. <function>PQfreemem</function> is the proper
15712 way to free libpq memory, especially on Windows, and is
15713 recommended for other platforms as well.
15718 <para>Document service capability, and add sample file (Bruce)</para>
15720 This allows clients to look up connection information in a
15721 central file on the client machine.
15727 Make <function>PQsetdbLogin</function> have the same defaults as
15728 <function>PQconnectdb</function> (Tom)
15732 <listitem><para>Allow libpq to cleanly fail when result sets are too large (Tom)</para></listitem>
15736 Improve performance of function <function>PQunescapeBytea</function> (Ben Lamb)
15742 Allow thread-safe libpq with <filename>configure</filename>
15743 option <option>--enable-thread-safety</option> (Lee Kindness,
15750 Allow function <function>pqInternalNotice</function> to accept a
15751 format string and arguments instead of just a preformatted
15752 message (Tom, Sean Chittenden)
15758 Control SSL negotiation with <literal>sslmode</literal> values
15759 <literal>disable</literal>, <literal>allow</literal>,
15760 <literal>prefer</literal>, and <literal>require</literal> (Jon
15766 <para>Allow new error codes and levels of text (Tom)</para>
15770 <para>Allow access to the underlying table and column of a query result (Tom)</para>
15772 This is helpful for query-builder applications that want to know
15773 the underlying table and column names associated with a specific
15778 <listitem><para>Allow access to the current transaction status (Tom)</para></listitem>
15779 <listitem><para>Add ability to pass binary data directly to the server (Tom)</para></listitem>
15783 Add function <function>PQexecPrepared</function> and
15784 <function>PQsendQueryPrepared</function> functions which perform
15785 bind/execute of previously prepared statements (Tom)
15792 <title>JDBC Changes</title>
15795 <listitem><para>Allow <function>setNull</function> on updateable result sets</para></listitem>
15796 <listitem><para>Allow <function>executeBatch</function> on a prepared statement (Barry)</para></listitem>
15797 <listitem><para>Support SSL connections (Barry)</para></listitem>
15798 <listitem><para>Handle schema names in result sets (Paul Sorenson)</para></listitem>
15799 <listitem><para>Add refcursor support (Nic Ferrier)</para></listitem>
15804 <title>Miscellaneous Interface Changes</title>
15808 <para>Prevent possible memory leak or core dump during libpgtcl shutdown (Tom)</para>
15811 <para>Add Informix compatibility to ECPG (Michael)</para>
15813 This allows ECPG to process embedded C programs that were
15814 written using certain Informix extensions.
15819 <para>Add type <type>decimal</type> to ECPG that is fixed length, for Informix (Michael)</para>
15824 Allow thread-safe embedded SQL programs with
15825 <filename>configure</filename> option
15826 <option>--enable-thread-safety</option> (Lee Kindness, Bruce)
15829 This allows multiple threads to access the database at the same
15835 <para>Moved Python client PyGreSQL to <ulink url="http://www.pygresql.org"></ulink> (Marc)</para>
15841 <title>Source Code Changes</title>
15844 <listitem><para>Prevent need for separate platform geometry regression result files (Tom)</para></listitem>
15845 <listitem><para>Improved PPC locking primitive (Reinhard Max)</para></listitem>
15846 <listitem><para>New function <function>palloc0</function> to allocate and clear memory (Bruce)</para></listitem>
15847 <listitem><para>Fix locking code for s390x CPU (64-bit) (Tom)</para></listitem>
15848 <listitem><para>Allow OpenBSD to use local ident credentials (William Ahern)</para></listitem>
15849 <listitem><para>Make query plan trees read-only to executor (Tom)</para></listitem>
15850 <listitem><para>Add Darwin startup scripts (David Wheeler)</para></listitem>
15851 <listitem><para>Allow libpq to compile with Borland C++ compiler (Lester Godwin, Karl Waclawek)</para></listitem>
15852 <listitem><para>Use our own version of <function>getopt_long()</function> if needed (Peter)</para></listitem>
15853 <listitem><para>Convert administration scripts to C (Peter)</para></listitem>
15854 <listitem><para> Bison >= 1.85 is now required to build the <productname>PostgreSQL</> grammar, if building from CVS</para></listitem>
15855 <listitem><para>Merge documentation into one book (Peter)</para></listitem>
15856 <listitem><para>Add Windows compatibility functions (Bruce)</para></listitem>
15857 <listitem><para>Allow client interfaces to compile under MinGW (Bruce)</para></listitem>
15858 <listitem><para>New <function>ereport()</function> function for error reporting (Tom)</para></listitem>
15859 <listitem><para>Support Intel compiler on Linux (Peter)</para></listitem>
15860 <listitem><para>Improve Linux startup scripts (Slawomir Sudnik, Darko Prenosil)</para></listitem>
15861 <listitem><para>Add support for AMD Opteron and Itanium (Jeffrey W. Baker, Bruce)</para></listitem>
15863 <para>Remove <option>--enable-recode</option> option from <command>configure</command></para>
15865 This was no longer needed now that we have <command>CREATE CONVERSION</command>.
15869 <para>Generate a compile error if spinlock code is not found (Bruce)</para>
15871 Platforms without spinlock code will now fail to compile, rather
15872 than silently using semaphores. This failure can be disabled
15873 with a new <command>configure</command> option.
15880 <title>Contrib Changes</title>
15883 <listitem><para>Change dbmirror license to BSD</para></listitem>
15884 <listitem><para>Improve earthdistance (Bruno Wolff III)</para></listitem>
15885 <listitem><para>Portability improvements to pgcrypto (Marko Kreen)</para></listitem>
15886 <listitem><para>Prevent crash in xml (John Gray, Michael Richards)</para></listitem>
15887 <listitem><para>Update oracle</para></listitem>
15888 <listitem><para>Update mysql</para></listitem>
15889 <listitem><para>Update cube (Bruno Wolff III)</para></listitem>
15890 <listitem><para>Update earthdistance to use cube (Bruno Wolff III)</para></listitem>
15891 <listitem><para>Update btree_gist (Oleg)</para></listitem>
15892 <listitem><para>New tsearch2 full-text search module (Oleg, Teodor)</para></listitem>
15893 <listitem><para>Add hash-based crosstab function to tablefuncs (Joe)</para></listitem>
15894 <listitem><para>Add serial column to order <function>connectby()</> siblings in tablefuncs (Nabil Sayegh,Joe)</para></listitem>
15895 <listitem><para>Add named persistent connections to dblink (Shridhar Daithanka)</para></listitem>
15896 <listitem><para>New pg_autovacuum allows automatic <command>VACUUM</command> (Matthew T. O'Connor)</para></listitem>
15897 <listitem><para>Make pgbench honor environment variables <envar>PGHOST</>, <envar>PGPORT</>, <envar>PGUSER</> (Tatsuo)</para></listitem>
15898 <listitem><para>Improve intarray (Teodor Sigaev)</para></listitem>
15899 <listitem><para>Improve pgstattuple (Rod)</para></listitem>
15900 <listitem><para>Fix bug in <function>metaphone()</function> in fuzzystrmatch</para></listitem>
15901 <listitem><para>Improve adddepend (Rod)</para></listitem>
15902 <listitem><para>Update spi/timetravel (Böjthe Zoltán)</para></listitem>
15903 <listitem><para>Fix dbase <option>-s</> option and improve non-ASCII handling (Thomas Behr, Márcio Smiderle)</para></listitem>
15904 <listitem><para>Remove array module because features now included by default (Joe)</para></listitem>
15910 <sect1 id="release-7-3-20">
15911 <title>Release 7.3.20</title>
15914 <title>Release date</title>
15915 <simpara>2007-09-17</simpara>
15919 This release contains fixes from 7.3.19.
15923 <title>Migration to Version 7.3.20</title>
15926 A dump/restore is not required for those running 7.3.X. However,
15927 if you are upgrading from a version earlier than 7.3.13, see the release
15934 <title>Changes</title>
15940 Prevent index corruption when a transaction inserts rows and
15941 then aborts close to the end of a concurrent <command>VACUUM</>
15942 on the same table (Tom)
15948 Make <command>CREATE DOMAIN ... DEFAULT NULL</> work properly (Tom)
15954 Fix crash when <varname>log_min_error_statement</> logging runs out
15961 Require non-superusers who use <filename>/contrib/dblink</> to use only
15962 password authentication, as a security measure (Joe)
15971 <sect1 id="release-7-3-19">
15972 <title>Release 7.3.19</title>
15975 <title>Release date</title>
15976 <simpara>2007-04-23</simpara>
15980 This release contains fixes from 7.3.18,
15981 including a security fix.
15985 <title>Migration to Version 7.3.19</title>
15988 A dump/restore is not required for those running 7.3.X. However,
15989 if you are upgrading from a version earlier than 7.3.13, see the release
15996 <title>Changes</title>
16002 Support explicit placement of the temporary-table schema within
16003 <varname>search_path</>, and disable searching it for functions
16004 and operators (Tom)
16007 This is needed to allow a security-definer function to set a
16008 truly secure value of <varname>search_path</>. Without it,
16009 an unprivileged SQL user can use temporary objects to execute code
16010 with the privileges of the security-definer function (CVE-2007-2138).
16011 See <command>CREATE FUNCTION</> for more information.
16017 Fix potential-data-corruption bug in how <command>VACUUM FULL</> handles
16018 <command>UPDATE</> chains (Tom, Pavan Deolasee)
16027 <sect1 id="release-7-3-18">
16028 <title>Release 7.3.18</title>
16031 <title>Release date</title>
16032 <simpara>2007-02-05</simpara>
16036 This release contains a variety of fixes from 7.3.17, including
16041 <title>Migration to Version 7.3.18</title>
16044 A dump/restore is not required for those running 7.3.X. However,
16045 if you are upgrading from a version earlier than 7.3.13, see the release
16052 <title>Changes</title>
16058 Remove security vulnerability that allowed connected users
16059 to read backend memory (Tom)
16062 The vulnerability involves changing the
16063 data type of a table column used in a SQL function (CVE-2007-0555).
16064 This error can easily be exploited to cause a backend crash, and in
16065 principle might be used to read database content that the user
16066 should not be able to access.
16072 Fix rare bug wherein btree index page splits could fail
16073 due to choosing an infeasible split point (Heikki Linnakangas)
16079 Tighten security of multi-byte character processing for UTF8 sequences
16080 over three bytes long (Tom)
16089 <sect1 id="release-7-3-17">
16090 <title>Release 7.3.17</title>
16093 <title>Release date</title>
16094 <simpara>2007-01-08</simpara>
16098 This release contains a variety of fixes from 7.3.16.
16102 <title>Migration to Version 7.3.17</title>
16105 A dump/restore is not required for those running 7.3.X. However,
16106 if you are upgrading from a version earlier than 7.3.13, see the release
16113 <title>Changes</title>
16119 <function>to_number()</> and <function>to_char(numeric)</>
16120 are now <literal>STABLE</>, not <literal>IMMUTABLE</>, for
16121 new <application>initdb</> installs (Tom)
16125 This is because <varname>lc_numeric</> can potentially
16126 change the output of these functions.
16132 Improve index usage of regular expressions that use parentheses (Tom)
16136 This improves <application>psql</> <literal>\d</> performance also.
16145 <sect1 id="release-7-3-16">
16146 <title>Release 7.3.16</title>
16149 <title>Release date</title>
16150 <simpara>2006-10-16</simpara>
16154 This release contains a variety of fixes from 7.3.15.
16158 <title>Migration to Version 7.3.16</title>
16161 A dump/restore is not required for those running 7.3.X. However,
16162 if you are upgrading from a version earlier than 7.3.13, see the release
16169 <title>Changes</title>
16172 <listitem><para>Fix corner cases in pattern matching for
16173 <application>psql</>'s <literal>\d</> commands</para></listitem>
16174 <listitem><para>Fix index-corrupting bugs in /contrib/ltree
16175 (Teodor)</para></listitem>
16176 <listitem><para>Back-port 7.4 spinlock code to improve performance and support
16177 64-bit architectures better</para> </listitem>
16178 <listitem><para>Fix SSL-related memory leak in libpq</para> </listitem>
16179 <listitem><para>Fix backslash escaping in /contrib/dbmirror</para></listitem>
16180 <listitem><para>Adjust regression tests for recent changes in US DST laws
16181 </para> </listitem>
16187 <sect1 id="release-7-3-15">
16188 <title>Release 7.3.15</title>
16191 <title>Release date</title>
16192 <simpara>2006-05-23</simpara>
16196 This release contains a variety of fixes from 7.3.14,
16197 including patches for extremely serious security issues.
16201 <title>Migration to Version 7.3.15</title>
16204 A dump/restore is not required for those running 7.3.X. However,
16205 if you are upgrading from a version earlier than 7.3.13, see the release
16210 Full security against the SQL-injection attacks described in
16211 CVE-2006-2313 and CVE-2006-2314 might require changes in application
16212 code. If you have applications that embed untrustworthy strings
16213 into SQL commands, you should examine them as soon as possible to
16214 ensure that they are using recommended escaping techniques. In
16215 most cases, applications should be using subroutines provided by
16216 libraries or drivers (such as <application>libpq</>'s
16217 <function>PQescapeStringConn()</>) to perform string escaping,
16218 rather than relying on <foreignphrase>ad hoc</> code to do it.
16223 <title>Changes</title>
16226 <listitem><para>Change the server to reject invalidly-encoded multibyte
16227 characters in all cases (Tatsuo, Tom)</para>
16228 <para>While <productname>PostgreSQL</> has been moving in this direction for
16229 some time, the checks are now applied uniformly to all encodings and all
16230 textual input, and are now always errors not merely warnings. This change
16231 defends against SQL-injection attacks of the type described in CVE-2006-2313.
16234 <listitem><para>Reject unsafe uses of <literal>\'</> in string literals</para>
16235 <para>As a server-side defense against SQL-injection attacks of the type
16236 described in CVE-2006-2314, the server now only accepts <literal>''</> and not
16237 <literal>\'</> as a representation of ASCII single quote in SQL string
16238 literals. By default, <literal>\'</> is rejected only when
16239 <varname>client_encoding</> is set to a client-only encoding (SJIS, BIG5, GBK,
16240 GB18030, or UHC), which is the scenario in which SQL injection is possible.
16241 A new configuration parameter <varname>backslash_quote</> is available to
16242 adjust this behavior when needed. Note that full security against
16243 CVE-2006-2314 might require client-side changes; the purpose of
16244 <varname>backslash_quote</> is in part to make it obvious that insecure
16245 clients are insecure.
16248 <listitem><para>Modify <application>libpq</>'s string-escaping routines to be
16249 aware of encoding considerations</para>
16250 <para>This fixes <application>libpq</>-using applications for the security
16251 issues described in CVE-2006-2313 and CVE-2006-2314.
16252 Applications that use multiple <productname>PostgreSQL</> connections
16253 concurrently should migrate to <function>PQescapeStringConn()</> and
16254 <function>PQescapeByteaConn()</> to ensure that escaping is done correctly
16255 for the settings in use in each database connection. Applications that
16256 do string escaping <quote>by hand</> should be modified to rely on library
16260 <listitem><para>Fix some incorrect encoding conversion functions</para>
16261 <para><function>win1251_to_iso</>, <function>alt_to_iso</>,
16262 <function>euc_tw_to_big5</>, <function>euc_tw_to_mic</>,
16263 <function>mic_to_euc_tw</> were all broken to varying
16267 <listitem><para>Clean up stray remaining uses of <literal>\'</> in strings
16268 (Bruce, Jan)</para></listitem>
16270 <listitem><para>Fix server to use custom DH SSL parameters correctly (Michael
16271 Fuhr)</para></listitem>
16273 <listitem><para>Fix various minor memory leaks</para></listitem>
16279 <sect1 id="release-7-3-14">
16280 <title>Release 7.3.14</title>
16283 <title>Release date</title>
16284 <simpara>2006-02-14</simpara>
16288 This release contains a variety of fixes from 7.3.13.
16292 <title>Migration to Version 7.3.14</title>
16295 A dump/restore is not required for those running 7.3.X. However,
16296 if you are upgrading from a version earlier than 7.3.13, see the release
16302 <title>Changes</title>
16306 <listitem><para>Fix potential crash in <command>SET
16307 SESSION AUTHORIZATION</> (CVE-2006-0553)</para>
16308 <para>An unprivileged user could crash the server process, resulting in
16309 momentary denial of service to other users, if the server has been compiled
16310 with Asserts enabled (which is not the default).
16311 Thanks to Akio Ishida for reporting this problem.
16314 <listitem><para>Fix bug with row visibility logic in self-inserted
16316 <para>Under rare circumstances a row inserted by the current command
16317 could be seen as already valid, when it should not be. Repairs bug
16318 created in 7.3.11 release.
16321 <listitem><para>Fix race condition that could lead to <quote>file already
16322 exists</> errors during pg_clog file creation
16323 (Tom)</para></listitem>
16325 <listitem><para>Fix to allow restoring dumps that have cross-schema
16326 references to custom operators (Tom)</para></listitem>
16328 <listitem><para>Portability fix for testing presence of <function>finite</>
16329 and <function>isinf</> during configure (Tom)</para></listitem>
16336 <sect1 id="release-7-3-13">
16337 <title>Release 7.3.13</title>
16340 <title>Release date</title>
16341 <simpara>2006-01-09</simpara>
16345 This release contains a variety of fixes from 7.3.12.
16349 <title>Migration to Version 7.3.13</title>
16352 A dump/restore is not required for those running 7.3.X. However,
16353 if you are upgrading from a version earlier than 7.3.10, see the release
16355 Also, you might need to <command>REINDEX</> indexes on textual
16356 columns after updating, if you are affected by the locale or
16357 <application>plperl</> issues described below.
16362 <title>Changes</title>
16366 <listitem><para>Fix character string comparison for locales that consider
16367 different character combinations as equal, such as Hungarian (Tom)</para>
16368 <para>This might require <command>REINDEX</> to fix existing indexes on
16369 textual columns.</para></listitem>
16371 <listitem><para>Set locale environment variables during postmaster startup
16372 to ensure that <application>plperl</> won't change the locale later</para>
16373 <para>This fixes a problem that occurred if the <application>postmaster</> was
16374 started with environment variables specifying a different locale than what
16375 <application>initdb</> had been told. Under these conditions, any use of
16376 <application>plperl</> was likely to lead to corrupt indexes. You might need
16377 <command>REINDEX</> to fix existing indexes on
16378 textual columns if this has happened to you.</para></listitem>
16380 <listitem><para>Fix longstanding bug in strpos() and regular expression
16381 handling in certain rarely used Asian multi-byte character sets (Tatsuo)
16384 <listitem><para>Fix bug in <filename>/contrib/pgcrypto</> gen_salt,
16385 which caused it not to use all available salt space for MD5 and
16386 XDES algorithms (Marko Kreen, Solar Designer)</para>
16387 <para>Salts for Blowfish and standard DES are unaffected.</para></listitem>
16389 <listitem><para>Fix <filename>/contrib/dblink</> to throw an error,
16390 rather than crashing, when the number of columns specified is different from
16391 what's actually returned by the query (Joe)</para></listitem>
16398 <sect1 id="release-7-3-12">
16399 <title>Release 7.3.12</title>
16402 <title>Release date</title>
16403 <simpara>2005-12-12</simpara>
16407 This release contains a variety of fixes from 7.3.11.
16411 <title>Migration to Version 7.3.12</title>
16414 A dump/restore is not required for those running 7.3.X. However,
16415 if you are upgrading from a version earlier than 7.3.10, see the release
16421 <title>Changes</title>
16425 <listitem><para>Fix race condition in transaction log management</para>
16426 <para>There was a narrow window in which an I/O operation could be initiated
16427 for the wrong page, leading to an Assert failure or data
16431 <listitem><para><filename>/contrib/ltree</> fixes (Teodor)</para></listitem>
16433 <listitem><para>Fix longstanding planning error for outer joins</para>
16434 <para>This bug sometimes caused a bogus error <quote>RIGHT JOIN is
16435 only supported with merge-joinable join conditions</>.</para></listitem>
16437 <listitem><para>Prevent core dump in <application>pg_autovacuum</> when a
16438 table has been dropped</para></listitem>
16445 <sect1 id="release-7-3-11">
16446 <title>Release 7.3.11</title>
16449 <title>Release date</title>
16450 <simpara>2005-10-04</simpara>
16454 This release contains a variety of fixes from 7.3.10.
16458 <title>Migration to Version 7.3.11</title>
16461 A dump/restore is not required for those running 7.3.X. However,
16462 if you are upgrading from a version earlier than 7.3.10, see the release
16468 <title>Changes</title>
16471 <listitem><para>Fix error that allowed <command>VACUUM</> to remove
16472 <literal>ctid</> chains too soon, and add more checking in code that follows
16473 <literal>ctid</> links</para>
16474 <para>This fixes a long-standing problem that could cause crashes in very rare
16475 circumstances.</para></listitem>
16476 <listitem><para>Fix <type>CHAR()</> to properly pad spaces to the specified
16477 length when using a multiple-byte character set (Yoshiyuki Asaba)</para>
16478 <para>In prior releases, the padding of <type>CHAR()</> was incorrect
16479 because it only padded to the specified number of bytes without
16480 considering how many characters were stored.</para></listitem>
16481 <listitem><para>Fix missing rows in queries like <literal>UPDATE a=... WHERE
16482 a...</> with GiST index on column <literal>a</></para></listitem>
16483 <listitem><para>Improve checking for partially-written WAL
16484 pages</para></listitem>
16485 <listitem><para>Improve robustness of signal handling when SSL is
16486 enabled</para></listitem>
16487 <listitem><para>Various memory leakage fixes</para></listitem>
16488 <listitem><para>Various portability improvements</para></listitem>
16489 <listitem><para>Fix PL/PgSQL to handle <literal>var := var</> correctly when
16490 the variable is of pass-by-reference type</para></listitem>
16496 <sect1 id="release-7-3-10">
16497 <title>Release 7.3.10</title>
16500 <title>Release date</title>
16501 <simpara>2005-05-09</simpara>
16505 This release contains a variety of fixes from 7.3.9, including several
16506 security-related issues.
16510 <title>Migration to Version 7.3.10</title>
16513 A dump/restore is not required for those running 7.3.X. However,
16514 it is one possible way of handling a significant security problem
16515 that has been found in the initial contents of 7.3.X system
16516 catalogs. A dump/initdb/reload sequence using 7.3.10's initdb will
16517 automatically correct this problem.
16521 The security problem is that the built-in character set encoding
16522 conversion functions can be invoked from SQL commands by unprivileged
16523 users, but the functions were not designed for such use and are not
16524 secure against malicious choices of arguments. The fix involves changing
16525 the declared parameter list of these functions so that they can no longer
16526 be invoked from SQL commands. (This does not affect their normal use
16527 by the encoding conversion machinery.)
16528 It is strongly recommended that all installations repair this error,
16529 either by initdb or by following the manual repair procedure given
16530 below. The error at least allows unprivileged database users to crash
16531 their server process, and might allow unprivileged users to gain the
16532 privileges of a database superuser.
16536 If you wish not to do an initdb, perform the following procedure instead.
16537 As the database superuser, do:
16541 UPDATE pg_proc SET proargtypes[3] = 'internal'::regtype
16542 WHERE pronamespace = 11 AND pronargs = 5
16543 AND proargtypes[2] = 'cstring'::regtype;
16544 -- The command should report having updated 90 rows;
16545 -- if not, rollback and investigate instead of committing!
16551 The above procedure must be carried out in <emphasis>each</> database
16552 of an installation, including <literal>template1</>, and ideally
16553 including <literal>template0</> as well. If you do not fix the
16554 template databases then any subsequently created databases will contain
16555 the same error. <literal>template1</> can be fixed in the same way
16556 as any other database, but fixing <literal>template0</> requires
16557 additional steps. First, from any database issue:
16559 UPDATE pg_database SET datallowconn = true WHERE datname = 'template0';
16561 Next connect to <literal>template0</> and perform the above repair
16562 procedure. Finally, do:
16564 -- re-freeze template0:
16566 -- and protect it against future alterations:
16567 UPDATE pg_database SET datallowconn = false WHERE datname = 'template0';
16573 <title>Changes</title>
16576 <listitem><para>Change encoding function signature to prevent
16577 misuse</para></listitem>
16578 <listitem><para>Repair ancient race condition that allowed a transaction to be
16579 seen as committed for some purposes (eg SELECT FOR UPDATE) slightly sooner
16580 than for other purposes</para>
16581 <para>This is an extremely serious bug since it could lead to apparent
16582 data inconsistencies being briefly visible to applications.</para></listitem>
16583 <listitem><para>Repair race condition between relation extension and
16585 <para>This could theoretically have caused loss of a page's worth of
16586 freshly-inserted data, although the scenario seems of very low probability.
16587 There are no known cases of it having caused more than an Assert failure.
16589 <listitem><para>Fix comparisons of <type>TIME WITH TIME ZONE</> values</para>
16591 The comparison code was wrong in the case where the
16592 <literal>--enable-integer-datetimes</> configuration switch had been used.
16593 NOTE: if you have an index on a <type>TIME WITH TIME ZONE</> column,
16594 it will need to be <command>REINDEX</>ed after installing this update, because
16595 the fix corrects the sort order of column values.
16597 <listitem><para>Fix <function>EXTRACT(EPOCH)</> for
16598 <type>TIME WITH TIME ZONE</> values</para></listitem>
16599 <listitem><para>Fix mis-display of negative fractional seconds in
16600 <type>INTERVAL</> values</para>
16602 This error only occurred when the
16603 <literal>--enable-integer-datetimes</> configuration switch had been used.
16605 <listitem><para>Additional buffer overrun checks in plpgsql
16606 (Neil)</para></listitem>
16607 <listitem><para>Fix pg_dump to dump trigger names containing <literal>%</>
16608 correctly (Neil)</para></listitem>
16609 <listitem><para>Prevent <function>to_char(interval)</> from dumping core for
16610 month-related formats</para></listitem>
16611 <listitem><para>Fix <filename>contrib/pgcrypto</> for newer OpenSSL builds
16612 (Marko Kreen)</para></listitem>
16613 <listitem><para>Still more 64-bit fixes for
16614 <filename>contrib/intagg</></para></listitem>
16615 <listitem><para>Prevent incorrect optimization of functions returning
16616 <type>RECORD</></para></listitem>
16622 <sect1 id="release-7-3-9">
16623 <title>Release 7.3.9</title>
16626 <title>Release date</title>
16627 <simpara>2005-01-31</simpara>
16631 This release contains a variety of fixes from 7.3.8, including several
16632 security-related issues.
16636 <title>Migration to Version 7.3.9</title>
16639 A dump/restore is not required for those running 7.3.X.
16644 <title>Changes</title>
16647 <listitem><para>Disallow <command>LOAD</> to non-superusers</para>
16649 On platforms that will automatically execute initialization functions of a
16650 shared library (this includes at least Windows and ELF-based Unixen),
16651 <command>LOAD</> can be used to make the server execute arbitrary code.
16652 Thanks to NGS Software for reporting this.</para></listitem>
16653 <listitem><para>Check that creator of an aggregate function has the right to
16654 execute the specified transition functions</para>
16656 This oversight made it possible to bypass denial of EXECUTE
16657 permission on a function.</para></listitem>
16658 <listitem><para>Fix security and 64-bit issues in
16659 contrib/intagg</para></listitem>
16660 <listitem><para>Add needed STRICT marking to some contrib functions (Kris
16661 Jurka)</para></listitem>
16662 <listitem><para>Avoid buffer overrun when plpgsql cursor declaration has too
16663 many parameters (Neil)</para></listitem>
16664 <listitem><para>Fix planning error for FULL and RIGHT outer joins</para>
16666 The result of the join was mistakenly supposed to be sorted the same as the
16667 left input. This could not only deliver mis-sorted output to the user, but
16668 in case of nested merge joins could give outright wrong answers.
16670 <listitem><para>Fix plperl for quote marks in tuple fields</para></listitem>
16671 <listitem><para>Fix display of negative intervals in SQL and GERMAN
16672 datestyles</para></listitem>
16678 <sect1 id="release-7-3-8">
16679 <title>Release 7.3.8</title>
16682 <title>Release date</title>
16683 <simpara>2004-10-22</simpara>
16687 This release contains a variety of fixes from 7.3.7.
16692 <title>Migration to Version 7.3.8</title>
16695 A dump/restore is not required for those running 7.3.X.
16700 <title>Changes</title>
16703 <listitem><para>Repair possible failure to update hint bits on disk</para>
16705 Under rare circumstances this oversight could lead to
16706 <quote>could not access transaction status</> failures, which qualifies
16707 it as a potential-data-loss bug.
16709 <listitem><para>Ensure that hashed outer join does not miss tuples</para>
16711 Very large left joins using a hash join plan could fail to output unmatched
16712 left-side rows given just the right data distribution.
16714 <listitem><para>Disallow running pg_ctl as root</para>
16716 This is to guard against any possible security issues.
16718 <listitem><para>Avoid using temp files in /tmp in make_oidjoins_check</para>
16720 This has been reported as a security issue, though it's hardly worthy of
16721 concern since there is no reason for non-developers to use this script anyway.
16728 <sect1 id="release-7-3-7">
16729 <title>Release 7.3.7</title>
16732 <title>Release date</title>
16733 <simpara>2004-08-16</simpara>
16737 This release contains one critical fix over 7.3.6, and some minor items.
16742 <title>Migration to Version 7.3.7</title>
16745 A dump/restore is not required for those running 7.3.X.
16750 <title>Changes</title>
16753 <listitem><para>Prevent possible loss of committed transactions during crash</para>
16755 Due to insufficient interlocking between transaction commit and checkpointing,
16756 it was possible for transactions committed just before the most recent
16757 checkpoint to be lost, in whole or in part, following a database crash and
16758 restart. This is a serious bug that has existed
16759 since <productname>PostgreSQL</productname> 7.1.
16761 <listitem><para>Remove asymmetrical word processing in tsearch (Teodor)</para></listitem>
16762 <listitem><para>Properly schema-qualify function names when pg_dump'ing a CAST</para></listitem>
16768 <sect1 id="release-7-3-6">
16769 <title>Release 7.3.6</title>
16772 <title>Release date</title>
16773 <simpara>2004-03-02</simpara>
16777 This release contains a variety of fixes from 7.3.5.
16782 <title>Migration to Version 7.3.6</title>
16785 A dump/restore is <emphasis>not</emphasis> required for those
16792 <title>Changes</title>
16795 <listitem><para>Revert erroneous changes in rule permissions checking</para>
16796 <para>A patch applied in 7.3.3 to fix a corner case in rule permissions checks
16797 turns out to have disabled rule-related permissions checks in many
16798 not-so-corner cases. This would for example allow users to insert into views
16799 they weren't supposed to have permission to insert into. We have therefore
16800 reverted the 7.3.3 patch. The original bug will be fixed in 8.0.
16802 <listitem><para>Repair incorrect order of operations in
16803 GetNewTransactionId()</para>
16805 This bug could result in failure under out-of-disk-space conditions, including
16806 inability to restart even after disk space is freed.
16808 <listitem><para>Ensure configure selects -fno-strict-aliasing even when
16809 an external value for CFLAGS is supplied</para>
16811 On some platforms, building with -fstrict-aliasing causes bugs.
16813 <listitem><para>Make pg_restore handle 64-bit off_t correctly</para>
16815 This bug prevented proper restoration from archive files exceeding 4 GB.
16817 <listitem><para>Make contrib/dblink not assume that local and remote type OIDs
16818 match (Joe)</para></listitem>
16819 <listitem><para>Quote connectby()'s start_with argument properly (Joe)</para></listitem>
16820 <listitem><para>Don't crash when a rowtype argument to a plpgsql function is
16821 NULL</para></listitem>
16822 <listitem><para>Avoid generating invalid character encoding sequences in
16823 corner cases when planning LIKE operations</para></listitem>
16824 <listitem><para>Ensure text_position() cannot scan past end of source string
16825 in multibyte cases (Korea PostgreSQL Users' Group)</para></listitem>
16826 <listitem><para>Fix index optimization and selectivity estimates for LIKE
16827 operations on bytea columns (Joe)</para></listitem>
16833 <sect1 id="release-7-3-5">
16834 <title>Release 7.3.5</title>
16837 <title>Release date</title>
16838 <simpara>2003-12-03</simpara>
16842 This has a variety of fixes from 7.3.4.
16847 <title>Migration to Version 7.3.5</title>
16850 A dump/restore is <emphasis>not</emphasis> required for those
16856 <title>Changes</title>
16859 <listitem><para>Force zero_damaged_pages to be on during recovery from WAL</para></listitem>
16860 <listitem><para>Prevent some obscure cases of <quote>variable not in subplan target lists</quote></para></listitem>
16861 <listitem><para>Force stats processes to detach from shared memory, ensuring cleaner shutdown</para></listitem>
16862 <listitem><para>Make PQescapeBytea and byteaout consistent with each other (Joe)</para></listitem>
16863 <listitem><para>Added missing SPI_finish() calls to dblink's get_tuple_of_interest() (Joe)</para></listitem>
16864 <listitem><para>Fix for possible foreign key violation when rule rewrites INSERT (Jan)</para></listitem>
16865 <listitem><para>Support qualified type names in PL/Tcl's spi_prepare command (Jan)</para></listitem>
16866 <listitem><para>Make pg_dump handle a procedural language handler located in pg_catalog</para></listitem>
16867 <listitem><para>Make pg_dump handle cases where a custom opclass is in another schema</para></listitem>
16868 <listitem><para>Make pg_dump dump binary-compatible casts correctly (Jan)</para></listitem>
16869 <listitem><para>Fix insertion of expressions containing subqueries into rule bodies</para></listitem>
16870 <listitem><para>Fix incorrect argument processing in clusterdb script (Anand Ranganathan)</para></listitem>
16871 <listitem><para>Fix problems with dropped columns in plpython triggers</para></listitem>
16872 <listitem><para>Repair problems with to_char() reading past end of its input string (Karel)</para></listitem>
16873 <listitem><para>Fix GB18030 mapping errors (Tatsuo)</para></listitem>
16874 <listitem><para>Fix several problems with SSL error handling and asynchronous SSL I/O</para></listitem>
16875 <listitem><para>Remove ability to bind a list of values to a single parameter in JDBC
16876 (prevents possible SQL-injection attacks)</para></listitem>
16877 <listitem><para>Fix some errors in HAVE_INT64_TIMESTAMP code paths</para></listitem>
16878 <listitem><para>Fix corner case for btree search in parallel with first root page split</para></listitem>
16884 <sect1 id="release-7-3-4">
16885 <title>Release 7.3.4</title>
16888 <title>Release date</title>
16889 <simpara>2003-07-24</simpara>
16893 This has a variety of fixes from 7.3.3.
16898 <title>Migration to Version 7.3.4</title>
16901 A dump/restore is <emphasis>not</emphasis> required for those
16907 <title>Changes</title>
16910 <listitem><para>Repair breakage in timestamp-to-date conversion for dates before 2000</para></listitem>
16911 <listitem><para>Prevent rare possibility of server startup failure (Tom)</para></listitem>
16912 <listitem><para>Fix bugs in interval-to-time conversion (Tom)</para></listitem>
16913 <listitem><para>Add constraint names in a few places in pg_dump (Rod)</para></listitem>
16914 <listitem><para>Improve performance of functions with many parameters (Tom)</para></listitem>
16915 <listitem><para>Fix to_ascii() buffer overruns (Tom)</para></listitem>
16916 <listitem><para>Prevent restore of database comments from throwing an error (Tom)</para></listitem>
16917 <listitem><para>Work around buggy strxfrm() present in some Solaris releases (Tom)</para></listitem>
16918 <listitem><para>Properly escape jdbc setObject() strings to improve security (Barry)</para></listitem>
16924 <sect1 id="release-7-3-3">
16925 <title>Release 7.3.3</title>
16928 <title>Release date</title>
16929 <simpara>2003-05-22</simpara>
16933 This release contains a variety of fixes for version 7.3.2.
16937 <title>Migration to Version 7.3.3</title>
16940 A dump/restore is <emphasis>not</emphasis> required for those
16941 running version 7.3.*.
16946 <title>Changes</title>
16949 <listitem><para>Repair sometimes-incorrect computation of StartUpID after a crash</para></listitem>
16950 <listitem><para>Avoid slowness with lots of deferred triggers in one transaction (Stephan)</para></listitem>
16951 <listitem><para>Don't lock referenced row when <command>UPDATE</command> doesn't change foreign key's value (Jan)</para></listitem>
16952 <listitem><para>Use <command>-fPIC</command> not <command>-fpic</command> on Sparc (Tom Callaway)</para></listitem>
16953 <listitem><para>Repair lack of schema-awareness in contrib/reindexdb</para></listitem>
16954 <listitem><para>Fix contrib/intarray error for zero-element result array (Teodor)</para></listitem>
16955 <listitem><para>Ensure createuser script will exit on control-C (Oliver)</para></listitem>
16956 <listitem><para>Fix errors when the type of a dropped column has itself been dropped</para></listitem>
16957 <listitem><para><command>CHECKPOINT</command> does not cause database panic on failure in noncritical steps</para></listitem>
16958 <listitem><para>Accept 60 in seconds fields of timestamp, time, interval input values</para></listitem>
16959 <listitem><para>Issue notice, not error, if <type>TIMESTAMP</type>,
16960 <type> TIME</type>, or <type>INTERVAL</type> precision too large</para></listitem>
16961 <listitem><para>Fix <function>abstime-to-time</function> cast function (fix is
16962 not applied unless you <application>initdb</application>)</para></listitem>
16963 <listitem><para>Fix <application>pg_proc</application> entry for
16964 <type>timestampt_izone</type> (fix is not applied unless you
16965 <application>initdb</application>)</para></listitem>
16966 <listitem><para>Make <function>EXTRACT(EPOCH FROM timestamp without time zone)</function> treat input as local time</para></listitem>
16967 <listitem><para><command>'now'::timestamptz</command> gave wrong answer if timezone changed earlier in transaction</para></listitem>
16968 <listitem><para><envar>HAVE_INT64_TIMESTAMP</envar> code for time with timezone overwrote its input</para></listitem>
16969 <listitem><para>Accept <command>GLOBAL TEMP/TEMPORARY</command> as a
16970 synonym for <command>TEMPORARY</command></para></listitem>
16971 <listitem><para>Avoid improper schema-privilege-check failure in foreign-key triggers</para></listitem>
16972 <listitem><para>Fix bugs in foreign-key triggers for <command>SET DEFAULT</command> action</para></listitem>
16973 <listitem><para>Fix incorrect time-qual check in row fetch for
16974 <command>UPDATE</command> and <command>DELETE</command> triggers</para></listitem>
16975 <listitem><para>Foreign-key clauses were parsed but ignored in
16976 <command>ALTER TABLE ADD COLUMN</command></para></listitem>
16977 <listitem><para>Fix createlang script breakage for case where handler function already exists</para></listitem>
16978 <listitem><para>Fix misbehavior on zero-column tables in <application>pg_dump</application>, COPY, ANALYZE, other places</para></listitem>
16979 <listitem><para>Fix misbehavior of <function>func_error()</function> on type names containing '%'</para></listitem>
16980 <listitem><para>Fix misbehavior of <function>replace()</function> on strings containing '%'</para></listitem>
16981 <listitem><para>Regular-expression patterns containing certain multibyte characters failed</para></listitem>
16982 <listitem><para>Account correctly for <command>NULL</command>s in more cases in join size estimation</para></listitem>
16983 <listitem><para>Avoid conflict with system definition of <function>isblank()</function> function or macro</para></listitem>
16984 <listitem><para>Fix failure to convert large code point values in EUC_TW conversions (Tatsuo)</para></listitem>
16985 <listitem><para>Fix error recovery for <function>SSL_read</function>/<function>SSL_write</function> calls</para></listitem>
16986 <listitem><para>Don't do early constant-folding of type coercion expressions</para></listitem>
16987 <listitem><para>Validate page header fields immediately after reading in any page</para></listitem>
16988 <listitem><para>Repair incorrect check for ungrouped variables in unnamed joins</para></listitem>
16989 <listitem><para>Fix buffer overrun in <function>to_ascii</function> (Guido Notari)</para></listitem>
16990 <listitem><para>contrib/ltree fixes (Teodor)</para></listitem>
16991 <listitem><para>Fix core dump in deadlock detection on machines where char is unsigned</para></listitem>
16992 <listitem><para>Avoid running out of buffers in many-way indexscan (bug introduced in 7.3)</para></listitem>
16993 <listitem><para>Fix planner's selectivity estimation functions to handle domains properly</para></listitem>
16994 <listitem><para>Fix <application>dbmirror</application> memory-allocation bug (Steven Singer)</para></listitem>
16995 <listitem><para>Prevent infinite loop in <function>ln(numeric)</function> due to roundoff error</para></listitem>
16996 <listitem><para><command>GROUP BY</command> got confused if there were multiple equal GROUP BY items</para></listitem>
16997 <listitem><para>Fix bad plan when inherited <command>UPDATE</command>/<command>DELETE</command> references another inherited table</para></listitem>
16998 <listitem><para>Prevent clustering on incomplete (partial or non-NULL-storing) indexes</para></listitem>
16999 <listitem><para>Service shutdown request at proper time if it arrives while still starting up</para></listitem>
17000 <listitem><para>Fix left-links in temporary indexes (could make backwards scans miss entries)</para></listitem>
17001 <listitem><para>Fix incorrect handling of client_encoding setting in postgresql.conf (Tatsuo)</para></listitem>
17002 <listitem><para>Fix failure to respond to <command>pg_ctl stop -m fast</command> after Async_NotifyHandler runs</para></listitem>
17003 <listitem><para>Fix SPI for case where rule contains multiple statements of the same type</para></listitem>
17004 <listitem><para>Fix problem with checking for wrong type of access privilege in rule query</para></listitem>
17005 <listitem><para>Fix problem with <command>EXCEPT</command> in <command>CREATE RULE</command></para></listitem>
17006 <listitem><para>Prevent problem with dropping temp tables having serial columns</para></listitem>
17007 <listitem><para>Fix replace_vars_with_subplan_refs failure in complex views</para></listitem>
17008 <listitem><para>Fix regexp slowness in single-byte encodings (Tatsuo)</para></listitem>
17009 <listitem><para>Allow qualified type names in <command>CREATE CAST</command>
17010 and <command> DROP CAST</command></para></listitem>
17011 <listitem><para>Accept <function>SETOF type[]</function>, which formerly had to
17012 be written <function>SETOF _type</function></para></listitem>
17013 <listitem><para>Fix <application>pg_dump</application> core dump in some cases with procedural languages</para></listitem>
17014 <listitem><para>Force ISO datestyle in <application>pg_dump</application> output, for portability (Oliver)</para></listitem>
17015 <listitem><para><application>pg_dump</application> failed to handle error return
17016 from <function>lo_read</function> (Oleg Drokin)</para></listitem>
17017 <listitem><para><application>pg_dumpall</application> failed with groups having no members (Nick Eskelinen)</para></listitem>
17018 <listitem><para><application>pg_dumpall</application> failed to recognize --globals-only switch</para></listitem>
17019 <listitem><para>pg_restore failed to restore blobs if -X disable-triggers is specified</para></listitem>
17020 <listitem><para>Repair intrafunction memory leak in plpgsql</para></listitem>
17021 <listitem><para>pltcl's <command>elog</command> command dumped core if given wrong parameters (Ian Harding)</para></listitem>
17022 <listitem><para>plpython used wrong value of <envar>atttypmod</envar> (Brad McLean)</para></listitem>
17023 <listitem><para>Fix improper quoting of boolean values in Python interface (D'Arcy)</para></listitem>
17024 <listitem><para>Added <function>addDataType()</function> method to PGConnection interface for JDBC</para></listitem>
17025 <listitem><para>Fixed various problems with updateable ResultSets for JDBC (Shawn Green)</para></listitem>
17026 <listitem><para>Fixed various problems with DatabaseMetaData for JDBC (Kris Jurka, Peter Royal)</para></listitem>
17027 <listitem><para>Fixed problem with parsing table ACLs in JDBC</para></listitem>
17028 <listitem><para>Better error message for character set conversion problems in JDBC</para></listitem>
17034 <sect1 id="release-7-3-2">
17035 <title>Release 7.3.2</title>
17038 <title>Release date</title>
17039 <simpara>2003-02-04</simpara>
17043 This release contains a variety of fixes for version 7.3.1.
17048 <title>Migration to Version 7.3.2</title>
17051 A dump/restore is <emphasis>not</emphasis> required for those
17052 running version 7.3.*.
17057 <title>Changes</title>
17060 <listitem><para>Restore creation of OID column in CREATE TABLE AS / SELECT INTO</para></listitem>
17061 <listitem><para>Fix <application>pg_dump</> core dump when dumping views having comments</para></listitem>
17062 <listitem><para>Dump DEFERRABLE/INITIALLY DEFERRED constraints properly</para></listitem>
17063 <listitem><para>Fix UPDATE when child table's column numbering differs from parent</para></listitem>
17064 <listitem><para>Increase default value of max_fsm_relations</para></listitem>
17065 <listitem><para>Fix problem when fetching backwards in a cursor for a single-row query</para></listitem>
17066 <listitem><para>Make backward fetch work properly with cursor on SELECT DISTINCT query</para></listitem>
17067 <listitem><para>Fix problems with loading <application>pg_dump</> files containing contrib/lo usage</para></listitem>
17068 <listitem><para>Fix problem with all-numeric user names</para></listitem>
17069 <listitem><para>Fix possible memory leak and core dump during disconnect in libpgtcl</para></listitem>
17070 <listitem><para>Make plpython's spi_execute command handle nulls properly (Andrew Bosma)</para></listitem>
17071 <listitem><para>Adjust plpython error reporting so that its regression test passes again</para></listitem>
17072 <listitem><para>Work with bison 1.875</para></listitem>
17073 <listitem><para>Handle mixed-case names properly in plpgsql's %type (Neil)</para></listitem>
17074 <listitem><para>Fix core dump in pltcl when executing a query rewritten by a rule</para></listitem>
17075 <listitem><para>Repair array subscript overruns (per report from Yichen Xie)</para></listitem>
17076 <listitem><para>Reduce MAX_TIME_PRECISION from 13 to 10 in floating-point case</para></listitem>
17077 <listitem><para>Correctly case-fold variable names in per-database and per-user settings</para></listitem>
17078 <listitem><para>Fix coredump in plpgsql's RETURN NEXT when SELECT into record returns no rows</para></listitem>
17079 <listitem><para>Fix outdated use of pg_type.typprtlen in python client interface</para></listitem>
17080 <listitem><para>Correctly handle fractional seconds in timestamps in JDBC driver</para></listitem>
17081 <listitem><para>Improve performance of getImportedKeys() in JDBC</para></listitem>
17082 <listitem><para>Make shared-library symlinks work standardly on HPUX (Giles)</para></listitem>
17083 <listitem><para>Repair inconsistent rounding behavior for timestamp, time, interval</para></listitem>
17084 <listitem><para>SSL negotiation fixes (Nathan Mueller)</para></listitem>
17085 <listitem><para>Make libpq's ~/.pgpass feature work when connecting with PQconnectDB</para></listitem>
17086 <listitem><para>Update my2pg, ora2pg</para></listitem>
17087 <listitem><para>Translation updates</para></listitem>
17088 <listitem><para>Add casts between types lo and oid in contrib/lo</para></listitem>
17089 <listitem><para>fastpath code now checks for privilege to call function</para></listitem>
17095 <sect1 id="release-7-3-1">
17096 <title>Release 7.3.1</title>
17099 <title>Release date</title>
17100 <simpara>2002-12-18</simpara>
17104 This release contains a variety of fixes for version 7.3.
17109 <title>Migration to Version 7.3.1</title>
17112 A dump/restore is <emphasis>not</emphasis> required for those
17113 running version 7.3. However, it should be noted that the main
17114 <productname>PostgreSQL</productname> interface library, libpq,
17115 has a new major version number for this release, which might require
17116 recompilation of client code in certain cases.
17121 <title>Changes</title>
17124 <listitem><para>Fix a core dump of COPY TO when client/server encodings don't match (Tom)</para></listitem>
17125 <listitem><para>Allow <application>pg_dump</> to work with pre-7.2 servers (Philip)</para></listitem>
17126 <listitem><para>contrib/adddepend fixes (Tom)</para></listitem>
17127 <listitem><para>Fix problem with deletion of per-user/per-database config settings (Tom)</para></listitem>
17128 <listitem><para>contrib/vacuumlo fix (Tom)</para></listitem>
17129 <listitem><para>Allow 'password' encryption even when pg_shadow contains MD5 passwords (Bruce)</para></listitem>
17130 <listitem><para>contrib/dbmirror fix (Steven Singer)</para></listitem>
17131 <listitem><para>Optimizer fixes (Tom)</para></listitem>
17132 <listitem><para>contrib/tsearch fixes (Teodor Sigaev, Magnus)</para></listitem>
17133 <listitem><para>Allow locale names to be mixed case (Nicolai Tufar)</para></listitem>
17134 <listitem><para>Increment libpq library's major version number (Bruce)</para></listitem>
17135 <listitem><para>pg_hba.conf error reporting fixes (Bruce, Neil)</para></listitem>
17136 <listitem><para>Add SCO Openserver 5.0.4 as a supported platform (Bruce)</para></listitem>
17137 <listitem><para>Prevent EXPLAIN from crashing server (Tom)</para></listitem>
17138 <listitem><para>SSL fixes (Nathan Mueller)</para></listitem>
17139 <listitem><para>Prevent composite column creation via ALTER TABLE (Tom)</para></listitem>
17145 <sect1 id="release-7-3">
17146 <title>Release 7.3</title>
17149 <title>Release date</title>
17150 <simpara>2002-11-27</simpara>
17154 <title>Overview</title>
17157 Major changes in this release:
17162 <term>Schemas</term>
17165 Schemas allow users to create objects in separate namespaces,
17166 so two people or applications can have tables with the same
17167 name. There is also a public schema for shared tables.
17168 Table/index creation can be restricted by removing privileges
17169 on the public schema.
17175 <term>Drop Column</term>
17178 <productname>PostgreSQL</productname> now supports the
17179 <literal>ALTER TABLE ... DROP COLUMN</literal> functionality.
17185 <term>Table Functions</term>
17188 Functions returning multiple rows and/or multiple columns are
17189 now much easier to use than before. You can call such a
17190 <quote>table function</quote> in the <literal>SELECT</literal>
17191 <literal>FROM</literal> clause, treating its output like a
17192 table. Also, <application>PL/pgSQL</application> functions can
17199 <term>Prepared Queries</term>
17202 <productname>PostgreSQL</productname> now supports prepared
17203 queries, for improved performance.
17209 <term>Dependency Tracking</term>
17212 <productname>PostgreSQL</productname> now records object
17213 dependencies, which allows improvements in many areas.
17214 <command>DROP</command> statements now take either
17215 <literal>CASCADE</> or <literal>RESTRICT</> to control whether
17216 dependent objects are also dropped.
17222 <term>Privileges</term>
17225 Functions and procedural languages now have privileges, and
17226 functions can be defined to run with the privileges of their
17233 <term>Internationalization</term>
17236 Both multibyte and locale support are now always enabled.
17242 <term>Logging</term>
17245 A variety of logging options have been enhanced.
17251 <term>Interfaces</term>
17254 A large number of interfaces have been moved to <ulink
17255 url="http://gborg.postgresql.org">http://gborg.postgresql.org</>
17256 where they can be developed and released independently.
17262 <term>Functions/Identifiers</term>
17265 By default, functions can now take up to 32 parameters, and
17266 identifiers can be up to 63 bytes long. Also, <literal>OPAQUE</>
17267 is now deprecated: there are specific <quote>pseudo-datatypes</>
17268 to represent each of the former meanings of <literal>OPAQUE</>
17269 in function argument and result types.
17278 <title>Migration to Version 7.3</title>
17281 A dump/restore using <application>pg_dump</> is required for those
17282 wishing to migrate data from any previous release. If your
17283 application examines the system catalogs, additional changes will
17284 be required due to the introduction of schemas in 7.3; for more
17285 information, see: <ulink
17286 url="http://developer.postgresql.org/~momjian/upgrade_tips_7.3"></>.
17290 Observe the following incompatibilities:
17296 Pre-6.3 clients are no longer supported.
17302 <filename>pg_hba.conf</filename> now has a column for the user
17303 name and additional features. Existing files need to be
17310 Several <filename>postgresql.conf</filename> logging parameters
17317 <literal>LIMIT #,#</literal> has been disabled; use
17318 <literal>LIMIT # OFFSET #</literal>.
17324 <command>INSERT</command> statements with column lists must
17325 specify a value for each specified column. For example,
17326 <literal>INSERT INTO tab (col1, col2) VALUES ('val1')</literal>
17327 is now invalid. It's still allowed to supply fewer columns than
17328 expected if the <command>INSERT</command> does not have a column list.
17334 <type>serial</type> columns are no longer automatically
17335 <literal>UNIQUE</>; thus, an index will not automatically be
17342 A <command>SET</command> command inside an aborted transaction
17343 is now rolled back.
17349 <command>COPY</command> no longer considers missing trailing
17350 columns to be null. All columns need to be specified.
17351 (However, one can achieve a similar effect by specifying a
17352 column list in the <command>COPY</command> command.)
17358 The data type <type>timestamp</type> is now equivalent to
17359 <type>timestamp without time zone</type>, instead of
17360 <type>timestamp with time zone</type>.
17366 Pre-7.3 databases loaded into 7.3 will not have the new object
17367 dependencies for <type>serial</type> columns, unique
17368 constraints, and foreign keys. See the directory
17369 <filename>contrib/adddepend/</filename> for a detailed
17370 description and a script that will add such dependencies.
17376 An empty string (<literal>''</literal>) is no longer allowed as
17377 the input into an integer field. Formerly, it was silently
17386 <title>Changes</title>
17389 <title>Server Operation</title>
17391 <listitem><para>Add pg_locks view to show locks (Neil)</para></listitem>
17392 <listitem><para>Security fixes for password negotiation memory allocation (Neil)</para></listitem>
17393 <listitem><para>Remove support for version 0 FE/BE protocol (<productname>PostgreSQL</productname> 6.2 and earlier) (Tom)</para></listitem>
17394 <listitem><para>Reserve the last few backend slots for superusers, add parameter superuser_reserved_connections to control this (Nigel J. Andrews)</para></listitem>
17399 <title>Performance</title>
17401 <listitem><para>Improve startup by calling localtime() only once (Tom)</para></listitem>
17402 <listitem><para>Cache system catalog information in flat files for faster startup (Tom)</para></listitem>
17403 <listitem><para>Improve caching of index information (Tom)</para></listitem>
17404 <listitem><para>Optimizer improvements (Tom, Fernando Nasser)</para></listitem>
17405 <listitem><para>Catalog caches now store failed lookups (Tom)</para></listitem>
17406 <listitem><para>Hash function improvements (Neil)</para></listitem>
17407 <listitem><para>Improve performance of query tokenization and network handling (Peter)</para></listitem>
17408 <listitem><para>Speed improvement for large object restore (Mario Weilguni)</para></listitem>
17409 <listitem><para>Mark expired index entries on first lookup, saving later heap fetches (Tom)</para></listitem>
17410 <listitem><para>Avoid excessive NULL bitmap padding (Manfred Koizar)</para></listitem>
17411 <listitem><para>Add BSD-licensed qsort() for Solaris, for performance (Bruce)</para></listitem>
17412 <listitem><para>Reduce per-row overhead by four bytes (Manfred Koizar)</para></listitem>
17413 <listitem><para>Fix GEQO optimizer bug (Neil Conway)</para></listitem>
17414 <listitem><para>Make WITHOUT OID actually save four bytes per row (Manfred Koizar)</para></listitem>
17415 <listitem><para>Add default_statistics_target variable to specify ANALYZE buckets (Neil)</para></listitem>
17416 <listitem><para>Use local buffer cache for temporary tables so no WAL overhead (Tom)</para></listitem>
17417 <listitem><para>Improve free space map performance on large tables (Stephen Marshall, Tom)</para></listitem>
17418 <listitem><para>Improved WAL write concurrency (Tom)</para></listitem>
17423 <title>Privileges</title>
17425 <listitem><para>Add privileges on functions and procedural languages (Peter)</para></listitem>
17426 <listitem><para>Add OWNER to CREATE DATABASE so superusers can create databases on behalf of unprivileged users (Gavin Sherry, Tom)</para></listitem>
17427 <listitem><para>Add new object privilege bits EXECUTE and USAGE (Tom)</para></listitem>
17428 <listitem><para>Add SET SESSION AUTHORIZATION DEFAULT and RESET SESSION AUTHORIZATION (Tom)</para></listitem>
17429 <listitem><para>Allow functions to be executed with the privilege of the function owner (Peter)</para></listitem>
17434 <title>Server Configuration</title>
17436 <listitem><para>Server log messages now tagged with LOG, not DEBUG (Bruce)</para></listitem>
17437 <listitem><para>Add user column to pg_hba.conf (Bruce)</para></listitem>
17438 <listitem><para>Have log_connections output two lines in log file (Tom)</para></listitem>
17439 <listitem><para>Remove debug_level from postgresql.conf, now server_min_messages (Bruce)</para></listitem>
17440 <listitem><para>New ALTER DATABASE/USER ... SET command for per-user/database initialization (Peter)</para></listitem>
17441 <listitem><para>New parameters server_min_messages and client_min_messages to control which messages are sent to the server logs or client applications (Bruce)</para></listitem>
17442 <listitem><para>Allow pg_hba.conf to specify lists of users/databases separated by commas, group names prepended with +, and file names prepended with @ (Bruce)</para></listitem>
17443 <listitem><para>Remove secondary password file capability and pg_password utility (Bruce)</para></listitem>
17444 <listitem><para>Add variable db_user_namespace for database-local user names (Bruce)</para></listitem>
17445 <listitem><para>SSL improvements (Bear Giles)</para></listitem>
17446 <listitem><para>Make encryption of stored passwords the default (Bruce)</para></listitem>
17447 <listitem><para>Allow pg_statistics to be reset by calling pg_stat_reset() (Christopher)</para></listitem>
17448 <listitem><para>Add log_duration parameter (Bruce)</para></listitem>
17449 <listitem><para>Rename debug_print_query to log_statement (Bruce)</para></listitem>
17450 <listitem><para>Rename show_query_stats to show_statement_stats (Bruce)</para></listitem>
17451 <listitem><para>Add param log_min_error_statement to print commands to logs on error (Gavin)</para></listitem>
17456 <title>Queries</title>
17458 <listitem><para>Make cursors insensitive, meaning their contents do not change (Tom)</para></listitem>
17459 <listitem><para>Disable LIMIT #,# syntax; now only LIMIT # OFFSET # supported (Bruce)</para></listitem>
17460 <listitem><para>Increase identifier length to 63 (Neil, Bruce)</para></listitem>
17461 <listitem><para>UNION fixes for merging >= 3 columns of different lengths (Tom)</para></listitem>
17462 <listitem><para>Add DEFAULT key word to INSERT, e.g., INSERT ... (..., DEFAULT, ...) (Rod)</para></listitem>
17463 <listitem><para>Allow views to have default values using ALTER COLUMN ... SET DEFAULT (Neil)</para></listitem>
17464 <listitem><para>Fail on INSERTs with column lists that don't supply all column values, e.g., INSERT INTO tab (col1, col2) VALUES ('val1'); (Rod)</para></listitem>
17465 <listitem><para>Fix for join aliases (Tom)</para></listitem>
17466 <listitem><para>Fix for FULL OUTER JOINs (Tom)</para></listitem>
17467 <listitem><para>Improve reporting of invalid identifier and location (Tom, Gavin)</para></listitem>
17468 <listitem><para>Fix OPEN cursor(args) (Tom)</para></listitem>
17469 <listitem><para>Allow 'ctid' to be used in a view and currtid(viewname) (Hiroshi)</para></listitem>
17470 <listitem><para>Fix for CREATE TABLE AS with UNION (Tom)</para></listitem>
17471 <listitem><para>SQL99 syntax improvements (Thomas)</para></listitem>
17472 <listitem><para>Add statement_timeout variable to cancel queries (Bruce)</para></listitem>
17473 <listitem><para>Allow prepared queries with PREPARE/EXECUTE (Neil)</para></listitem>
17474 <listitem><para>Allow FOR UPDATE to appear after LIMIT/OFFSET (Bruce)</para></listitem>
17475 <listitem><para>Add variable autocommit (Tom, David Van Wie)</para></listitem>
17480 <title>Object Manipulation</title>
17482 <listitem><para>Make equals signs optional in CREATE DATABASE (Gavin Sherry)</para></listitem>
17483 <listitem><para>Make ALTER TABLE OWNER change index ownership too (Neil)</para></listitem>
17484 <listitem><para>New ALTER TABLE tabname ALTER COLUMN colname SET STORAGE controls TOAST storage, compression (John Gray)</para></listitem>
17485 <listitem><para>Add schema support, CREATE/DROP SCHEMA (Tom)</para></listitem>
17486 <listitem><para>Create schema for temporary tables (Tom)</para></listitem>
17487 <listitem><para>Add variable search_path for schema search (Tom)</para></listitem>
17488 <listitem><para>Add ALTER TABLE SET/DROP NOT NULL (Christopher)</para></listitem>
17489 <listitem><para>New CREATE FUNCTION volatility levels (Tom)</para></listitem>
17490 <listitem><para>Make rule names unique only per table (Tom)</para></listitem>
17491 <listitem><para>Add 'ON tablename' clause to DROP RULE and COMMENT ON RULE (Tom)</para></listitem>
17492 <listitem><para>Add ALTER TRIGGER RENAME (Joe)</para></listitem>
17493 <listitem><para>New current_schema() and current_schemas() inquiry functions (Tom)</para></listitem>
17494 <listitem><para>Allow functions to return multiple rows (table functions) (Joe)</para></listitem>
17495 <listitem><para>Make WITH optional in CREATE DATABASE, for consistency (Bruce)</para></listitem>
17496 <listitem><para>Add object dependency tracking (Rod, Tom)</para></listitem>
17497 <listitem><para>Add RESTRICT/CASCADE to DROP commands (Rod)</para></listitem>
17498 <listitem><para>Add ALTER TABLE DROP for non-CHECK CONSTRAINT (Rod)</para></listitem>
17499 <listitem><para>Autodestroy sequence on DROP of table with SERIAL (Rod)</para></listitem>
17500 <listitem><para>Prevent column dropping if column is used by foreign key (Rod)</para></listitem>
17501 <listitem><para>Automatically drop constraints/functions when object is dropped (Rod)</para></listitem>
17502 <listitem><para>Add CREATE/DROP OPERATOR CLASS (Bill Studenmund, Tom)</para></listitem>
17503 <listitem><para>Add ALTER TABLE DROP COLUMN (Christopher, Tom, Hiroshi)</para></listitem>
17504 <listitem><para>Prevent inherited columns from being removed or renamed (Alvaro Herrera)</para></listitem>
17505 <listitem><para>Fix foreign key constraints to not error on intermediate database states (Stephan)</para></listitem>
17506 <listitem><para>Propagate column or table renaming to foreign key constraints</para></listitem>
17507 <listitem><para>Add CREATE OR REPLACE VIEW (Gavin, Neil, Tom)</para></listitem>
17508 <listitem><para>Add CREATE OR REPLACE RULE (Gavin, Neil, Tom)</para></listitem>
17509 <listitem><para>Have rules execute alphabetically, returning more predictable values (Tom)</para></listitem>
17510 <listitem><para>Triggers are now fired in alphabetical order (Tom)</para></listitem>
17511 <listitem><para>Add /contrib/adddepend to handle pre-7.3 object dependencies (Rod)</para></listitem>
17512 <listitem><para>Allow better casting when inserting/updating values (Tom)</para></listitem>
17517 <title>Utility Commands</title>
17519 <listitem><para>Have COPY TO output embedded carriage returns and newlines as \r and \n (Tom)</para></listitem>
17520 <listitem><para>Allow DELIMITER in COPY FROM to be 8-bit clean (Tatsuo)</para></listitem>
17521 <listitem><para>Make <application>pg_dump</> use ALTER TABLE ADD PRIMARY KEY, for performance (Neil)</para></listitem>
17522 <listitem><para>Disable brackets in multistatement rules (Bruce)</para></listitem>
17523 <listitem><para>Disable VACUUM from being called inside a function (Bruce)</para></listitem>
17524 <listitem><para>Allow dropdb and other scripts to use identifiers with spaces (Bruce)</para></listitem>
17525 <listitem><para>Restrict database comment changes to the current database</para></listitem>
17526 <listitem><para>Allow comments on operators, independent of the underlying function (Rod)</para></listitem>
17527 <listitem><para>Rollback SET commands in aborted transactions (Tom)</para></listitem>
17528 <listitem><para>EXPLAIN now outputs as a query (Tom)</para></listitem>
17529 <listitem><para>Display condition expressions and sort keys in EXPLAIN (Tom)</para></listitem>
17530 <listitem><para>Add 'SET LOCAL var = value' to set configuration variables for a single transaction (Tom)</para></listitem>
17531 <listitem><para>Allow ANALYZE to run in a transaction (Bruce)</para></listitem>
17532 <listitem><para>Improve COPY syntax using new WITH clauses, keep backward compatibility (Bruce)</para></listitem>
17533 <listitem><para>Fix <application>pg_dump</> to consistently output tags in non-ASCII dumps (Bruce)</para></listitem>
17534 <listitem><para>Make foreign key constraints clearer in dump file (Rod)</para></listitem>
17535 <listitem><para>Add COMMENT ON CONSTRAINT (Rod)</para></listitem>
17536 <listitem><para>Allow COPY TO/FROM to specify column names (Brent Verner)</para></listitem>
17537 <listitem><para>Dump UNIQUE and PRIMARY KEY constraints as ALTER TABLE (Rod)</para></listitem>
17538 <listitem><para>Have SHOW output a query result (Joe)</para></listitem>
17539 <listitem><para>Generate failure on short COPY lines rather than pad NULLs (Neil)</para></listitem>
17540 <listitem><para>Fix CLUSTER to preserve all table attributes (Alvaro Herrera)</para></listitem>
17541 <listitem><para>New pg_settings table to view/modify GUC settings (Joe)</para></listitem>
17542 <listitem><para>Add smart quoting, portability improvements to <application>pg_dump</> output (Peter)</para></listitem>
17543 <listitem><para>Dump serial columns out as SERIAL (Tom)</para></listitem>
17544 <listitem><para>Enable large file support, >2G for <application>pg_dump</> (Peter, Philip Warner, Bruce)</para></listitem>
17545 <listitem><para>Disallow TRUNCATE on tables that are involved in referential constraints (Rod)</para></listitem>
17546 <listitem><para>Have TRUNCATE also auto-truncate the toast table of the relation (Tom)</para></listitem>
17547 <listitem><para>Add clusterdb utility that will auto-cluster an entire database based on previous CLUSTER operations (Alvaro Herrera)</para></listitem>
17548 <listitem><para>Overhaul pg_dumpall (Peter)</para></listitem>
17549 <listitem><para>Allow REINDEX of TOAST tables (Tom)</para></listitem>
17550 <listitem><para>Implemented START TRANSACTION, per SQL99 (Neil)</para></listitem>
17551 <listitem><para>Fix rare index corruption when a page split affects bulk delete (Tom)</para></listitem>
17552 <listitem><para>Fix ALTER TABLE ... ADD COLUMN for inheritance (Alvaro Herrera)</para></listitem>
17557 <title>Data Types and Functions</title>
17559 <listitem><para>Fix factorial(0) to return 1 (Bruce)</para></listitem>
17560 <listitem><para>Date/time/timezone improvements (Thomas)</para></listitem>
17561 <listitem><para>Fix for array slice extraction (Tom)</para></listitem>
17562 <listitem><para>Fix extract/date_part to report proper microseconds for timestamp (Tatsuo)</para></listitem>
17563 <listitem><para>Allow text_substr() and bytea_substr() to read TOAST values more efficiently (John Gray)</para></listitem>
17564 <listitem><para>Add domain support (Rod)</para></listitem>
17565 <listitem><para>Make WITHOUT TIME ZONE the default for TIMESTAMP and TIME data types (Thomas)</para></listitem>
17566 <listitem><para>Allow alternate storage scheme of 64-bit integers for date/time types using --enable-integer-datetimes in configure (Thomas)</para></listitem>
17567 <listitem><para>Make timezone(timestamptz) return timestamp rather than a string (Thomas)</para></listitem>
17568 <listitem><para>Allow fractional seconds in date/time types for dates prior to 1BC (Thomas)</para></listitem>
17569 <listitem><para>Limit timestamp data types to 6 decimal places of precision (Thomas)</para></listitem>
17570 <listitem><para>Change timezone conversion functions from timetz() to timezone() (Thomas)</para></listitem>
17571 <listitem><para>Add configuration variables datestyle and timezone (Tom)</para></listitem>
17572 <listitem><para>Add OVERLAY(), which allows substitution of a substring in a string (Thomas)</para></listitem>
17573 <listitem><para>Add SIMILAR TO (Thomas, Tom)</para></listitem>
17574 <listitem><para>Add regular expression SUBSTRING(string FROM pat FOR escape) (Thomas)</para></listitem>
17575 <listitem><para>Add LOCALTIME and LOCALTIMESTAMP functions (Thomas)</para></listitem>
17576 <listitem><para>Add named composite types using CREATE TYPE typename AS (column) (Joe)</para></listitem>
17577 <listitem><para>Allow composite type definition in the table alias clause (Joe)</para></listitem>
17578 <listitem><para>Add new API to simplify creation of C language table functions (Joe)</para></listitem>
17579 <listitem><para>Remove ODBC-compatible empty parentheses from calls to SQL99 functions for which these parentheses do not match the standard (Thomas)</para></listitem>
17580 <listitem><para>Allow macaddr data type to accept 12 hex digits with no separators (Mike Wyer)</para></listitem>
17581 <listitem><para>Add CREATE/DROP CAST (Peter)</para></listitem>
17582 <listitem><para>Add IS DISTINCT FROM operator (Thomas)</para></listitem>
17583 <listitem><para>Add SQL99 TREAT() function, synonym for CAST() (Thomas)</para></listitem>
17584 <listitem><para>Add pg_backend_pid() to output backend pid (Bruce)</para></listitem>
17585 <listitem><para>Add IS OF / IS NOT OF type predicate (Thomas)</para></listitem>
17586 <listitem><para>Allow bit string constants without fully-specified length (Thomas)</para></listitem>
17587 <listitem><para>Allow conversion between 8-byte integers and bit strings (Thomas)</para></listitem>
17588 <listitem><para>Implement hex literal conversion to bit string literal (Thomas)</para></listitem>
17589 <listitem><para>Allow table functions to appear in the FROM clause (Joe)</para></listitem>
17590 <listitem><para>Increase maximum number of function parameters to 32 (Bruce)</para></listitem>
17591 <listitem><para>No longer automatically create index for SERIAL column (Tom)</para></listitem>
17592 <listitem><para>Add current_database() (Rod)</para></listitem>
17593 <listitem><para>Fix cash_words() to not overflow buffer (Tom)</para></listitem>
17594 <listitem><para>Add functions replace(), split_part(), to_hex() (Joe)</para></listitem>
17595 <listitem><para>Fix LIKE for bytea as a right-hand argument (Joe)</para></listitem>
17596 <listitem><para>Prevent crashes caused by SELECT cash_out(2) (Tom)</para></listitem>
17597 <listitem><para>Fix to_char(1,'FM999.99') to return a period (Karel)</para></listitem>
17598 <listitem><para>Fix trigger/type/language functions returning OPAQUE to return proper type (Tom)</para></listitem>
17603 <title>Internationalization</title>
17605 <listitem><para>Add additional encodings: Korean (JOHAB), Thai (WIN874), Vietnamese (TCVN), Arabic (WIN1256), Simplified Chinese (GBK), Korean (UHC) (Eiji Tokuya)</para></listitem>
17606 <listitem><para>Enable locale support by default (Peter)</para></listitem>
17607 <listitem><para>Add locale variables (Peter)</para></listitem>
17608 <listitem><para>Escape byes >= 0x7f for multibyte in PQescapeBytea/PQunescapeBytea (Tatsuo)</para></listitem>
17609 <listitem><para>Add locale awareness to regular expression character classes</para></listitem>
17610 <listitem><para>Enable multibyte support by default (Tatsuo)</para></listitem>
17611 <listitem><para>Add GB18030 multibyte support (Bill Huang)</para></listitem>
17612 <listitem><para>Add CREATE/DROP CONVERSION, allowing loadable encodings (Tatsuo, Kaori)</para></listitem>
17613 <listitem><para>Add pg_conversion table (Tatsuo)</para></listitem>
17614 <listitem><para>Add SQL99 CONVERT() function (Tatsuo)</para></listitem>
17615 <listitem><para>pg_dumpall, pg_controldata, and pg_resetxlog now national-language aware (Peter)</para></listitem>
17616 <listitem><para>New and updated translations</para></listitem>
17621 <title>Server-side Languages</title>
17623 <listitem><para>Allow recursive SQL function (Peter)</para></listitem>
17624 <listitem><para>Change PL/Tcl build to use configured compiler and Makefile.shlib (Peter)</para></listitem>
17625 <listitem><para>Overhaul the PL/pgSQL FOUND variable to be more Oracle-compatible (Neil, Tom)</para></listitem>
17626 <listitem><para>Allow PL/pgSQL to handle quoted identifiers (Tom)</para></listitem>
17627 <listitem><para>Allow set-returning PL/pgSQL functions (Neil)</para></listitem>
17628 <listitem><para>Make PL/pgSQL schema-aware (Joe)</para></listitem>
17629 <listitem><para>Remove some memory leaks (Nigel J. Andrews, Tom)</para></listitem>
17634 <title>psql</title>
17636 <listitem><para>Don't lowercase psql \connect database name for 7.2.0 compatibility (Tom)</para></listitem>
17637 <listitem><para>Add psql \timing to time user queries (Greg Sabino Mullane)</para></listitem>
17638 <listitem><para>Have psql \d show index information (Greg Sabino Mullane)</para></listitem>
17639 <listitem><para>New psql \dD shows domains (Jonathan Eisler)</para></listitem>
17640 <listitem><para>Allow psql to show rules on views (Paul ?)</para></listitem>
17641 <listitem><para>Fix for psql variable substitution (Tom)</para></listitem>
17642 <listitem><para>Allow psql \d to show temporary table structure (Tom)</para></listitem>
17643 <listitem><para>Allow psql \d to show foreign keys (Rod)</para></listitem>
17644 <listitem><para>Fix \? to honor \pset pager (Bruce)</para></listitem>
17645 <listitem><para>Have psql reports its version number on startup (Tom)</para></listitem>
17646 <listitem><para>Allow \copy to specify column names (Tom)</para></listitem>
17651 <title>libpq</title>
17653 <listitem><para>Add ~/.pgpass to store host/user password combinations (Alvaro Herrera)</para></listitem>
17654 <listitem><para>Add PQunescapeBytea() function to libpq (Patrick Welche)</para></listitem>
17655 <listitem><para>Fix for sending large queries over non-blocking connections (Bernhard Herzog)</para></listitem>
17656 <listitem><para>Fix for libpq using timers on Win9X (David Ford)</para></listitem>
17657 <listitem><para>Allow libpq notify to handle servers with different-length identifiers (Tom)</para></listitem>
17658 <listitem><para>Add libpq PQescapeString() and PQescapeBytea() to Windows (Bruce)</para></listitem>
17659 <listitem><para>Fix for SSL with non-blocking connections (Jack Bates)</para></listitem>
17660 <listitem><para>Add libpq connection timeout parameter (Denis A Ustimenko)</para></listitem>
17665 <title>JDBC</title>
17667 <listitem><para>Allow JDBC to compile with JDK 1.4 (Dave)</para></listitem>
17668 <listitem><para>Add JDBC 3 support (Barry)</para></listitem>
17669 <listitem><para>Allows JDBC to set loglevel by adding ?loglevel=X to the connection URL (Barry)</para></listitem>
17670 <listitem><para>Add Driver.info() message that prints out the version number (Barry)</para></listitem>
17671 <listitem><para>Add updateable result sets (Raghu Nidagal, Dave)</para></listitem>
17672 <listitem><para>Add support for callable statements (Paul Bethe)</para></listitem>
17673 <listitem><para>Add query cancel capability</para></listitem>
17674 <listitem><para>Add refresh row (Dave)</para></listitem>
17675 <listitem><para>Fix MD5 encryption handling for multibyte servers (Jun Kawai)</para></listitem>
17676 <listitem><para>Add support for prepared statements (Barry)</para></listitem>
17681 <title>Miscellaneous Interfaces</title>
17683 <listitem><para>Fixed ECPG bug concerning octal numbers in single quotes (Michael)</para></listitem>
17684 <listitem><para>Move src/interfaces/libpgeasy to http://gborg.postgresql.org (Marc, Bruce)</para></listitem>
17685 <listitem><para>Improve Python interface (Elliot Lee, Andrew Johnson, Greg Copeland)</para></listitem>
17686 <listitem><para>Add libpgtcl connection close event (Gerhard Hintermayer)</para></listitem>
17687 <listitem><para>Move src/interfaces/libpq++ to http://gborg.postgresql.org (Marc, Bruce)</para></listitem>
17688 <listitem><para>Move src/interfaces/odbc to http://gborg.postgresql.org (Marc)</para></listitem>
17689 <listitem><para>Move src/interfaces/libpgeasy to http://gborg.postgresql.org (Marc, Bruce)</para></listitem>
17690 <listitem><para>Move src/interfaces/perl5 to http://gborg.postgresql.org (Marc, Bruce)</para></listitem>
17691 <listitem><para>Remove src/bin/pgaccess from main tree, now at http://www.pgaccess.org (Bruce)</para></listitem>
17692 <listitem><para>Add pg_on_connection_loss command to libpgtcl (Gerhard Hintermayer, Tom)</para></listitem>
17697 <title>Source Code</title>
17699 <listitem><para>Fix for parallel make (Peter)</para></listitem>
17700 <listitem><para>AIX fixes for linking Tcl (Andreas Zeugswetter)</para></listitem>
17701 <listitem><para>Allow PL/Perl to build under Cygwin (Jason Tishler)</para></listitem>
17702 <listitem><para>Improve MIPS compiles (Peter, Oliver Elphick)</para></listitem>
17703 <listitem><para>Require Autoconf version 2.53 (Peter)</para></listitem>
17704 <listitem><para>Require readline and zlib by default in configure (Peter)</para></listitem>
17705 <listitem><para>Allow Solaris to use Intimate Shared Memory (ISM), for performance (Scott Brunza, P.J. Josh Rovero)</para></listitem>
17706 <listitem><para>Always enable syslog in compile, remove --enable-syslog option (Tatsuo)</para></listitem>
17707 <listitem><para>Always enable multibyte in compile, remove --enable-multibyte option (Tatsuo)</para></listitem>
17708 <listitem><para>Always enable locale in compile, remove --enable-locale option (Peter)</para></listitem>
17709 <listitem><para>Fix for Win9x DLL creation (Magnus Naeslund)</para></listitem>
17710 <listitem><para>Fix for link() usage by WAL code on Windows, BeOS (Jason Tishler)</para></listitem>
17711 <listitem><para>Add sys/types.h to c.h, remove from main files (Peter, Bruce)</para></listitem>
17712 <listitem><para>Fix AIX hang on SMP machines (Tomoyuki Niijima)</para></listitem>
17713 <listitem><para>AIX SMP hang fix (Tomoyuki Niijima)</para></listitem>
17714 <listitem><para>Fix pre-1970 date handling on newer glibc libraries (Tom)</para></listitem>
17715 <listitem><para>Fix PowerPC SMP locking (Tom)</para></listitem>
17716 <listitem><para>Prevent gcc -ffast-math from being used (Peter, Tom)</para></listitem>
17717 <listitem><para>Bison >= 1.50 now required for developer builds</para></listitem>
17718 <listitem><para>Kerberos 5 support now builds with Heimdal (Peter)</para></listitem>
17719 <listitem><para>Add appendix in the User's Guide which lists SQL features (Thomas)</para></listitem>
17720 <listitem><para>Improve loadable module linking to use RTLD_NOW (Tom)</para></listitem>
17721 <listitem><para>New error levels WARNING, INFO, LOG, DEBUG[1-5] (Bruce)</para></listitem>
17722 <listitem><para>New src/port directory holds replaced libc functions (Peter, Bruce)</para></listitem>
17723 <listitem><para>New pg_namespace system catalog for schemas (Tom)</para></listitem>
17724 <listitem><para>Add pg_class.relnamespace for schemas (Tom)</para></listitem>
17725 <listitem><para>Add pg_type.typnamespace for schemas (Tom)</para></listitem>
17726 <listitem><para>Add pg_proc.pronamespace for schemas (Tom)</para></listitem>
17727 <listitem><para>Restructure aggregates to have pg_proc entries (Tom)</para></listitem>
17728 <listitem><para>System relations now have their own namespace, pg_* test not required (Fernando Nasser)</para></listitem>
17729 <listitem><para>Rename TOAST index names to be *_index rather than *_idx (Neil)</para></listitem>
17730 <listitem><para>Add namespaces for operators, opclasses (Tom)</para></listitem>
17731 <listitem><para>Add additional checks to server control file (Thomas)</para></listitem>
17732 <listitem><para>New Polish FAQ (Marcin Mazurek)</para></listitem>
17733 <listitem><para>Add Posix semaphore support (Tom)</para></listitem>
17734 <listitem><para>Document need for reindex (Bruce)</para></listitem>
17735 <listitem><para>Rename some internal identifiers to simplify Windows compile (Jan, Katherine Ward)</para></listitem>
17736 <listitem><para>Add documentation on computing disk space (Bruce)</para></listitem>
17737 <listitem><para>Remove KSQO from GUC (Bruce)</para></listitem>
17738 <listitem><para>Fix memory leak in rtree (Kenneth Been)</para></listitem>
17739 <listitem><para>Modify a few error messages for consistency (Bruce)</para></listitem>
17740 <listitem><para>Remove unused system table columns (Peter)</para></listitem>
17741 <listitem><para>Make system columns NOT NULL where appropriate (Tom)</para></listitem>
17742 <listitem><para>Clean up use of sprintf in favor of snprintf() (Neil, Jukka Holappa)</para></listitem>
17743 <listitem><para>Remove OPAQUE and create specific subtypes (Tom)</para></listitem>
17744 <listitem><para>Cleanups in array internal handling (Joe, Tom)</para></listitem>
17745 <listitem><para>Disallow pg_atoi('') (Bruce)</para></listitem>
17746 <listitem><para>Remove parameter wal_files because WAL files are now recycled (Bruce)</para></listitem>
17747 <listitem><para>Add version numbers to heap pages (Tom)</para></listitem>
17752 <title>Contrib</title>
17754 <listitem><para>Allow inet arrays in /contrib/array (Neil)</para></listitem>
17755 <listitem><para>GiST fixes (Teodor Sigaev, Neil)</para></listitem>
17756 <listitem><para>Upgrade /contrib/mysql</para></listitem>
17757 <listitem><para>Add /contrib/dbsize which shows table sizes without vacuum (Peter)</para></listitem>
17758 <listitem><para>Add /contrib/intagg, integer aggregator routines (mlw)</para></listitem>
17759 <listitem><para>Improve /contrib/oid2name (Neil, Bruce)</para></listitem>
17760 <listitem><para>Improve /contrib/tsearch (Oleg, Teodor Sigaev)</para></listitem>
17761 <listitem><para>Cleanups of /contrib/rserver (Alexey V. Borzov)</para></listitem>
17762 <listitem><para>Update /contrib/oracle conversion utility (Gilles Darold)</para></listitem>
17763 <listitem><para>Update /contrib/dblink (Joe)</para></listitem>
17764 <listitem><para>Improve options supported by /contrib/vacuumlo (Mario Weilguni)</para></listitem>
17765 <listitem><para>Improvements to /contrib/intarray (Oleg, Teodor Sigaev, Andrey Oktyabrski)</para></listitem>
17766 <listitem><para>Add /contrib/reindexdb utility (Shaun Thomas)</para></listitem>
17767 <listitem><para>Add indexing to /contrib/isbn_issn (Dan Weston)</para></listitem>
17768 <listitem><para>Add /contrib/dbmirror (Steven Singer)</para></listitem>
17769 <listitem><para>Improve /contrib/pgbench (Neil)</para></listitem>
17770 <listitem><para>Add /contrib/tablefunc table function examples (Joe)</para></listitem>
17771 <listitem><para>Add /contrib/ltree data type for tree structures (Teodor Sigaev, Oleg Bartunov)</para></listitem>
17772 <listitem><para>Move /contrib/pg_controldata, pg_resetxlog into main tree (Bruce)</para></listitem>
17773 <listitem><para>Fixes to /contrib/cube (Bruno Wolff)</para></listitem>
17774 <listitem><para>Improve /contrib/fulltextindex (Christopher)</para></listitem>
17781 <sect1 id="release-7-2-8">
17782 <title>Release 7.2.8</title>
17785 <title>Release date</title>
17786 <simpara>2005-05-09</simpara>
17790 This release contains a variety of fixes from 7.2.7, including one
17791 security-related issue.
17795 <title>Migration to Version 7.2.8</title>
17798 A dump/restore is not required for those running 7.2.X.
17803 <title>Changes</title>
17806 <listitem><para>Repair ancient race condition that allowed a transaction to be
17807 seen as committed for some purposes (eg SELECT FOR UPDATE) slightly sooner
17808 than for other purposes</para>
17809 <para>This is an extremely serious bug since it could lead to apparent
17810 data inconsistencies being briefly visible to applications.</para></listitem>
17811 <listitem><para>Repair race condition between relation extension and
17813 <para>This could theoretically have caused loss of a page's worth of
17814 freshly-inserted data, although the scenario seems of very low probability.
17815 There are no known cases of it having caused more than an Assert failure.
17817 <listitem><para>Fix <function>EXTRACT(EPOCH)</> for
17818 <type>TIME WITH TIME ZONE</> values</para></listitem>
17819 <listitem><para>Additional buffer overrun checks in plpgsql
17820 (Neil)</para></listitem>
17821 <listitem><para>Fix pg_dump to dump index names and trigger names containing
17822 <literal>%</> correctly (Neil)</para></listitem>
17823 <listitem><para>Prevent <function>to_char(interval)</> from dumping core for
17824 month-related formats</para></listitem>
17825 <listitem><para>Fix <filename>contrib/pgcrypto</> for newer OpenSSL builds
17826 (Marko Kreen)</para></listitem>
17832 <sect1 id="release-7-2-7">
17833 <title>Release 7.2.7</title>
17836 <title>Release date</title>
17837 <simpara>2005-01-31</simpara>
17841 This release contains a variety of fixes from 7.2.6, including several
17842 security-related issues.
17846 <title>Migration to Version 7.2.7</title>
17849 A dump/restore is not required for those running 7.2.X.
17854 <title>Changes</title>
17857 <listitem><para>Disallow <command>LOAD</> to non-superusers</para>
17859 On platforms that will automatically execute initialization functions of a
17860 shared library (this includes at least Windows and ELF-based Unixen),
17861 <command>LOAD</> can be used to make the server execute arbitrary code.
17862 Thanks to NGS Software for reporting this.</para></listitem>
17863 <listitem><para>Add needed STRICT marking to some contrib functions (Kris
17864 Jurka)</para></listitem>
17865 <listitem><para>Avoid buffer overrun when plpgsql cursor declaration has too
17866 many parameters (Neil)</para></listitem>
17867 <listitem><para>Fix planning error for FULL and RIGHT outer joins</para>
17869 The result of the join was mistakenly supposed to be sorted the same as the
17870 left input. This could not only deliver mis-sorted output to the user, but
17871 in case of nested merge joins could give outright wrong answers.
17873 <listitem><para>Fix display of negative intervals in SQL and GERMAN
17874 datestyles</para></listitem>
17880 <sect1 id="release-7-2-6">
17881 <title>Release 7.2.6</title>
17884 <title>Release date</title>
17885 <simpara>2004-10-22</simpara>
17889 This release contains a variety of fixes from 7.2.5.
17894 <title>Migration to Version 7.2.6</title>
17897 A dump/restore is not required for those running 7.2.X.
17902 <title>Changes</title>
17905 <listitem><para>Repair possible failure to update hint bits on disk</para>
17907 Under rare circumstances this oversight could lead to
17908 <quote>could not access transaction status</> failures, which qualifies
17909 it as a potential-data-loss bug.
17911 <listitem><para>Ensure that hashed outer join does not miss tuples</para>
17913 Very large left joins using a hash join plan could fail to output unmatched
17914 left-side rows given just the right data distribution.
17916 <listitem><para>Disallow running pg_ctl as root</para>
17918 This is to guard against any possible security issues.
17920 <listitem><para>Avoid using temp files in /tmp in make_oidjoins_check</para>
17922 This has been reported as a security issue, though it's hardly worthy of
17923 concern since there is no reason for non-developers to use this script anyway.
17925 <listitem><para>Update to newer versions of Bison</para></listitem>
17931 <sect1 id="release-7-2-5">
17932 <title>Release 7.2.5</title>
17935 <title>Release date</title>
17936 <simpara>2004-08-16</simpara>
17940 This release contains a variety of fixes from 7.2.4.
17945 <title>Migration to Version 7.2.5</title>
17948 A dump/restore is not required for those running 7.2.X.
17953 <title>Changes</title>
17956 <listitem><para>Prevent possible loss of committed transactions during crash</para>
17958 Due to insufficient interlocking between transaction commit and checkpointing,
17959 it was possible for transactions committed just before the most recent
17960 checkpoint to be lost, in whole or in part, following a database crash and
17961 restart. This is a serious bug that has existed
17962 since <productname>PostgreSQL</productname> 7.1.
17964 <listitem><para>Fix corner case for btree search in parallel with first root page split</para></listitem>
17965 <listitem><para>Fix buffer overrun in <function>to_ascii</function> (Guido Notari)</para></listitem>
17966 <listitem><para>Fix core dump in deadlock detection on machines where char is unsigned</para></listitem>
17967 <listitem><para>Fix failure to respond to <command>pg_ctl stop -m fast</command> after Async_NotifyHandler runs</para></listitem>
17968 <listitem><para>Repair memory leaks in pg_dump</para></listitem>
17969 <listitem><para>Avoid conflict with system definition of <function>isblank()</function> function or macro</para></listitem>
17974 <sect1 id="release-7-2-4">
17975 <title>Release 7.2.4</title>
17978 <title>Release date</title>
17979 <simpara>2003-01-30</simpara>
17983 This release contains a variety of fixes for version 7.2.3,
17984 including fixes to prevent possible data loss.
17988 <title>Migration to Version 7.2.4</title>
17991 A dump/restore is <emphasis>not</emphasis> required for those
17992 running version 7.2.*.
17997 <title>Changes</title>
18000 <listitem><para>Fix some additional cases of VACUUM "No one parent tuple was found" error</para></listitem>
18001 <listitem><para>Prevent VACUUM from being called inside a function (Bruce)</para></listitem>
18002 <listitem><para>Ensure pg_clog updates are sync'd to disk before marking checkpoint complete</para></listitem>
18003 <listitem><para>Avoid integer overflow during large hash joins</para></listitem>
18004 <listitem><para>Make GROUP commands work when pg_group.grolist is large enough to be toasted</para></listitem>
18005 <listitem><para>Fix errors in datetime tables; some timezone names weren't being recognized</para></listitem>
18006 <listitem><para>Fix integer overflows in circle_poly(), path_encode(), path_add() (Neil)</para></listitem>
18007 <listitem><para>Repair long-standing logic errors in lseg_eq(), lseg_ne(), lseg_center()</para></listitem>
18013 <sect1 id="release-7-2-3">
18014 <title>Release 7.2.3</title>
18017 <title>Release date</title>
18018 <simpara>2002-10-01</simpara>
18022 This release contains a variety of fixes for version 7.2.2,
18023 including fixes to prevent possible data loss.
18027 <title>Migration to Version 7.2.3</title>
18030 A dump/restore is <emphasis>not</emphasis> required for those
18031 running version 7.2.*.
18036 <title>Changes</title>
18039 <listitem><para>Prevent possible compressed transaction log loss (Tom)</para></listitem>
18040 <listitem><para>Prevent non-superuser from increasing most recent vacuum info (Tom)</para></listitem>
18041 <listitem><para>Handle pre-1970 date values in newer versions of glibc (Tom)</para></listitem>
18042 <listitem><para>Fix possible hang during server shutdown</para></listitem>
18043 <listitem><para>Prevent spinlock hangs on SMP PPC machines (Tomoyuki Niijima)</para></listitem>
18044 <listitem><para>Fix <application>pg_dump</> to properly dump FULL JOIN USING (Tom)</para></listitem>
18050 <sect1 id="release-7-2-2">
18051 <title>Release 7.2.2</title>
18054 <title>Release date</title>
18055 <simpara>2002-08-23</simpara>
18059 This release contains a variety of fixes for version 7.2.1.
18063 <title>Migration to Version 7.2.2</title>
18066 A dump/restore is <emphasis>not</emphasis> required for those
18067 running version 7.2.*.
18072 <title>Changes</title>
18075 <listitem><para>Allow EXECUTE of "CREATE TABLE AS ... SELECT" in PL/pgSQL (Tom)</para></listitem>
18076 <listitem><para>Fix for compressed transaction log id wraparound (Tom)</para></listitem>
18077 <listitem><para>Fix PQescapeBytea/PQunescapeBytea so that they handle bytes > 0x7f (Tatsuo)</para></listitem>
18078 <listitem><para>Fix for psql and <application>pg_dump</> crashing when invoked with non-existent long options (Tatsuo)</para></listitem>
18079 <listitem><para>Fix crash when invoking geometric operators (Tom)</para></listitem>
18080 <listitem><para>Allow OPEN cursor(args) (Tom)</para></listitem>
18081 <listitem><para>Fix for rtree_gist index build (Teodor)</para></listitem>
18082 <listitem><para>Fix for dumping user-defined aggregates (Tom)</para></listitem>
18083 <listitem><para>contrib/intarray fixes (Oleg)</para></listitem>
18084 <listitem><para>Fix for complex UNION/EXCEPT/INTERSECT queries using parens (Tom)</para></listitem>
18085 <listitem><para>Fix to pg_convert (Tatsuo)</para></listitem>
18086 <listitem><para>Fix for crash with long DATA strings (Thomas, Neil)</para></listitem>
18087 <listitem><para>Fix for repeat(), lpad(), rpad() and long strings (Neil)</para></listitem>
18093 <sect1 id="release-7-2-1">
18094 <title>Release 7.2.1</title>
18097 <title>Release date</title>
18098 <simpara>2002-03-21</simpara>
18102 This release contains a variety of fixes for version 7.2.
18106 <title>Migration to Version 7.2.1</title>
18109 A dump/restore is <emphasis>not</emphasis> required for those
18110 running version 7.2.
18115 <title>Changes</title>
18118 <listitem><para>Ensure that sequence counters do not go backwards after a crash (Tom)</para></listitem>
18119 <listitem><para>Fix pgaccess kanji-conversion key binding (Tatsuo)</para></listitem>
18120 <listitem><para>Optimizer improvements (Tom)</para></listitem>
18121 <listitem><para>Cash I/O improvements (Tom)</para></listitem>
18122 <listitem><para>New Russian FAQ</para></listitem>
18123 <listitem><para>Compile fix for missing AuthBlockSig (Heiko)</para></listitem>
18124 <listitem><para>Additional time zones and time zone fixes (Thomas)</para></listitem>
18125 <listitem><para>Allow psql \connect to handle mixed case database and user names (Tom)</para></listitem>
18126 <listitem><para>Return proper OID on command completion even with ON INSERT rules (Tom)</para></listitem>
18127 <listitem><para>Allow COPY FROM to use 8-bit DELIMITERS (Tatsuo)</para></listitem>
18128 <listitem><para>Fix bug in extract/date_part for milliseconds/microseconds (Tatsuo)</para></listitem>
18129 <listitem><para>Improve handling of multiple UNIONs with different lengths (Tom)</para></listitem>
18130 <listitem><para>contrib/btree_gist improvements (Teodor Sigaev)</para></listitem>
18131 <listitem><para>contrib/tsearch dictionary improvements, see README.tsearch for an additional installation step (Thomas T. Thai, Teodor Sigaev)</para></listitem>
18132 <listitem><para>Fix for array subscripts handling (Tom)</para></listitem>
18133 <listitem><para>Allow EXECUTE of "CREATE TABLE AS ... SELECT" in PL/pgSQL (Tom)</para></listitem>
18139 <sect1 id="release-7-2">
18140 <title>Release 7.2</title>
18143 <title>Release date</title>
18144 <simpara>2002-02-04</simpara>
18148 <title>Overview</title>
18151 This release improves <productname>PostgreSQL</> for use in
18152 high-volume applications.
18156 Major changes in this release:
18161 <term>VACUUM</term>
18164 Vacuuming no longer locks tables, thus allowing normal user
18165 access during the vacuum. A new <command>VACUUM FULL</>
18166 command does old-style vacuum by locking the table and
18167 shrinking the on-disk copy of the table.
18173 <term>Transactions</term>
18176 There is no longer a problem with installations that exceed
18177 four billion transactions.
18186 OIDs are now optional. Users can now create tables without
18187 OIDs for cases where OID usage is excessive.
18193 <term>Optimizer</term>
18196 The system now computes histogram column statistics during
18197 <command>ANALYZE</>, allowing much better optimizer choices.
18203 <term>Security</term>
18206 A new MD5 encryption option allows more secure storage and
18207 transfer of passwords. A new Unix-domain socket
18208 authentication option is available on Linux and BSD systems.
18214 <term>Statistics</term>
18217 Administrators can use the new table access statistics module
18218 to get fine-grained information about table and index usage.
18224 <term>Internationalization</term>
18227 Program and library messages can now be displayed in several
18237 <title>Migration to Version 7.2</title>
18240 A dump/restore using <command>pg_dump</command> is required for
18241 those wishing to migrate data from any previous release.
18245 Observe the following incompatibilities:
18251 The semantics of the <command>VACUUM</command> command have
18252 changed in this release. You might wish to update your
18253 maintenance procedures accordingly.
18259 In this release, comparisons using <literal>= NULL</literal>
18260 will always return false (or NULL, more precisely). Previous
18261 releases automatically transformed this syntax to <literal>IS
18262 NULL</literal>. The old behavior can be re-enabled using a
18263 <filename>postgresql.conf</filename> parameter.
18269 The <filename>pg_hba.conf</> and <filename>pg_ident.conf</>
18270 configuration is now only reloaded after receiving a
18271 <systemitem>SIGHUP</> signal, not with each connection.
18277 The function <filename>octet_length()</> now returns the uncompressed data length.
18283 The date/time value <literal>'current'</literal> is no longer
18284 available. You will need to rewrite your applications.
18290 The <literal>timestamp()</literal>, <literal>time()</literal>,
18291 and <literal>interval()</literal> functions are no longer
18292 available. Instead of <literal>timestamp()</literal>, use
18293 <literal>timestamp 'string'</literal> or <literal>CAST</literal>.
18300 The <literal>SELECT ... LIMIT #,#</literal> syntax will be removed
18301 in the next release. You should change your queries to use
18302 separate LIMIT and OFFSET clauses, e.g. <literal>LIMIT 10 OFFSET
18308 <title>Changes</title>
18311 <title>Server Operation</title>
18313 <listitem><para>Create temporary files in a separate directory (Bruce)</para></listitem>
18314 <listitem><para>Delete orphaned temporary files on postmaster startup (Bruce)</para></listitem>
18315 <listitem><para>Added unique indexes to some system tables (Tom)</para></listitem>
18316 <listitem><para>System table operator reorganization (Oleg Bartunov, Teodor Sigaev, Tom)</para></listitem>
18317 <listitem><para>Renamed pg_log to pg_clog (Tom)</para></listitem>
18318 <listitem><para>Enable SIGTERM, SIGQUIT to kill backends (Jan)</para></listitem>
18319 <listitem><para>Removed compile-time limit on number of backends (Tom)</para></listitem>
18320 <listitem><para>Better cleanup for semaphore resource failure (Tatsuo, Tom)</para></listitem>
18321 <listitem><para>Allow safe transaction ID wraparound (Tom)</para></listitem>
18322 <listitem><para>Removed OIDs from some system tables (Tom)</para></listitem>
18323 <listitem><para>Removed "triggered data change violation" error check (Tom)</para></listitem>
18324 <listitem><para>SPI portal creation of prepared/saved plans (Jan)</para></listitem>
18325 <listitem><para>Allow SPI column functions to work for system columns (Tom)</para></listitem>
18326 <listitem><para>Long value compression improvement (Tom)</para></listitem>
18327 <listitem><para>Statistics collector for table, index access (Jan)</para></listitem>
18328 <listitem><para>Truncate extra-long sequence names to a reasonable value (Tom)</para></listitem>
18329 <listitem><para>Measure transaction times in milliseconds (Thomas)</para></listitem>
18330 <listitem><para>Fix TID sequential scans (Hiroshi)</para></listitem>
18331 <listitem><para>Superuser ID now fixed at 1 (Peter E)</para></listitem>
18332 <listitem><para>New pg_ctl "reload" option (Tom)</para></listitem>
18337 <title>Performance</title>
18339 <listitem><para>Optimizer improvements (Tom)</para></listitem>
18340 <listitem><para>New histogram column statistics for optimizer (Tom)</para></listitem>
18341 <listitem><para>Reuse write-ahead log files rather than discarding them (Tom)</para></listitem>
18342 <listitem><para>Cache improvements (Tom)</para></listitem>
18343 <listitem><para>IS NULL, IS NOT NULL optimizer improvement (Tom)</para></listitem>
18344 <listitem><para>Improve lock manager to reduce lock contention (Tom)</para></listitem>
18345 <listitem><para>Keep relcache entries for index access support functions (Tom)</para></listitem>
18346 <listitem><para>Allow better selectivity with NaN and infinities in NUMERIC (Tom)</para></listitem>
18347 <listitem><para>R-tree performance improvements (Kenneth Been)</para></listitem>
18348 <listitem><para>B-tree splits more efficient (Tom)</para></listitem>
18353 <title>Privileges</title>
18355 <listitem><para>Change UPDATE, DELETE privileges to be distinct (Peter E)</para></listitem>
18356 <listitem><para>New REFERENCES, TRIGGER privileges (Peter E)</para></listitem>
18357 <listitem><para>Allow GRANT/REVOKE to/from more than one user at a time (Peter E)</para></listitem>
18358 <listitem><para>New has_table_privilege() function (Joe Conway)</para></listitem>
18359 <listitem><para>Allow non-superuser to vacuum database (Tom)</para></listitem>
18360 <listitem><para>New SET SESSION AUTHORIZATION command (Peter E)</para></listitem>
18361 <listitem><para>Fix bug in privilege modifications on newly created tables (Tom)</para></listitem>
18362 <listitem><para>Disallow access to pg_statistic for non-superuser, add user-accessible views (Tom)</para></listitem>
18367 <title>Client Authentication</title>
18369 <listitem><para>Fork postmaster before doing authentication to prevent hangs (Peter E)</para></listitem>
18370 <listitem><para>Add ident authentication over Unix domain sockets on Linux, *BSD (Helge Bahmann, Oliver Elphick, Teodor Sigaev, Bruce)</para></listitem>
18371 <listitem><para>Add a password authentication method that uses MD5 encryption (Bruce)</para></listitem>
18372 <listitem><para>Allow encryption of stored passwords using MD5 (Bruce)</para></listitem>
18373 <listitem><para>PAM authentication (Dominic J. Eidson)</para></listitem>
18374 <listitem><para>Load pg_hba.conf and pg_ident.conf only on startup and SIGHUP (Bruce)</para></listitem>
18379 <title>Server Configuration</title>
18381 <listitem><para>Interpretation of some time zone abbreviations as Australian rather than North American now settable at run time (Bruce)</para></listitem>
18382 <listitem><para>New parameter to set default transaction isolation level (Peter E)</para></listitem>
18383 <listitem><para>New parameter to enable conversion of "expr = NULL" into "expr IS NULL", off by default (Peter E)</para></listitem>
18384 <listitem><para>New parameter to control memory usage by VACUUM (Tom)</para></listitem>
18385 <listitem><para>New parameter to set client authentication timeout (Tom)</para></listitem>
18386 <listitem><para>New parameter to set maximum number of open files (Tom)</para></listitem>
18391 <title>Queries</title>
18393 <listitem><para>Statements added by INSERT rules now execute after the INSERT (Jan)</para></listitem>
18394 <listitem><para>Prevent unadorned relation names in target list (Bruce)</para></listitem>
18395 <listitem><para>NULLs now sort after all normal values in ORDER BY (Tom)</para></listitem>
18396 <listitem><para>New IS UNKNOWN, IS NOT UNKNOWN Boolean tests (Tom)</para></listitem>
18397 <listitem><para>New SHARE UPDATE EXCLUSIVE lock mode (Tom)</para></listitem>
18398 <listitem><para>New EXPLAIN ANALYZE command that shows run times and row counts (Martijn van Oosterhout)</para></listitem>
18399 <listitem><para>Fix problem with LIMIT and subqueries (Tom)</para></listitem>
18400 <listitem><para>Fix for LIMIT, DISTINCT ON pushed into subqueries (Tom)</para></listitem>
18401 <listitem><para>Fix nested EXCEPT/INTERSECT (Tom)</para></listitem>
18406 <title>Schema Manipulation</title>
18408 <listitem><para>Fix SERIAL in temporary tables (Bruce)</para></listitem>
18409 <listitem><para>Allow temporary sequences (Bruce)</para></listitem>
18410 <listitem><para>Sequences now use int8 internally (Tom)</para></listitem>
18411 <listitem><para>New SERIAL8 creates int8 columns with sequences, default still SERIAL4 (Tom)</para></listitem>
18412 <listitem><para>Make OIDs optional using WITHOUT OIDS (Tom)</para></listitem>
18413 <listitem><para>Add %TYPE syntax to CREATE TYPE (Ian Lance Taylor)</para></listitem>
18414 <listitem><para>Add ALTER TABLE / DROP CONSTRAINT for CHECK constraints (Christopher Kings-Lynne)</para></listitem>
18415 <listitem><para>New CREATE OR REPLACE FUNCTION to alter existing function (preserving the function OID) (Gavin Sherry)</para></listitem>
18416 <listitem><para>Add ALTER TABLE / ADD [ UNIQUE | PRIMARY ] (Christopher Kings-Lynne)</para></listitem>
18417 <listitem><para>Allow column renaming in views</para></listitem>
18418 <listitem><para>Make ALTER TABLE / RENAME COLUMN update column names of indexes (Brent Verner)</para></listitem>
18419 <listitem><para>Fix for ALTER TABLE / ADD CONSTRAINT ... CHECK with inherited tables (Stephan Szabo)</para></listitem>
18420 <listitem><para>ALTER TABLE RENAME update foreign-key trigger arguments correctly (Brent Verner)</para></listitem>
18421 <listitem><para>DROP AGGREGATE and COMMENT ON AGGREGATE now accept an aggtype (Tom)</para></listitem>
18422 <listitem><para>Add automatic return type data casting for SQL functions (Tom)</para></listitem>
18423 <listitem><para>Allow GiST indexes to handle NULLs and multikey indexes (Oleg Bartunov, Teodor Sigaev, Tom)</para></listitem>
18424 <listitem><para>Enable partial indexes (Martijn van Oosterhout)</para></listitem>
18429 <title>Utility Commands</title>
18431 <listitem><para>Add RESET ALL, SHOW ALL (Marko Kreen)</para></listitem>
18432 <listitem><para>CREATE/ALTER USER/GROUP now allow options in any order (Vince)</para></listitem>
18433 <listitem><para>Add LOCK A, B, C functionality (Neil Padgett)</para></listitem>
18434 <listitem><para>New ENCRYPTED/UNENCRYPTED option to CREATE/ALTER USER (Bruce)</para></listitem>
18435 <listitem><para>New light-weight VACUUM does not lock table; old semantics are available as VACUUM FULL (Tom)</para></listitem>
18436 <listitem><para>Disable COPY TO/FROM on views (Bruce)</para></listitem>
18437 <listitem><para>COPY DELIMITERS string must be exactly one character (Tom)</para></listitem>
18438 <listitem><para>VACUUM warning about index tuples fewer than heap now only appears when appropriate (Martijn van Oosterhout)</para></listitem>
18439 <listitem><para>Fix privilege checks for CREATE INDEX (Tom)</para></listitem>
18440 <listitem><para>Disallow inappropriate use of CREATE/DROP INDEX/TRIGGER/VIEW (Tom)</para></listitem>
18445 <title>Data Types and Functions</title>
18447 <listitem><para>SUM(), AVG(), COUNT() now uses int8 internally for speed (Tom)</para></listitem>
18448 <listitem><para>Add convert(), convert2() (Tatsuo)</para></listitem>
18449 <listitem><para>New function bit_length() (Peter E)</para></listitem>
18450 <listitem><para>Make the "n" in CHAR(n)/VARCHAR(n) represents letters, not bytes (Tatsuo)</para></listitem>
18451 <listitem><para>CHAR(), VARCHAR() now reject strings that are too long (Peter E)</para></listitem>
18452 <listitem><para>BIT VARYING now rejects bit strings that are too long (Peter E)</para></listitem>
18453 <listitem><para>BIT now rejects bit strings that do not match declared size (Peter E)</para></listitem>
18454 <listitem><para>INET, CIDR text conversion functions (Alex Pilosov)</para></listitem>
18455 <listitem><para>INET, CIDR operators << and <<= indexable (Alex Pilosov)</para></listitem>
18456 <listitem><para>Bytea \### now requires valid three digit octal number</para></listitem>
18457 <listitem><para>Bytea comparison improvements, now supports =, <>, >, >=, <, and <=</para></listitem>
18458 <listitem><para>Bytea now supports B-tree indexes</para></listitem>
18459 <listitem><para>Bytea now supports LIKE, LIKE...ESCAPE, NOT LIKE, NOT LIKE...ESCAPE</para></listitem>
18460 <listitem><para>Bytea now supports concatenation</para></listitem>
18461 <listitem><para>New bytea functions: position, substring, trim, btrim, and length</para></listitem>
18462 <listitem><para>New encode() function mode, "escaped", converts minimally escaped bytea to/from text</para></listitem>
18463 <listitem><para>Add pg_database_encoding_max_length() (Tatsuo)</para></listitem>
18464 <listitem><para>Add pg_client_encoding() function (Tatsuo)</para></listitem>
18465 <listitem><para>now() returns time with millisecond precision (Thomas)</para></listitem>
18466 <listitem><para>New TIMESTAMP WITHOUT TIMEZONE data type (Thomas)</para></listitem>
18467 <listitem><para>Add ISO date/time specification with "T", yyyy-mm-ddThh:mm:ss (Thomas)</para></listitem>
18468 <listitem><para>New xid/int comparison functions (Hiroshi)</para></listitem>
18469 <listitem><para>Add precision to TIME, TIMESTAMP, and INTERVAL data types (Thomas)</para></listitem>
18470 <listitem><para>Modify type coercion logic to attempt binary-compatible functions first (Tom)</para></listitem>
18471 <listitem><para>New encode() function installed by default (Marko Kreen)</para></listitem>
18472 <listitem><para>Improved to_*() conversion functions (Karel Zak)</para></listitem>
18473 <listitem><para>Optimize LIKE/ILIKE when using single-byte encodings (Tatsuo)</para></listitem>
18474 <listitem><para>New functions in contrib/pgcrypto: crypt(), hmac(), encrypt(), gen_salt() (Marko Kreen)</para></listitem>
18475 <listitem><para>Correct description of translate() function (Bruce)</para></listitem>
18476 <listitem><para>Add INTERVAL argument for SET TIME ZONE (Thomas)</para></listitem>
18477 <listitem><para>Add INTERVAL YEAR TO MONTH (etc.) syntax (Thomas)</para></listitem>
18478 <listitem><para>Optimize length functions when using single-byte encodings (Tatsuo)</para></listitem>
18479 <listitem><para>Fix path_inter, path_distance, path_length, dist_ppath to handle closed paths (Curtis Barrett, Tom)</para></listitem>
18480 <listitem><para>octet_length(text) now returns non-compressed length (Tatsuo, Bruce)</para></listitem>
18481 <listitem><para>Handle "July" full name in date/time literals (Greg Sabino Mullane)</para></listitem>
18482 <listitem><para>Some datatype() function calls now evaluated differently</para></listitem>
18483 <listitem><para>Add support for Julian and ISO time specifications (Thomas)</para></listitem>
18488 <title>Internationalization</title>
18490 <listitem><para>National language support in psql, <application>pg_dump</>, libpq, and server (Peter E)</para></listitem>
18491 <listitem><para>Message translations in Chinese (simplified, traditional), Czech, French, German, Hungarian, Russian, Swedish (Peter E, Serguei A. Mokhov, Karel Zak, Weiping He, Zhenbang Wei, Kovacs Zoltan)</para></listitem>
18492 <listitem><para>Make trim, ltrim, rtrim, btrim, lpad, rpad, translate multibyte aware (Tatsuo)</para></listitem>
18493 <listitem><para>Add LATIN5,6,7,8,9,10 support (Tatsuo)</para></listitem>
18494 <listitem><para>Add ISO 8859-5,6,7,8 support (Tatsuo)</para></listitem>
18495 <listitem><para>Correct LATIN5 to mean ISO-8859-9, not ISO-8859-5 (Tatsuo)</para></listitem>
18496 <listitem><para>Make mic2ascii() non-ASCII aware (Tatsuo)</para></listitem>
18497 <listitem><para>Reject invalid multibyte character sequences (Tatsuo)</para></listitem>
18502 <title><application>PL/pgSQL</></title>
18504 <listitem><para>Now uses portals for SELECT loops, allowing huge result sets (Jan)</para></listitem>
18505 <listitem><para>CURSOR and REFCURSOR support (Jan)</para></listitem>
18506 <listitem><para>Can now return open cursors (Jan)</para></listitem>
18507 <listitem><para>Add ELSEIF (Klaus Reger)</para></listitem>
18508 <listitem><para>Improve PL/pgSQL error reporting, including location of error (Tom)</para></listitem>
18509 <listitem><para>Allow IS or FOR key words in cursor declaration, for compatibility (Bruce)</para></listitem>
18510 <listitem><para>Fix for SELECT ... FOR UPDATE (Tom)</para></listitem>
18511 <listitem><para>Fix for PERFORM returning multiple rows (Tom)</para></listitem>
18512 <listitem><para>Make PL/pgSQL use the server's type coercion code (Tom)</para></listitem>
18513 <listitem><para>Memory leak fix (Jan, Tom)</para></listitem>
18514 <listitem><para>Make trailing semicolon optional (Tom)</para></listitem>
18519 <title>PL/Perl</title>
18521 <listitem><para>New untrusted PL/Perl (Alex Pilosov)</para></listitem>
18522 <listitem><para>PL/Perl is now built on some platforms even if libperl is not shared (Peter E)</para></listitem>
18527 <title>PL/Tcl</title>
18529 <listitem><para>Now reports errorInfo (Vsevolod Lobko)</para></listitem>
18530 <listitem><para>Add spi_lastoid function (bob@redivi.com)</para></listitem>
18535 <title>PL/Python</title>
18537 <listitem><para>...is new (Andrew Bosma)</para></listitem>
18542 <title><application>psql</></title>
18544 <listitem><para>\d displays indexes in unique, primary groupings (Christopher Kings-Lynne)</para></listitem>
18545 <listitem><para>Allow trailing semicolons in backslash commands (Greg Sabino Mullane)</para></listitem>
18546 <listitem><para>Read password from /dev/tty if possible</para></listitem>
18547 <listitem><para>Force new password prompt when changing user and database (Tatsuo, Tom)</para></listitem>
18548 <listitem><para>Format the correct number of columns for Unicode (Patrice)</para></listitem>
18553 <title><application>libpq</></title>
18555 <listitem><para>New function PQescapeString() to escape quotes in command strings (Florian Weimer)</para></listitem>
18556 <listitem><para>New function PQescapeBytea() escapes binary strings for use as SQL string literals</para></listitem>
18561 <title>JDBC</title>
18563 <listitem><para>Return OID of INSERT (Ken K)</para></listitem>
18564 <listitem><para>Handle more data types (Ken K)</para></listitem>
18565 <listitem><para>Handle single quotes and newlines in strings (Ken K)</para></listitem>
18566 <listitem><para>Handle NULL variables (Ken K)</para></listitem>
18567 <listitem><para>Fix for time zone handling (Barry Lind)</para></listitem>
18568 <listitem><para>Improved Druid support</para></listitem>
18569 <listitem><para>Allow eight-bit characters with non-multibyte server (Barry Lind)</para></listitem>
18570 <listitem><para>Support BIT, BINARY types (Ned Wolpert)</para></listitem>
18571 <listitem><para>Reduce memory usage (Michael Stephens, Dave Cramer)</para></listitem>
18572 <listitem><para>Update DatabaseMetaData (Peter E)</para></listitem>
18573 <listitem><para>Add DatabaseMetaData.getCatalogs() (Peter E)</para></listitem>
18574 <listitem><para>Encoding fixes (Anders Bengtsson)</para></listitem>
18575 <listitem><para>Get/setCatalog methods (Jason Davies)</para></listitem>
18576 <listitem><para>DatabaseMetaData.getColumns() now returns column defaults (Jason Davies)</para></listitem>
18577 <listitem><para>DatabaseMetaData.getColumns() performance improvement (Jeroen van Vianen)</para></listitem>
18578 <listitem><para>Some JDBC1 and JDBC2 merging (Anders Bengtsson)</para></listitem>
18579 <listitem><para>Transaction performance improvements (Barry Lind)</para></listitem>
18580 <listitem><para>Array fixes (Greg Zoller)</para></listitem>
18581 <listitem><para>Serialize addition </para></listitem>
18582 <listitem><para>Fix batch processing (Rene Pijlman)</para></listitem>
18583 <listitem><para>ExecSQL method reorganization (Anders Bengtsson)</para></listitem>
18584 <listitem><para>GetColumn() fixes (Jeroen van Vianen)</para></listitem>
18585 <listitem><para>Fix isWriteable() function (Rene Pijlman)</para></listitem>
18586 <listitem><para>Improved passage of JDBC2 conformance tests (Rene Pijlman)</para></listitem>
18587 <listitem><para>Add bytea type capability (Barry Lind)</para></listitem>
18588 <listitem><para>Add isNullable() (Rene Pijlman)</para></listitem>
18589 <listitem><para>JDBC date/time test suite fixes (Liam Stewart)</para></listitem>
18590 <listitem><para>Fix for SELECT 'id' AS xxx FROM table (Dave Cramer)</para></listitem>
18591 <listitem><para>Fix DatabaseMetaData to show precision properly (Mark Lillywhite)</para></listitem>
18592 <listitem><para>New getImported/getExported keys (Jason Davies)</para></listitem>
18593 <listitem><para>MD5 password encryption support (Jeremy Wohl)</para></listitem>
18594 <listitem><para>Fix to actually use type cache (Ned Wolpert)</para></listitem>
18599 <title>ODBC</title>
18601 <listitem><para>Remove query size limit (Hiroshi)</para></listitem>
18602 <listitem><para>Remove text field size limit (Hiroshi)</para></listitem>
18603 <listitem><para>Fix for SQLPrimaryKeys in multibyte mode (Hiroshi)</para></listitem>
18604 <listitem><para>Allow ODBC procedure calls (Hiroshi)</para></listitem>
18605 <listitem><para>Improve boolean handing (Aidan Mountford)</para></listitem>
18606 <listitem><para>Most configuration options now settable via DSN (Hiroshi)</para></listitem>
18607 <listitem><para>Multibyte, performance fixes (Hiroshi)</para></listitem>
18608 <listitem><para>Allow driver to be used with iODBC or unixODBC (Peter E)</para></listitem>
18609 <listitem><para>MD5 password encryption support (Bruce)</para></listitem>
18610 <listitem><para>Add more compatibility functions to odbc.sql (Peter E)</para></listitem>
18615 <title><application>ECPG</></title>
18617 <listitem><para>EXECUTE ... INTO implemented (Christof Petig)</para></listitem>
18618 <listitem><para>Multiple row descriptor support (e.g. CARDINALITY) (Christof Petig)</para></listitem>
18619 <listitem><para>Fix for GRANT parameters (Lee Kindness)</para></listitem>
18620 <listitem><para>Fix INITIALLY DEFERRED bug</para></listitem>
18621 <listitem><para>Various bug fixes (Michael, Christof Petig)</para></listitem>
18622 <listitem><para>Auto allocation for indicator variable arrays (int *ind_p=NULL)</para></listitem>
18623 <listitem><para>Auto allocation for string arrays (char **foo_pp=NULL)</para></listitem>
18624 <listitem><para>ECPGfree_auto_mem fixed</para></listitem>
18625 <listitem><para>All function names with external linkage are now prefixed by ECPG</para></listitem>
18626 <listitem><para>Fixes for arrays of structures (Michael)</para></listitem>
18631 <title>Misc. Interfaces</title>
18633 <listitem><para>Python fix fetchone() (Gerhard Haring)</para></listitem>
18634 <listitem><para>Use UTF, Unicode in Tcl where appropriate (Vsevolod Lobko, Reinhard Max)</para></listitem>
18635 <listitem><para>Add Tcl COPY TO/FROM (ljb)</para></listitem>
18636 <listitem><para>Prevent output of default index op class in <application>pg_dump</> (Tom)</para></listitem>
18637 <listitem><para>Fix libpgeasy memory leak (Bruce)</para></listitem>
18642 <title>Build and Install</title>
18644 <listitem><para>Configure, dynamic loader, and shared library fixes (Peter E)</para></listitem>
18645 <listitem><para>Fixes in QNX 4 port (Bernd Tegge)</para></listitem>
18646 <listitem><para>Fixes in Cygwin and Windows ports (Jason Tishler, Gerhard Haring, Dmitry Yurtaev, Darko Prenosil, Mikhail Terekhov)</para></listitem>
18647 <listitem><para>Fix for Windows socket communication failures (Magnus, Mikhail Terekhov)</para></listitem>
18648 <listitem><para>Hurd compile fix (Oliver Elphick)</para></listitem>
18649 <listitem><para>BeOS fixes (Cyril Velter)</para></listitem>
18650 <listitem><para>Remove configure --enable-unicode-conversion, now enabled by multibyte (Tatsuo)</para></listitem>
18651 <listitem><para>AIX fixes (Tatsuo, Andreas)</para></listitem>
18652 <listitem><para>Fix parallel make (Peter E)</para></listitem>
18653 <listitem><para>Install SQL language manual pages into OS-specific directories (Peter E)</para></listitem>
18654 <listitem><para>Rename config.h to pg_config.h (Peter E)</para></listitem>
18655 <listitem><para>Reorganize installation layout of header files (Peter E)</para></listitem>
18660 <title>Source Code</title>
18662 <listitem><para>Remove SEP_CHAR (Bruce)</para></listitem>
18663 <listitem><para>New GUC hooks (Tom)</para></listitem>
18664 <listitem><para>Merge GUC and command line handling (Marko Kreen)</para></listitem>
18665 <listitem><para>Remove EXTEND INDEX (Martijn van Oosterhout, Tom)</para></listitem>
18666 <listitem><para>New pgjindent utility to indent java code (Bruce)</para></listitem>
18667 <listitem><para>Remove define of true/false when compiling under C++ (Leandro Fanzone, Tom)</para></listitem>
18668 <listitem><para>pgindent fixes (Bruce, Tom)</para></listitem>
18669 <listitem><para>Replace strcasecmp() with strcmp() where appropriate (Peter E)</para></listitem>
18670 <listitem><para>Dynahash portability improvements (Tom)</para></listitem>
18671 <listitem><para>Add 'volatile' usage in spinlock structures</para></listitem>
18672 <listitem><para>Improve signal handling logic (Tom)</para></listitem>
18677 <title>Contrib</title>
18679 <listitem><para>New contrib/rtree_gist (Oleg Bartunov, Teodor Sigaev)</para></listitem>
18680 <listitem><para>New contrib/tsearch full-text indexing (Oleg, Teodor Sigaev)</para></listitem>
18681 <listitem><para>Add contrib/dblink for remote database access (Joe Conway)</para></listitem>
18682 <listitem><para>contrib/ora2pg Oracle conversion utility (Gilles Darold)</para></listitem>
18683 <listitem><para>contrib/xml XML conversion utility (John Gray)</para></listitem>
18684 <listitem><para>contrib/fulltextindex fixes (Christopher Kings-Lynne)</para></listitem>
18685 <listitem><para>New contrib/fuzzystrmatch with levenshtein and metaphone, soundex merged (Joe Conway)</para></listitem>
18686 <listitem><para>Add contrib/intarray boolean queries, binary search, fixes (Oleg Bartunov)</para></listitem>
18687 <listitem><para>New pg_upgrade utility (Bruce)</para></listitem>
18688 <listitem><para>Add new pg_resetxlog options (Bruce, Tom)</para></listitem>
18695 <sect1 id="release-7-1-3">
18696 <title>Release 7.1.3</title>
18699 <title>Release date</title>
18700 <simpara>2001-08-15</simpara>
18704 <title>Migration to Version 7.1.3</title>
18707 A dump/restore is <emphasis>not</emphasis> required for those running
18713 <title>Changes</title>
18717 Remove unused WAL segments of large transactions (Tom)
18718 Multiaction rule fix (Tom)
18719 PL/pgSQL memory allocation fix (Jan)
18720 VACUUM buffer fix (Tom)
18721 Regression test fixes (Tom)
18722 pg_dump fixes for GRANT/REVOKE/comments on views, user-defined types (Tom)
18723 Fix subselects with DISTINCT ON or LIMIT (Tom)
18725 Disable COPY TO/FROM a view (Tom)
18726 Cygwin build (Jason Tishler)
18733 <sect1 id="release-7-1-2">
18734 <title>Release 7.1.2</title>
18737 <title>Release date</title>
18738 <simpara>2001-05-11</simpara>
18742 This has one fix from 7.1.1.
18747 <title>Migration to Version 7.1.2</title>
18750 A dump/restore is <emphasis>not</emphasis> required for those running
18756 <title>Changes</title>
18760 Fix PL/pgSQL SELECTs when returning no rows
18761 Fix for psql backslash core dump
18762 Referential integrity privilege fix
18771 <sect1 id="release-7-1-1">
18772 <title>Release 7.1.1</title>
18775 <title>Release date</title>
18776 <simpara>2001-05-05</simpara>
18780 This has a variety of fixes from 7.1.
18785 <title>Migration to Version 7.1.1</title>
18788 A dump/restore is <emphasis>not</emphasis> required for those running
18794 <title>Changes</title>
18798 Fix for numeric MODULO operator (Tom)
18799 pg_dump fixes (Philip)
18800 pg_dump can dump 7.0 databases (Philip)
18801 readline 4.2 fixes (Peter E)
18803 AIX, MSWIN, VAX, N32K fixes (Tom)
18804 Multibytes fixes (Tom)
18805 Unicode fixes (Tatsuo)
18806 Optimizer improvements (Tom)
18807 Fix for whole rows in functions (Tom)
18808 Fix for pg_ctl and option strings with spaces (Peter E)
18809 ODBC fixes (Hiroshi)
18810 EXTRACT can now take string argument (Thomas)
18811 Python fixes (Darcy)
18818 <sect1 id="release-7-1">
18819 <title>Release 7.1</title>
18822 <title>Release date</title>
18823 <simpara>2001-04-13</simpara>
18827 This release focuses on removing limitations that have existed in the
18828 <productname>PostgreSQL</productname> code for many years.
18832 Major changes in this release:
18838 Write-ahead Log (WAL)
18842 To maintain database consistency in case of an operating system crash,
18843 previous releases of <productname>PostgreSQL</productname> have forced
18844 all data modifications to disk before each transaction commit. With
18845 WAL, only one log file must be flushed to disk, greatly improving
18846 performance. If you have been using -F in previous releases to
18847 disable disk flushes, you might want to consider discontinuing its use.
18858 TOAST - Previous releases had a compiled-in row length limit,
18859 typically 8k - 32k. This limit made storage of long text fields
18860 difficult. With TOAST, long rows of any length can be stored with good
18872 We now support outer joins. The UNION/NOT IN
18873 workaround for outer joins is no longer required. We use the SQL92
18885 The previous C function manager did not
18886 handle null values properly, nor did it support 64-bit <acronym>CPU</acronym>'s (Alpha). The new
18887 function manager does. You can continue using your old custom
18888 functions, but you might want to rewrite them in the future to use the new
18889 function manager call interface.
18900 A large number of complex queries that were
18901 unsupported in previous releases now work. Many combinations of views,
18902 aggregates, UNION, LIMIT, cursors, subqueries, and inherited tables
18903 now work properly. Inherited tables are now accessed by default.
18904 Subqueries in FROM are now supported.
18912 <title>Migration to Version 7.1</title>
18915 A dump/restore using pg_dump is required for those wishing to migrate
18916 data from any previous release.
18921 <title>Changes</title>
18927 Many multibyte/Unicode/locale fixes (Tatsuo and others)
18928 More reliable ALTER TABLE RENAME (Tom)
18929 Kerberos V fixes (David Wragg)
18930 Fix for INSERT INTO...SELECT where targetlist has subqueries (Tom)
18931 Prompt username/password on standard error (Bruce)
18932 Large objects inv_read/inv_write fixes (Tom)
18933 Fixes for to_char(), to_date(), to_ascii(), and to_timestamp() (Karel,
18935 Prevent query expressions from leaking memory (Tom)
18936 Allow UPDATE of arrays elements (Tom)
18937 Wake up lock waiters during cancel (Hiroshi)
18938 Fix rare cursor crash when using hash join (Tom)
18939 Fix for DROP TABLE/INDEX in rolled-back transaction (Hiroshi)
18940 Fix psql crash from \l+ if MULTIBYTE enabled (Peter E)
18941 Fix truncation of rule names during CREATE VIEW (Ross Reedstrom)
18942 Fix PL/perl (Alex Kapranoff)
18943 Disallow LOCK on views (Mark Hollomon)
18944 Disallow INSERT/UPDATE/DELETE on views (Mark Hollomon)
18945 Disallow DROP RULE, CREATE INDEX, TRUNCATE on views (Mark Hollomon)
18946 Allow PL/pgSQL accept non-ASCII identifiers (Tatsuo)
18947 Allow views to proper handle GROUP BY, aggregates, DISTINCT (Tom)
18948 Fix rare failure with TRUNCATE command (Tom)
18949 Allow UNION/INTERSECT/EXCEPT to be used with ALL, subqueries, views,
18950 DISTINCT, ORDER BY, SELECT...INTO (Tom)
18951 Fix parser failures during aborted transactions (Tom)
18952 Allow temporary relations to properly clean up indexes (Bruce)
18953 Fix VACUUM problem with moving rows in same page (Tom)
18954 Modify pg_dump to better handle user-defined items in template1 (Philip)
18955 Allow LIMIT in VIEW (Tom)
18956 Require cursor FETCH to honor LIMIT (Tom)
18957 Allow PRIMARY/FOREIGN Key definitions on inherited columns (Stephan)
18958 Allow ORDER BY, LIMIT in subqueries (Tom)
18959 Allow UNION in CREATE RULE (Tom)
18960 Make ALTER/DROP TABLE rollback-able (Vadim, Tom)
18961 Store initdb collation in pg_control so collation cannot be changed (Tom)
18962 Fix INSERT...SELECT with rules (Tom)
18963 Fix FOR UPDATE inside views and subselects (Tom)
18964 Fix OVERLAPS operators conform to SQL92 spec regarding NULLs (Tom)
18965 Fix lpad() and rpad() to handle length less than input string (Tom)
18966 Fix use of NOTIFY in some rules (Tom)
18967 Overhaul btree code (Tom)
18968 Fix NOT NULL use in Pl/pgSQL variables (Tom)
18969 Overhaul GIST code (Oleg)
18970 Fix CLUSTER to preserve constraints and column default (Tom)
18971 Improved deadlock detection handling (Tom)
18972 Allow multiple SERIAL columns in a table (Tom)
18973 Prevent occasional index corruption (Vadim)
18977 Add OUTER JOINs (Tom)
18978 Function manager overhaul (Tom)
18979 Allow ALTER TABLE RENAME on indexes (Tom)
18980 Improve CLUSTER (Tom)
18981 Improve ps status display for more platforms (Peter E, Marc)
18982 Improve CREATE FUNCTION failure message (Ross)
18983 JDBC improvements (Peter, Travis Bauer, Christopher Cain, William Webber,
18985 Grand Unified Configuration scheme/GUC. Many options can now be set in
18986 data/postgresql.conf, postmaster/postgres flags, or SET commands (Peter E)
18987 Improved handling of file descriptor cache (Tom)
18988 New warning code about auto-created table alias entries (Bruce)
18989 Overhaul initdb process (Tom, Peter E)
18990 Overhaul of inherited tables; inherited tables now accessed by default;
18991 new ONLY key word prevents it (Chris Bitmead, Tom)
18992 ODBC cleanups/improvements (Nick Gorham, Stephan Szabo, Zoltan Kovacs,
18994 Allow renaming of temp tables (Tom)
18995 Overhaul memory manager contexts (Tom)
18996 pg_dumpall uses CREATE USER or CREATE GROUP rather using COPY (Peter E)
18997 Overhaul pg_dump (Philip Warner)
18998 Allow pg_hba.conf secondary password file to specify only username (Peter E)
18999 Allow TEMPORARY or TEMP key word when creating temporary tables (Bruce)
19000 New memory leak checker (Karel)
19001 New SET SESSION CHARACTERISTICS (Thomas)
19002 Allow nested block comments (Thomas)
19003 Add WITHOUT TIME ZONE type qualifier (Thomas)
19004 New ALTER TABLE ADD CONSTRAINT (Stephan)
19005 Use NUMERIC accumulators for INTEGER aggregates (Tom)
19006 Overhaul aggregate code (Tom)
19007 New VARIANCE and STDDEV() aggregates
19008 Improve dependency ordering of pg_dump (Philip)
19009 New pg_restore command (Philip)
19010 New pg_dump tar output option (Philip)
19011 New pg_dump of large objects (Philip)
19012 New ESCAPE option to LIKE (Thomas)
19013 New case-insensitive LIKE - ILIKE (Thomas)
19014 Allow functional indexes to use binary-compatible type (Tom)
19015 Allow SQL functions to be used in more contexts (Tom)
19016 New pg_config utility (Peter E)
19017 New PL/pgSQL EXECUTE command which allows dynamic SQL and utility statements
19019 New PL/pgSQL GET DIAGNOSTICS statement for SPI value access (Jan)
19020 New quote_identifiers() and quote_literal() functions (Jan)
19021 New ALTER TABLE table OWNER TO user command (Mark Hollomon)
19022 Allow subselects in FROM, i.e. FROM (SELECT ...) [AS] alias (Tom)
19023 Update PyGreSQL to version 3.1 (D'Arcy)
19024 Store tables as files named by OID (Vadim)
19025 New SQL function setval(seq,val,bool) for use in pg_dump (Philip)
19026 Require DROP VIEW to remove views, no DROP TABLE (Mark)
19027 Allow DROP VIEW view1, view2 (Mark)
19028 Allow multiple objects in DROP INDEX, DROP RULE, and DROP TYPE (Tom)
19029 Allow automatic conversion to/from Unicode (Tatsuo, Eiji)
19030 New /contrib/pgcrypto hashing functions (Marko Kreen)
19031 New pg_dumpall --globals-only option (Peter E)
19032 New CHECKPOINT command for WAL which creates new WAL log file (Vadim)
19033 New AT TIME ZONE syntax (Thomas)
19034 Allow location of Unix domain socket to be configurable (David J. MacKenzie)
19035 Allow postmaster to listen on a specific IP address (David J. MacKenzie)
19036 Allow socket path name to be specified in hostname by using leading slash
19037 (David J. MacKenzie)
19038 Allow CREATE DATABASE to specify template database (Tom)
19039 New utility to convert MySQL schema dumps to SQL92 and PostgreSQL (Thomas)
19040 New /contrib/rserv replication toolkit (Vadim)
19041 New file format for COPY BINARY (Tom)
19042 New /contrib/oid2name to map numeric files to table names (B Palmer)
19043 New "idle in transaction" ps status message (Marc)
19044 Update to pgaccess 0.98.7 (Constantin Teodorescu)
19045 pg_ctl now defaults to -w (wait) on shutdown, new -l (log) option
19046 Add rudimentary dependency checking to pg_dump (Philip)
19050 Fix INET/CIDR type ordering and add new functions (Tom)
19051 Make OID behave as an unsigned type (Tom)
19052 Allow BIGINT as synonym for INT8 (Peter E)
19053 New int2 and int8 comparison operators (Tom)
19054 New BIT and BIT VARYING types (Adriaan Joubert, Tom, Peter E)
19055 CHAR() no longer faster than VARCHAR() because of TOAST (Tom)
19056 New GIST seg/cube examples (Gene Selkov)
19057 Improved round(numeric) handling (Tom)
19058 Fix CIDR output formatting (Tom)
19059 New CIDR abbrev() function (Tom)
19063 Write-Ahead Log (WAL) to provide crash recovery with less performance
19065 ANALYZE stage of VACUUM no longer exclusively locks table (Bruce)
19066 Reduced file seeks (Denis Perchine)
19067 Improve BTREE code for duplicate keys (Tom)
19068 Store all large objects in a single table (Denis Perchine, Tom)
19069 Improve memory allocation performance (Karel, Tom)
19073 New function manager call conventions (Tom)
19074 SGI portability fixes (David Kaelbling)
19075 New configure --enable-syslog option (Peter E)
19076 New BSDI README (Bruce)
19077 configure script moved to top level, not /src (Peter E)
19078 Makefile/configuration/compilation overhaul (Peter E)
19079 New configure --with-python option (Peter E)
19080 Solaris cleanups (Peter E)
19081 Overhaul /contrib Makefiles (Karel)
19082 New OpenSSL configuration option (Magnus, Peter E)
19083 AIX fixes (Andreas)
19084 QNX fixes (Maurizio)
19085 New heap_open(), heap_openr() API (Tom)
19086 Remove colon and semi-colon operators (Thomas)
19087 New pg_class.relkind value for views (Mark Hollomon)
19088 Rename ichar() to chr() (Karel)
19089 New documentation for btrim(), ascii(), chr(), repeat() (Karel)
19090 Fixes for NT/Cygwin (Pete Forman)
19091 AIX port fixes (Andreas)
19092 New BeOS port (David Reid, Cyril Velter)
19093 Add proofreader's changes to docs (Addison-Wesley, Bruce)
19094 New Alpha spinlock code (Adriaan Joubert, Compaq)
19095 UnixWare port overhaul (Peter E)
19096 New Darwin/MacOS X port (Peter Bierman, Bruce Hartzler)
19097 New FreeBSD Alpha port (Alfred)
19098 Overhaul shared memory segments (Tom)
19099 Add IBM S/390 support (Neale Ferguson)
19100 Moved macmanuf to /contrib (Larry Rosenman)
19101 Syslog improvements (Larry Rosenman)
19102 New template0 database that contains no user additions (Tom)
19103 New /contrib/cube and /contrib/seg GIST sample code (Gene Selkov)
19104 Allow NetBSD's libedit instead of readline (Peter)
19105 Improved assembly language source code format (Bruce)
19106 New contrib/pg_logger
19107 New --template option to createdb
19108 New contrib/pg_control utility (Oliver)
19109 New FreeBSD tools ipc_check, start-scripts/freebsd
19116 <sect1 id="release-7-0-3">
19117 <title>Release 7.0.3</title>
19120 <title>Release date</title>
19121 <simpara>2000-11-11</simpara>
19125 This has a variety of fixes from 7.0.2.
19130 <title>Migration to Version 7.0.3</title>
19133 A dump/restore is <emphasis>not</emphasis> required for those running
19139 <title>Changes</title>
19144 Large object fix (Tom)
19145 Fix lean in COPY WITH OIDS leak (Tom)
19146 Fix backwards-index-scan (Tom)
19147 Fix SELECT ... FOR UPDATE so it checks for duplicate keys (Hiroshi)
19148 Add --enable-syslog to configure (Marc)
19149 Fix abort transaction at backend exit in rare cases (Tom)
19150 Fix for psql \l+ when multibyte enabled (Tatsuo)
19151 Allow PL/pgSQL to accept non ascii identifiers (Tatsuo)
19152 Make vacuum always flush buffers (Tom)
19153 Fix to allow cancel while waiting for a lock (Hiroshi)
19154 Fix for memory allocation problem in user authentication code (Tom)
19155 Remove bogus use of int4out() (Tom)
19156 Fixes for multiple subqueries in COALESCE or BETWEEN (Tom)
19157 Fix for failure of triggers on heap open in certain cases (Jeroen van
19159 Fix for erroneous selectivity of not-equals (Tom)
19160 Fix for erroneous use of strcmp() (Tom)
19161 Fix for bug where storage manager accesses items beyond end of file
19163 Fix to include kernel errno message in all smgr elog messages (Tom)
19164 Fix for '.' not in PATH at build time (SL Baur)
19165 Fix for out-of-file-descriptors error (Tom)
19166 Fix to make pg_dump dump 'iscachable' flag for functions (Tom)
19167 Fix for subselect in targetlist of Append node (Tom)
19168 Fix for mergejoin plans (Tom)
19169 Fix TRUNCATE failure on relations with indexes (Tom)
19170 Avoid database-wide restart on write error (Hiroshi)
19171 Fix nodeMaterial to honor chgParam by recomputing its output (Tom)
19172 Fix VACUUM problem with moving chain of update row versions when source
19173 and destination of a row version lie on the same page (Tom)
19174 Fix user.c CommandCounterIncrement (Tom)
19175 Fix for AM/PM boundary problem in to_char() (Karel Zak)
19176 Fix TIME aggregate handling (Tom)
19177 Fix to_char() to avoid coredump on NULL input (Tom)
19179 Fix for inserting/copying longer multibyte strings into char() data
19181 Fix for crash of backend, on abort (Tom)
19188 <sect1 id="release-7-0-2">
19189 <title>Release 7.0.2</title>
19192 <title>Release date</title>
19193 <simpara>2000-06-05</simpara>
19197 This is a repackaging of 7.0.1 with added documentation.
19202 <title>Migration to Version 7.0.2</title>
19205 A dump/restore is <emphasis>not</emphasis> required for those running
19211 <title>Changes</title>
19215 Added documentation to tarball.
19222 <sect1 id="release-7-0-1">
19223 <title>Release 7.0.1</title>
19226 <title>Release date</title>
19227 <simpara>2000-06-01</simpara>
19231 This is a cleanup release for 7.0.
19235 <title>Migration to Version 7.0.1</title>
19238 A dump/restore is <emphasis>not</emphasis> required for those running
19244 <title>Changes</title>
19248 Fix many CLUSTER failures (Tom)
19249 Allow ALTER TABLE RENAME works on indexes (Tom)
19250 Fix plpgsql to handle datetime->timestamp and timespan->interval (Bruce)
19251 New configure --with-setproctitle switch to use setproctitle() (Marc, Bruce)
19252 Fix the off by one errors in ResultSet from 6.5.3, and more.
19253 jdbc ResultSet fixes (Joseph Shraibman)
19254 optimizer tunings (Tom)
19255 Fix create user for pgaccess
19256 Fix for UNLISTEN failure
19257 IRIX fixes (David Kaelbling)
19258 QNX fixes (Andreas Kardos)
19259 Reduce COPY IN lock level (Tom)
19260 Change libpqeasy to use PQconnectdb() style parameters (Bruce)
19261 Fix pg_dump to handle OID indexes (Tom)
19262 Fix small memory leak (Tom)
19263 Solaris fix for createdb/dropdb (Tatsuo)
19264 Fix for non-blocking connections (Alfred Perlstein)
19265 Fix improper recovery after RENAME TABLE failures (Tom)
19266 Copy pg_ident.conf.sample into /lib directory in install (Bruce)
19267 Add SJIS UDC (NEC selection IBM kanji) support (Eiji Tokuya)
19268 Fix too long syslog message (Tatsuo)
19269 Fix problem with quoted indexes that are too long (Tom)
19270 JDBC ResultSet.getTimestamp() fix (Gregory Krasnow & Floyd Marinescu)
19271 ecpg changes (Michael)
19277 <sect1 id="release-7-0">
19278 <title>Release 7.0</title>
19281 <title>Release date</title>
19282 <simpara>2000-05-08</simpara>
19286 This release contains improvements in many areas, demonstrating
19287 the continued growth of <productname>PostgreSQL</productname>.
19288 There are more improvements and fixes in 7.0 than in any previous
19289 release. The developers have confidence that this is the best
19290 release yet; we do our best to put out only solid releases, and
19291 this one is no exception.
19295 Major changes in this release:
19305 Foreign keys are now implemented, with the exception of PARTIAL MATCH
19306 foreign keys. Many users have been asking for this feature, and we are
19307 pleased to offer it.
19318 Continuing on work started a year ago, the optimizer has been
19319 improved, allowing better query plan selection and faster performance
19320 with less memory usage.
19327 Updated <application>psql</application>
19331 <application>psql</application>, our interactive terminal monitor, has been
19332 updated with a variety of new features. See the <application>psql</application> manual page for details.
19343 SQL92 join syntax is now supported, though only as
19344 <literal>INNER JOIN</> for this release. <literal>JOIN</>,
19345 <literal>NATURAL JOIN</>, <literal>JOIN</>/<literal>USING</>,
19346 and <literal>JOIN</>/<literal>ON</> are available, as are
19347 column correlation names.
19355 <title>Migration to Version 7.0</title>
19358 A dump/restore using <application>pg_dump</application>
19359 is required for those wishing to migrate data from any
19360 previous release of <productname>PostgreSQL</productname>.
19361 For those upgrading from 6.5.*, you can instead use
19362 <application>pg_upgrade</application> to upgrade to this
19363 release; however, a full dump/reload installation is always the
19364 most robust method for upgrades.
19368 Interface and compatibility issues to consider for the new
19375 The date/time types <type>datetime</type> and
19376 <type>timespan</type> have been superseded by the
19377 SQL92-defined types <type>timestamp</type> and
19378 <type>interval</type>. Although there has been some effort to
19379 ease the transition by allowing
19380 <productname>PostgreSQL</productname> to recognize
19381 the deprecated type names and translate them to the new type
19382 names, this mechanism cannot be completely transparent to
19383 your existing application.
19389 The optimizer has been substantially improved in the area of
19390 query cost estimation. In some cases, this will result in
19391 decreased query times as the optimizer makes a better choice
19392 for the preferred plan. However, in a small number of cases,
19393 usually involving pathological distributions of data, your
19394 query times might go up. If you are dealing with large amounts
19395 of data, you might want to check your queries to verify
19402 The <acronym>JDBC</acronym> and <acronym>ODBC</acronym>
19403 interfaces have been upgraded and extended.
19409 The string function <function>CHAR_LENGTH</function> is now a
19410 native function. Previous versions translated this into a call
19411 to <function>LENGTH</function>, which could result in
19412 ambiguity with other types implementing
19413 <function>LENGTH</function> such as the geometric types.
19420 <title>Changes</title>
19426 Prevent function calls exceeding maximum number of arguments (Tom)
19427 Improve CASE construct (Tom)
19428 Fix SELECT coalesce(f1,0) FROM int4_tbl GROUP BY f1 (Tom)
19429 Fix SELECT sentence.words[0] FROM sentence GROUP BY sentence.words[0] (Tom)
19430 Fix GROUP BY scan bug (Tom)
19431 Improvements in SQL grammar processing (Tom)
19432 Fix for views involved in INSERT ... SELECT ... (Tom)
19433 Fix for SELECT a/2, a/2 FROM test_missing_target GROUP BY a/2 (Tom)
19434 Fix for subselects in INSERT ... SELECT (Tom)
19435 Prevent INSERT ... SELECT ... ORDER BY (Tom)
19436 Fixes for relations greater than 2GB, including vacuum
19437 Improve propagating system table changes to other backends (Tom)
19438 Improve propagating user table changes to other backends (Tom)
19439 Fix handling of temp tables in complex situations (Bruce, Tom)
19440 Allow table locking at table open, improving concurrent reliability (Tom)
19441 Properly quote sequence names in pg_dump (Ross J. Reedstrom)
19442 Prevent DROP DATABASE while others accessing
19443 Prevent any rows from being returned by GROUP BY if no rows processed (Tom)
19444 Fix SELECT COUNT(1) FROM table WHERE ...' if no rows matching WHERE (Tom)
19445 Fix pg_upgrade so it works for MVCC (Tom)
19446 Fix for SELECT ... WHERE x IN (SELECT ... HAVING SUM(x) > 1) (Tom)
19447 Fix for "f1 datetime DEFAULT 'now'" (Tom)
19448 Fix problems with CURRENT_DATE used in DEFAULT (Tom)
19449 Allow comment-only lines, and ;;; lines too. (Tom)
19450 Improve recovery after failed disk writes, disk full (Hiroshi)
19451 Fix cases where table is mentioned in FROM but not joined (Tom)
19452 Allow HAVING clause without aggregate functions (Tom)
19453 Fix for "--" comment and no trailing newline, as seen in perl interface
19454 Improve pg_dump failure error reports (Bruce)
19455 Allow sorts and hashes to exceed 2GB file sizes (Tom)
19456 Fix for pg_dump dumping of inherited rules (Tom)
19457 Fix for NULL handling comparisons (Tom)
19458 Fix inconsistent state caused by failed CREATE/DROP commands (Hiroshi)
19459 Fix for dbname with dash
19460 Prevent DROP INDEX from interfering with other backends (Tom)
19461 Fix file descriptor leak in verify_password()
19462 Fix for "Unable to identify an operator =$" problem
19463 Fix ODBC so no segfault if CommLog and Debug enabled (Dirk Niggemann)
19464 Fix for recursive exit call (Massimo)
19465 Fix for extra-long timezones (Jeroen van Vianen)
19466 Make pg_dump preserve primary key information (Peter E)
19467 Prevent databases with single quotes (Peter E)
19468 Prevent DROP DATABASE inside transaction (Peter E)
19469 ecpg memory leak fixes (Stephen Birch)
19470 Fix for SELECT null::text, SELECT int4fac(null) and SELECT 2 + (null) (Tom)
19471 Y2K timestamp fix (Massimo)
19472 Fix for VACUUM 'HEAP_MOVED_IN was not expected' errors (Tom)
19473 Fix for views with tables/columns containing spaces (Tom)
19474 Prevent privileges on indexes (Peter E)
19475 Fix for spinlock stuck problem when error is generated (Hiroshi)
19476 Fix ipcclean on Linux
19477 Fix handling of NULL constraint conditions (Tom)
19478 Fix memory leak in odbc driver (Nick Gorham)
19479 Fix for privilege check on UNION tables (Tom)
19480 Fix to allow SELECT 'a' LIKE 'a' (Tom)
19481 Fix for SELECT 1 + NULL (Tom)
19483 Fix log() on numeric type (Tom)
19484 Deprecate ':' and ';' operators
19485 Allow vacuum of temporary tables
19486 Disallow inherited columns with the same name as new columns
19487 Recover or force failure when disk space is exhausted (Hiroshi)
19488 Fix INSERT INTO ... SELECT with AS columns matching result columns
19489 Fix INSERT ... SELECT ... GROUP BY groups by target columns not source columns (Tom)
19490 Fix CREATE TABLE test (a char(5) DEFAULT text '', b int4) with INSERT (Tom)
19491 Fix UNION with LIMIT
19492 Fix CREATE TABLE x AS SELECT 1 UNION SELECT 2
19493 Fix CREATE TABLE test(col char(2) DEFAULT user)
19494 Fix mismatched types in CREATE TABLE ... DEFAULT
19495 Fix SELECT * FROM pg_class where oid in (0,-1)
19496 Fix SELECT COUNT('asdf') FROM pg_class WHERE oid=12
19497 Prevent user who can create databases can modifying pg_database table (Peter E)
19498 Fix btree to give a useful elog when key > 1/2 (page - overhead) (Tom)
19499 Fix INSERT of 0.0 into DECIMAL(4,4) field (Tom)
19503 New CLI interface include file sqlcli.h, based on SQL3/SQL98
19504 Remove all limits on query length, row length limit still exists (Tom)
19505 Update jdbc protocol to 2.0 (Jens Glaser <email>jens@jens.de</email>)
19506 Add TRUNCATE command to quickly truncate relation (Mike Mascari)
19507 Fix to give super user and createdb user proper update catalog rights (Peter E)
19508 Allow ecpg bool variables to have NULL values (Christof)
19509 Issue ecpg error if NULL value for variable with no NULL indicator (Christof)
19510 Allow ^C to cancel COPY command (Massimo)
19511 Add SET FSYNC and SHOW PG_OPTIONS commands(Massimo)
19512 Function name overloading for dynamically-loaded C functions (Frankpitt)
19513 Add CmdTuples() to libpq++(Vince)
19514 New CREATE CONSTRAINT TRIGGER and SET CONSTRAINTS commands(Jan)
19515 Allow CREATE FUNCTION/WITH clause to be used for all language types
19516 configure --enable-debug adds -g (Peter E)
19517 configure --disable-debug removes -g (Peter E)
19518 Allow more complex default expressions (Tom)
19519 First real FOREIGN KEY constraint trigger functionality (Jan)
19520 Add FOREIGN KEY ... MATCH FULL ... ON DELETE CASCADE (Jan)
19521 Add FOREIGN KEY ... MATCH <unspecified> referential actions (Don Baccus)
19522 Allow WHERE restriction on ctid (physical heap location) (Hiroshi)
19523 Move pginterface from contrib to interface directory, rename to pgeasy (Bruce)
19524 Change pgeasy connectdb() parameter ordering (Bruce)
19525 Require SELECT DISTINCT target list to have all ORDER BY columns (Tom)
19526 Add Oracle's COMMENT ON command (Mike Mascari <email>mascarim@yahoo.com</email>)
19527 libpq's PQsetNoticeProcessor function now returns previous hook(Peter E)
19528 Prevent PQsetNoticeProcessor from being set to NULL (Peter E)
19529 Make USING in COPY optional (Bruce)
19530 Allow subselects in the target list (Tom)
19531 Allow subselects on the left side of comparison operators (Tom)
19532 New parallel regression test (Jan)
19533 Change backend-side COPY to write files with permissions 644 not 666 (Tom)
19534 Force permissions on PGDATA directory to be secure, even if it exists (Tom)
19535 Added psql LASTOID variable to return last inserted oid (Peter E)
19536 Allow concurrent vacuum and remove pg_vlock vacuum lock file (Tom)
19537 Add privilege check for vacuum (Peter E)
19538 New libpq functions to allow asynchronous connections: PQconnectStart(),
19539 PQconnectPoll(), PQresetStart(), PQresetPoll(), PQsetenvStart(),
19540 PQsetenvPoll(), PQsetenvAbort (Ewan Mellor)
19541 New libpq PQsetenv() function (Ewan Mellor)
19542 create/alter user extension (Peter E)
19543 New postmaster.pid and postmaster.opts under $PGDATA (Tatsuo)
19544 New scripts for create/drop user/db (Peter E)
19545 Major psql overhaul (Peter E)
19546 Add const to libpq interface (Peter E)
19547 New libpq function PQoidValue (Peter E)
19548 Show specific non-aggregate causing problem with GROUP BY (Tom)
19549 Make changes to pg_shadow recreate pg_pwd file (Peter E)
19550 Add aggregate(DISTINCT ...) (Tom)
19551 Allow flag to control COPY input/output of NULLs (Peter E)
19552 Make postgres user have a password by default (Peter E)
19553 Add CREATE/ALTER/DROP GROUP (Peter E)
19554 All administration scripts now support --long options (Peter E, Karel)
19555 Vacuumdb script now supports --all option (Peter E)
19556 ecpg new portable FETCH syntax
19557 Add ecpg EXEC SQL IFDEF, EXEC SQL IFNDEF, EXEC SQL ELSE, EXEC SQL ELIF
19558 and EXEC SQL ENDIF directives
19559 Add pg_ctl script to control backend start-up (Tatsuo)
19560 Add postmaster.opts.default file to store start-up flags (Tatsuo)
19561 Allow --with-mb=SQL_ASCII
19562 Increase maximum number of index keys to 16 (Bruce)
19563 Increase maximum number of function arguments to 16 (Bruce)
19564 Allow configuration of maximum number of index keys and arguments (Bruce)
19565 Allow unprivileged users to change their passwords (Peter E)
19566 Password authentication enabled; required for new users (Peter E)
19567 Disallow dropping a user who owns a database (Peter E)
19568 Change initdb option --with-mb to --enable-multibyte
19569 Add option for initdb to prompts for superuser password (Peter E)
19570 Allow complex type casts like col::numeric(9,2) and col::int2::float8 (Tom)
19571 Updated user interfaces on initdb, initlocation, pg_dump, ipcclean (Peter E)
19572 New pg_char_to_encoding() and pg_encoding_to_char() functions (Tatsuo)
19573 libpq non-blocking mode (Alfred Perlstein)
19574 Improve conversion of types in casts that don't specify a length
19575 New plperl internal programming language (Mark Hollomon)
19576 Allow COPY IN to read file that do not end with a newline (Tom)
19577 Indicate when long identifiers are truncated (Tom)
19578 Allow aggregates to use type equivalency (Peter E)
19579 Add Oracle's to_char(), to_date(), to_datetime(), to_timestamp(), to_number()
19580 conversion functions (Karel Zak <zakkr@zf.jcu.cz>)
19581 Add SELECT DISTINCT ON (expr [, expr ...]) targetlist ... (Tom)
19582 Check to be sure ORDER BY is compatible with the DISTINCT operation (Tom)
19583 Add NUMERIC and int8 types to ODBC
19584 Improve EXPLAIN results for Append, Group, Agg, Unique (Tom)
19585 Add ALTER TABLE ... ADD FOREIGN KEY (Stephan Szabo)
19586 Allow SELECT .. FOR UPDATE in PL/pgSQL (Hiroshi)
19587 Enable backward sequential scan even after reaching EOF (Hiroshi)
19588 Add btree indexing of boolean values, >= and <= (Don Baccus)
19589 Print current line number when COPY FROM fails (Massimo)
19590 Recognize POSIX time zone e.g. "PST+8" and "GMT-8" (Thomas)
19591 Add DEC as synonym for DECIMAL (Thomas)
19592 Add SESSION_USER as SQL92 key word, same as CURRENT_USER (Thomas)
19593 Implement SQL92 column aliases (aka correlation names) (Thomas)
19594 Implement SQL92 join syntax (Thomas)
19595 Make INTERVAL reserved word allowed as a column identifier (Thomas)
19596 Implement REINDEX command (Hiroshi)
19597 Accept ALL in aggregate function SUM(ALL col) (Tom)
19598 Prevent GROUP BY from using column aliases (Tom)
19599 New psql \encoding option (Tatsuo)
19600 Allow PQrequestCancel() to terminate when in waiting-for-lock state (Hiroshi)
19601 Allow negation of a negative number in all cases
19602 Add ecpg descriptors (Christof, Michael)
19603 Allow CREATE VIEW v AS SELECT f1::char(8) FROM tbl
19604 Allow casts with length, like foo::char(8)
19605 New libpq functions PQsetClientEncoding(), PQclientEncoding() (Tatsuo)
19606 Add support for SJIS user defined characters (Tatsuo)
19607 Larger views/rules supported
19608 Make libpq's PQconndefaults() thread-safe (Tom)
19609 Disable // as comment to be ANSI conforming, should use -- (Tom)
19610 Allow column aliases on views CREATE VIEW name (collist)
19611 Fixes for views with subqueries (Tom)
19612 Allow UPDATE table SET fld = (SELECT ...) (Tom)
19613 SET command options no longer require quotes
19614 Update pgaccess to 0.98.6
19615 New SET SEED command
19616 New pg_options.sample file
19617 New SET FSYNC command (Massimo)
19618 Allow pg_descriptions when creating tables
19619 Allow pg_descriptions when creating types, columns, and functions
19620 Allow psql \copy to allow delimiters (Peter E)
19621 Allow psql to print nulls as distinct from "" [null] (Peter E)
19625 Many array fixes (Tom)
19626 Allow bare column names to be subscripted as arrays (Tom)
19627 Improve type casting of int and float constants (Tom)
19628 Cleanups for int8 inputs, range checking, and type conversion (Tom)
19629 Fix for SELECT timespan('21:11:26'::time) (Tom)
19630 netmask('x.x.x.x/0') is 255.255.255.255 instead of 0.0.0.0 (Oleg Sharoiko)
19631 Add btree index on NUMERIC (Jan)
19632 Perl fix for large objects containing NUL characters (Douglas Thomson)
19633 ODBC fix for large objects (free)
19634 Fix indexing of cidr data type
19635 Fix for Ethernet MAC addresses (macaddr type) comparisons
19636 Fix for date/time types when overflows happened in computations (Tom)
19637 Allow array on int8 (Peter E)
19638 Fix for rounding/overflow of NUMERIC type, like NUMERIC(4,4) (Tom)
19639 Allow NUMERIC arrays
19640 Fix bugs in NUMERIC ceil() and floor() functions (Tom)
19641 Make char_length()/octet_length including trailing blanks (Tom)
19642 Made abstime/reltime use int4 instead of time_t (Peter E)
19643 New lztext data type for compressed text fields
19644 Revise code to handle coercion of int and float constants (Tom)
19645 Start at new code to implement a BIT and BIT VARYING type (Adriaan Joubert)
19646 NUMERIC now accepts scientific notation (Tom)
19647 NUMERIC to int4 rounds (Tom)
19648 Convert float4/8 to NUMERIC properly (Tom)
19649 Allow type conversion with NUMERIC (Thomas)
19650 Make ISO date style (2000-02-16 09:33) the default (Thomas)
19651 Add NATIONAL CHAR [ VARYING ] (Thomas)
19652 Allow NUMERIC round and trunc to accept negative scales (Tom)
19653 New TIME WITH TIME ZONE type (Thomas)
19654 Add MAX()/MIN() on time type (Thomas)
19655 Add abs(), mod(), fac() for int8 (Thomas)
19656 Rename functions to round(), sqrt(), cbrt(), pow() for float8 (Thomas)
19657 Add transcendental math functions (e.g. sin(), acos()) for float8 (Thomas)
19658 Add exp() and ln() for NUMERIC type
19659 Rename NUMERIC power() to pow() (Thomas)
19660 Improved TRANSLATE() function (Edwin Ramirez, Tom)
19661 Allow X=-Y operators (Tom)
19662 Allow SELECT float8(COUNT(*))/(SELECT COUNT(*) FROM t) FROM t GROUP BY f1; (Tom)
19663 Allow LOCALE to use indexes in regular expression searches (Tom)
19664 Allow creation of functional indexes to use default types
19668 Prevent exponential space consumption with many AND's and OR's (Tom)
19669 Collect attribute selectivity values for system columns (Tom)
19670 Reduce memory usage of aggregates (Tom)
19671 Fix for LIKE optimization to use indexes with multibyte encodings (Tom)
19672 Fix r-tree index optimizer selectivity (Thomas)
19673 Improve optimizer selectivity computations and functions (Tom)
19674 Optimize btree searching for cases where many equal keys exist (Tom)
19675 Enable fast LIKE index processing only if index present (Tom)
19676 Re-use free space on index pages with duplicates (Tom)
19677 Improve hash join processing (Tom)
19678 Prevent descending sort if result is already sorted(Hiroshi)
19679 Allow commuting of index scan query qualifications (Tom)
19680 Prefer index scans in cases where ORDER BY/GROUP BY is required (Tom)
19681 Allocate large memory requests in fix-sized chunks for performance (Tom)
19682 Fix vacuum's performance by reducing memory allocation requests (Tom)
19683 Implement constant-expression simplification (Bernard Frankpitt, Tom)
19684 Use secondary columns to be used to determine start of index scan (Hiroshi)
19685 Prevent quadruple use of disk space when doing internal sorting (Tom)
19686 Faster sorting by calling fewer functions (Tom)
19687 Create system indexes to match all system caches (Bruce, Hiroshi)
19688 Make system caches use system indexes (Bruce)
19689 Make all system indexes unique (Bruce)
19690 Improve pg_statistics management for VACUUM speed improvement (Tom)
19691 Flush backend cache less frequently (Tom, Hiroshi)
19692 COPY now reuses previous memory allocation, improving performance (Tom)
19693 Improve optimization cost estimation (Tom)
19694 Improve optimizer estimate of range queries x > lowbound AND x < highbound (Tom)
19695 Use DNF instead of CNF where appropriate (Tom, Taral)
19696 Further cleanup for OR-of-AND WHERE-clauses (Tom)
19697 Make use of index in OR clauses (x = 1 AND y = 2) OR (x = 2 AND y = 4) (Tom)
19698 Smarter optimizer computations for random index page access (Tom)
19699 New SET variable to control optimizer costs (Tom)
19700 Optimizer queries based on LIMIT, OFFSET, and EXISTS qualifications (Tom)
19701 Reduce optimizer internal housekeeping of join paths for speedup (Tom)
19702 Major subquery speedup (Tom)
19703 Fewer fsync writes when fsync is not disabled (Tom)
19704 Improved LIKE optimizer estimates (Tom)
19705 Prevent fsync in SELECT-only queries (Vadim)
19706 Make index creation use psort code, because it is now faster (Tom)
19707 Allow creation of sort temp tables > 1 Gig
19709 Source Tree Changes
19710 -------------------
19711 Fix for linux PPC compile
19712 New generic expression-tree-walker subroutine (Tom)
19713 Change form() to varargform() to prevent portability problems
19714 Improved range checking for large integers on Alphas
19715 Clean up #include in /include directory (Bruce)
19716 Add scripts for checking includes (Bruce)
19717 Remove un-needed #include's from *.c files (Bruce)
19718 Change #include's to use <> and "" as appropriate (Bruce)
19719 Enable Windows compilation of libpq
19720 Alpha spinlock fix from Uncle George <email>gatgul@voicenet.com</email>
19721 Overhaul of optimizer data structures (Tom)
19722 Fix to cygipc library (Yutaka Tanida)
19723 Allow pgsql to work on newer Cygwin snapshots (Dan)
19724 New catalog version number (Tom)
19726 Rename heap_replace to heap_update
19727 Update for QNX (Dr. Andreas Kardos)
19728 New platform-specific regression handling (Tom)
19729 Rename oid8 -> oidvector and int28 -> int2vector (Bruce)
19730 Included all yacc and lex files into the distribution (Peter E.)
19731 Remove lextest, no longer needed (Peter E)
19732 Fix for libpq and psql on Windows (Magnus)
19733 Internally change datetime and timespan into timestamp and interval (Thomas)
19734 Fix for plpgsql on BSD/OS
19735 Add SQL_ASCII test case to the regression test (Tatsuo)
19736 configure --with-mb now deprecated (Tatsuo)
19738 NetBSD fixes (Johnny C. Lam <email>lamj@stat.cmu.edu</email>)
19739 Fixes for Alpha compiles
19740 New multibyte encodings
19746 <sect1 id="release-6-5-3">
19747 <title>Release 6.5.3</title>
19750 <title>Release date</title>
19751 <simpara>1999-10-13</simpara>
19755 This is basically a cleanup release for 6.5.2. We have added a new
19756 <application>PgAccess</> that was missing in 6.5.2, and installed an NT-specific fix.
19761 <title>Migration to Version 6.5.3</title>
19764 A dump/restore is <emphasis>not</emphasis> required for those running
19769 <title>Changes</title>
19773 Updated version of pgaccess 0.98
19775 Fix dumping rules on inherited tables
19782 <sect1 id="release-6-5-2">
19783 <title>Release 6.5.2</title>
19786 <title>Release date</title>
19787 <simpara>1999-09-15</simpara>
19791 This is basically a cleanup release for 6.5.1. We have fixed a variety of
19792 problems reported by 6.5.1 users.
19797 <title>Migration to Version 6.5.2</title>
19800 A dump/restore is <emphasis>not</emphasis> required for those running
19806 <title>Changes</title>
19810 subselect+CASE fixes(Tom)
19811 Add SHLIB_LINK setting for solaris_i386 and solaris_sparc ports(Daren Sefcik)
19812 Fixes for CASE in WHERE join clauses(Tom)
19813 Fix BTScan abort(Tom)
19814 Repair the check for redundant UNIQUE and PRIMARY KEY indexes(Thomas)
19815 Improve it so that it checks for multicolumn constraints(Thomas)
19816 Fix for Windows making problem with MB enabled(Hiroki Kataoka)
19817 Allow BSD yacc and bison to compile pl code(Bruce)
19818 Fix SET NAMES working
19820 Fix vacuum's memory consumption(Hiroshi,Tatsuo)
19821 Reduce the total memory consumption of vacuum(Tom)
19822 Fix for timestamp(datetime)
19823 Rule deparsing bugfixes(Tom)
19824 Fix quoting problems in mkMakefile.tcldefs.sh.in and mkMakefile.tkdefs.sh.in(Tom)
19825 This is to re-use space on index pages freed by vacuum(Vadim)
19826 document -x for pg_dump(Bruce)
19827 Fix for unary operators in rule deparser(Tom)
19828 Comment out FileUnlink of excess segments during mdtruncate()(Tom)
19829 IRIX linking fix from Yu Cao >yucao@falcon.kla-tencor.com<
19830 Repair logic error in LIKE: should not return LIKE_ABORT
19831 when reach end of pattern before end of text(Tom)
19832 Repair incorrect cleanup of heap memory allocation during transaction abort(Tom)
19833 Updated version of pgaccess 0.98
19839 <sect1 id="release-6-5-1">
19840 <title>Release 6.5.1</title>
19843 <title>Release date</title>
19844 <simpara>1999-07-15</simpara>
19848 This is basically a cleanup release for 6.5. We have fixed a variety of
19849 problems reported by 6.5 users.
19853 <title>Migration to Version 6.5.1</title>
19856 A dump/restore is <emphasis>not</emphasis> required for those running
19862 <title>Changes</title>
19867 Portability fixes for linux_ppc, IRIX, linux_alpha, OpenBSD, alpha
19868 Remove QUERY_LIMIT, use SELECT...LIMIT
19869 Fix for EXPLAIN on inheritance(Tom)
19870 Patch to allow vacuum on multisegment tables(Hiroshi)
19871 R-Tree optimizer selectivity fix(Tom)
19872 ACL file descriptor leak fix(Atsushi Ogawa)
19873 New expression subtree code(Tom)
19874 Avoid disk writes for read-only transactions(Vadim)
19875 Fix for removal of temp tables if last transaction was aborted(Bruce)
19876 Fix to prevent too large row from being created(Bruce)
19878 Allow port numbers 32k - 64k(Bruce)
19879 Add ^ precedence(Bruce)
19880 Rename sort files called pg_temp to pg_sorttemp(Bruce)
19881 Fix for microseconds in time values(Tom)
19882 Tutorial source cleanup
19883 New linux_m68k port
19884 Fix for sorting of NULL's in some cases(Tom)
19885 Shared library dependencies fixed (Tom)
19886 Fixed glitches affecting GROUP BY in subselects(Tom)
19887 Fix some compiler warnings (Tomoaki Nishiyama)
19888 Add Win1250 (Czech) support (Pavel Behal)
19894 <sect1 id="release-6-5">
19895 <title>Release 6.5</title>
19898 <title>Release date</title>
19899 <simpara>1999-06-09</simpara>
19903 This release marks a major step in the development team's mastery of the source
19904 code we inherited from Berkeley. You will see we are now easily adding
19905 major features, thanks to the increasing size and experience of our
19906 world-wide development team.
19910 Here is a brief summary of the more notable changes:
19915 Multiversion concurrency control(MVCC)
19919 This removes our old table-level locking, and replaces it with
19920 a locking system that is superior to most commercial database
19921 systems. In a traditional system, each row that is modified
19922 is locked until committed, preventing reads by other users.
19923 MVCC uses the natural multiversion nature of
19924 <productname>PostgreSQL</productname> to allow readers to
19925 continue reading consistent data during writer activity.
19926 Writers continue to use the compact pg_log transaction system.
19927 This is all performed without having to allocate a lock for
19928 every row like traditional database systems. So, basically,
19929 we no longer are restricted by simple table-level locking; we
19930 have something better than row-level locking.
19937 Hot backups from <application>pg_dump</application>
19941 <application>pg_dump</application> takes advantage of the new
19942 MVCC features to give a consistent database dump/backup while
19943 the database stays online and available for queries.
19954 We now have a true numeric data type, with
19955 user-specified precision.
19966 Temporary tables are guaranteed to have unique names
19967 within a database session, and are destroyed on session exit.
19978 We now have CASE, INTERSECT, and EXCEPT statement
19979 support. We have new LIMIT/OFFSET, SET TRANSACTION ISOLATION LEVEL,
19980 SELECT ... FOR UPDATE, and an improved LOCK TABLE command.
19991 We continue to speed up <productname>PostgreSQL</productname>,
19992 thanks to the variety of talents within our team. We have
19993 sped up memory allocation, optimization, table joins, and row
20005 We continue to expand our port list, this time including
20006 <systemitem class="osname">Windows NT</>/<systemitem>ix86</> and <systemitem class="osname">NetBSD</>/<systemitem>arm32</>.
20017 Most interfaces have new versions, and existing functionality
20029 New and updated material is present throughout the
20030 documentation. New <acronym>FAQ</acronym>s have been
20031 contributed for <systemitem class="osname">SGI</> and <systemitem class="osname">AIX</> platforms.
20032 The <citetitle>Tutorial</citetitle> has introductory information
20033 on <acronym>SQL</acronym> from Stefan Simkovics.
20034 For the <citetitle>User's Guide</citetitle>, there are
20035 reference pages covering the postmaster and more utility
20036 programs, and a new appendix
20037 contains details on date/time behavior.
20038 The <citetitle>Administrator's Guide</citetitle> has a new
20039 chapter on troubleshooting from Tom Lane.
20040 And the <citetitle>Programmer's Guide</citetitle> has a
20041 description of query processing, also from Stefan, and details
20042 on obtaining the <productname>PostgreSQL</productname> source
20043 tree via anonymous <productname>CVS</productname> and
20044 <productname>CVSup</productname>.
20052 <title>Migration to Version 6.5</title>
20055 A dump/restore using <application>pg_dump</application>
20056 is required for those wishing to migrate data from any
20057 previous release of <productname>PostgreSQL</productname>.
20058 <application>pg_upgrade</application> can <emphasis>not</emphasis>
20059 be used to upgrade to this release because the on-disk structure
20060 of the tables has changed compared to previous releases.
20064 The new Multiversion Concurrency Control (MVCC) features can
20065 give somewhat different behaviors in multiuser
20066 environments. <emphasis>Read and understand the following section
20067 to ensure that your existing applications will give you the
20068 behavior you need.</emphasis>
20072 <title>Multiversion Concurrency Control</title>
20075 Because readers in 6.5 don't lock data, regardless of transaction
20076 isolation level, data read by one transaction can be overwritten by
20077 another. In other words, if a row is returned by
20078 <command>SELECT</command> it doesn't mean that this row really exists
20079 at the time it is returned (i.e. sometime after the statement or
20080 transaction began) nor that the row is protected from being deleted or
20081 updated by concurrent transactions before the current transaction does
20082 a commit or rollback.
20086 To ensure the actual existence of a row and protect it against
20087 concurrent updates one must use <command>SELECT FOR UPDATE</command> or
20088 an appropriate <command>LOCK TABLE</command> statement. This should be
20089 taken into account when porting applications from previous releases of
20090 <productname>PostgreSQL</productname> and other environments.
20094 Keep the above in mind if you are using
20095 <filename>contrib/refint.*</filename> triggers for
20096 referential integrity. Additional techniques are required now. One way is
20097 to use <command>LOCK parent_table IN SHARE ROW EXCLUSIVE MODE</command>
20098 command if a transaction is going to update/delete a primary key and
20099 use <command>LOCK parent_table IN SHARE MODE</command> command if a
20100 transaction is going to update/insert a foreign key.
20104 Note that if you run a transaction in SERIALIZABLE mode then you must
20105 execute the <command>LOCK</command> commands above before execution of any
20106 <acronym>DML</acronym> statement
20107 (<command>SELECT/INSERT/DELETE/UPDATE/FETCH/COPY_TO</command>) in the
20114 These inconveniences will disappear in the future
20115 when the ability to read dirty
20116 (uncommitted) data (regardless of isolation level) and true referential
20117 integrity will be implemented.
20123 <title>Changes</title>
20129 Fix text<->float8 and text<->float4 conversion functions(Thomas)
20130 Fix for creating tables with mixed-case constraints(Billy)
20131 Change exp()/pow() behavior to generate error on underflow/overflow(Jan)
20132 Fix bug in pg_dump -z
20133 Memory overrun cleanups(Tatsuo)
20134 Fix for lo_import crash(Tatsuo)
20135 Adjust handling of data type names to suppress double quotes(Thomas)
20136 Use type coercion for matching columns and DEFAULT(Thomas)
20137 Fix deadlock so it only checks once after one second of sleep(Bruce)
20138 Fixes for aggregates and PL/pgsql(Hiroshi)
20139 Fix for subquery crash(Vadim)
20140 Fix for libpq function PQfnumber and case-insensitive names(Bahman Rafatjoo)
20141 Fix for large object write-in-middle, no extra block, memory consumption(Tatsuo)
20142 Fix for pg_dump -d or -D and quote special characters in INSERT
20143 Repair serious problems with dynahash(Tom)
20144 Fix INET/CIDR portability problems
20145 Fix problem with selectivity error in ALTER TABLE ADD COLUMN(Bruce)
20146 Fix executor so mergejoin of different column types works(Tom)
20147 Fix for Alpha OR selectivity bug
20148 Fix OR index selectivity problem(Bruce)
20149 Fix so \d shows proper length for char()/varchar()(Ryan)
20150 Fix tutorial code(Clark)
20151 Improve destroyuser checking(Oliver)
20152 Fix for Kerberos(Rodney McDuff)
20153 Fix for dropping database while dirty buffers(Bruce)
20154 Fix so sequence nextval() can be case-sensitive(Bruce)
20156 Drop buffers before destroying database files(Bruce)
20157 Fix case where executor evaluates functions twice(Tatsuo)
20158 Allow sequence nextval actions to be case-sensitive(Bruce)
20159 Fix optimizer indexing not working for negative numbers(Bruce)
20160 Fix for memory leak in executor with fjIsNull
20161 Fix for aggregate memory leaks(Erik Riedel)
20162 Allow user name containing a dash to grant privileges
20163 Cleanup of NULL in inet types
20164 Clean up system table bugs(Tom)
20165 Fix problems of PAGER and \? command(Masaaki Sakaida)
20166 Reduce default multisegment file size limit to 1GB(Peter)
20167 Fix for dumping of CREATE OPERATOR(Tom)
20168 Fix for backward scanning of cursors(Hiroshi Inoue)
20169 Fix for COPY FROM STDIN when using \i(Tom)
20170 Fix for subselect is compared inside an expression(Jan)
20171 Fix handling of error reporting while returning rows(Tom)
20172 Fix problems with reference to array types(Tom,Jan)
20173 Prevent UPDATE SET oid(Jan)
20174 Fix pg_dump so -t option can handle case-sensitive tablenames
20175 Fixes for GROUP BY in special cases(Tom, Jan)
20176 Fix for memory leak in failed queries(Tom)
20177 DEFAULT now supports mixed-case identifiers(Tom)
20178 Fix for multisegment uses of DROP/RENAME table, indexes(Ole Gjerde)
20179 Disable use of pg_dump with both -o and -d options(Bruce)
20180 Allow pg_dump to properly dump group privileges(Bruce)
20181 Fix GROUP BY in INSERT INTO table SELECT * FROM table2(Jan)
20182 Fix for computations in views(Jan)
20183 Fix for aggregates on array indexes(Tom)
20184 Fix for DEFAULT handles single quotes in value requiring too many quotes
20185 Fix security problem with non-super users importing/exporting large objects(Tom)
20186 Rollback of transaction that creates table cleaned up properly(Tom)
20187 Fix to allow long table and column names to generate proper serial names(Tom)
20191 Add "vacuumdb" utility
20192 Speed up libpq by allocating memory better(Tom)
20193 EXPLAIN all indexes used(Tom)
20194 Implement CASE, COALESCE, NULLIF expression(Thomas)
20195 New pg_dump table output format(Constantin)
20196 Add string min()/max() functions(Thomas)
20197 Extend new type coercion techniques to aggregates(Thomas)
20198 New moddatetime contrib(Terry)
20199 Update to pgaccess 0.96(Constantin)
20200 Add routines for single-byte "char" type(Thomas)
20201 Improved substr() function(Thomas)
20202 Improved multibyte handling(Tatsuo)
20203 Multiversion concurrency control/MVCC(Vadim)
20204 New Serialized mode(Vadim)
20205 Fix for tables over 2gigs(Peter)
20206 New SET TRANSACTION ISOLATION LEVEL(Vadim)
20207 New LOCK TABLE IN ... MODE(Vadim)
20208 Update ODBC driver(Byron)
20209 New NUMERIC data type(Jan)
20210 New SELECT FOR UPDATE(Vadim)
20211 Handle "NaN" and "Infinity" for input values(Jan)
20212 Improved date/year handling(Thomas)
20213 Improved handling of backend connections(Magnus)
20214 New options ELOG_TIMESTAMPS and USE_SYSLOG options for log files(Massimo)
20215 New TCL_ARRAYS option(Massimo)
20216 New INTERSECT and EXCEPT(Stefan)
20217 New pg_index.indisprimary for primary key tracking(D'Arcy)
20218 New pg_dump option to allow dropping of tables before creation(Brook)
20219 Speedup of row output routines(Tom)
20220 New READ COMMITTED isolation level(Vadim)
20221 New TEMP tables/indexes(Bruce)
20222 Prevent sorting if result is already sorted(Jan)
20223 New memory allocation optimization(Jan)
20224 Allow psql to do \p\g(Bruce)
20225 Allow multiple rule actions(Jan)
20226 Added LIMIT/OFFSET functionality(Jan)
20227 Improve optimizer when joining a large number of tables(Bruce)
20228 New intro to SQL from S. Simkovics' Master's Thesis (Stefan, Thomas)
20229 New intro to backend processing from S. Simkovics' Master's Thesis (Stefan)
20230 Improved int8 support(Ryan Bradetich, Thomas, Tom)
20231 New routines to convert between int8 and text/varchar types(Thomas)
20232 New bushy plans, where meta-tables are joined(Bruce)
20233 Enable right-hand queries by default(Bruce)
20234 Allow reliable maximum number of backends to be set at configure time
20235 (--with-maxbackends and postmaster switch (-N backends))(Tom)
20236 GEQO default now 10 tables because of optimizer speedups(Tom)
20237 Allow NULL=Var for MS-SQL portability(Michael, Bruce)
20238 Modify contrib check_primary_key() so either "automatic" or "dependent"(Anand)
20239 Allow psql \d on a view show query(Ryan)
20240 Speedup for LIKE(Bruce)
20241 Ecpg fixes/features, see src/interfaces/ecpg/ChangeLog file(Michael)
20242 JDBC fixes/features, see src/interfaces/jdbc/CHANGELOG(Peter)
20243 Make % operator have precedence like /(Bruce)
20244 Add new postgres -O option to allow system table structure changes(Bruce)
20245 Update contrib/pginterface/findoidjoins script(Tom)
20246 Major speedup in vacuum of deleted rows with indexes(Vadim)
20247 Allow non-SQL functions to run different versions based on arguments(Tom)
20248 Add -E option that shows actual queries sent by \dt and friends(Masaaki Sakaida)
20249 Add version number in start-up banners for psql(Masaaki Sakaida)
20250 New contrib/vacuumlo removes large objects not referenced(Peter)
20251 New initialization for table sizes so non-vacuumed tables perform better(Tom)
20252 Improve error messages when a connection is rejected(Tom)
20253 Support for arrays of char() and varchar() fields(Massimo)
20254 Overhaul of hash code to increase reliability and performance(Tom)
20255 Update to PyGreSQL 2.4(D'Arcy)
20256 Changed debug options so -d4 and -d5 produce different node displays(Jan)
20257 New pg_options: pretty_plan, pretty_parse, pretty_rewritten(Jan)
20258 Better optimization statistics for system table access(Tom)
20259 Better handling of non-default block sizes(Massimo)
20260 Improve GEQO optimizer memory consumption(Tom)
20261 UNION now supports ORDER BY of columns not in target list(Jan)
20262 Major libpq++ improvements(Vince Vielhaber)
20263 pg_dump now uses -z(ACL's) as default(Bruce)
20264 backend cache, memory speedups(Tom)
20265 have pg_dump do everything in one snapshot transaction(Vadim)
20266 fix for large object memory leakage, fix for pg_dumping(Tom)
20267 INET type now respects netmask for comparisons
20268 Make VACUUM ANALYZE only use a readlock(Vadim)
20269 Allow VIEWs on UNIONS(Jan)
20270 pg_dump now can generate consistent snapshots on active databases(Vadim)
20272 Source Tree Changes
20273 -------------------
20274 Improve port matching(Tom)
20275 Portability fixes for SunOS
20276 Add Windows NT backend port and enable dynamic loading(Magnus and Daniel Horak)
20277 New port to Cobalt Qube(Mips) running Linux(Tatsuo)
20278 Port to NetBSD/m68k(Mr. Mutsuki Nakajima)
20279 Port to NetBSD/sun3(Mr. Mutsuki Nakajima)
20280 Port to NetBSD/macppc(Toshimi Aoki)
20281 Fix for tcl/tk configuration(Vince)
20282 Removed CURRENT key word for rule queries(Jan)
20283 NT dynamic loading now works(Daniel Horak)
20284 Add ARM32 support(Andrew McMurry)
20285 Better support for HP-UX 11 and UnixWare
20286 Improve file handling to be more uniform, prevent file descriptor leak(Tom)
20287 New install commands for plpgsql(Jan)
20294 <sect1 id="release-6-4-2">
20295 <title>Release 6.4.2</title>
20298 <title>Release date</title>
20299 <simpara>1998-12-20</simpara>
20303 The 6.4.1 release was improperly packaged. This also has one additional
20309 <title>Migration to Version 6.4.2</title>
20312 A dump/restore is <emphasis>not</emphasis> required for those running
20317 <title>Changes</title>
20321 Fix for datetime constant problem on some platforms(Thomas)
20329 <sect1 id="release-6-4-1">
20330 <title>Release 6.4.1</title>
20333 <title>Release date</title>
20334 <simpara>1998-12-18</simpara>
20338 This is basically a cleanup release for 6.4. We have fixed a variety of
20339 problems reported by 6.4 users.
20344 <title>Migration to Version 6.4.1</title>
20347 A dump/restore is <emphasis>not</emphasis> required for those running
20352 <title>Changes</title>
20356 Add pg_dump -N flag to force double quotes around identifiers. This is
20357 the default(Thomas)
20358 Fix for NOT in where clause causing crash(Bruce)
20359 EXPLAIN VERBOSE coredump fix(Vadim)
20360 Fix shared-library problems on Linux
20361 Fix test for table existence to allow mixed-case and whitespace in
20362 the table name(Thomas)
20363 Fix a couple of pg_dump bugs
20364 Configure matches template/.similar entries better(Tom)
20365 Change builtin function names from SPI_* to spi_*
20366 OR WHERE clause fix(Vadim)
20367 Fixes for mixed-case table names(Billy)
20368 contrib/linux/postgres.init.csh/sh fix(Thomas)
20369 libpq memory overrun fix
20371 Change exp() behavior to generate error on underflow(Thomas)
20372 pg_dump fixes for memory leak, inheritance constraints, layout change
20373 update pgaccess to 0.93
20374 Fix prototype for 64-bit platforms
20375 Multibyte fixes(Tatsuo)
20377 Fix memory overruns(Tatsuo)
20378 Fix for lo_import() crash(Bruce)
20379 Better search for install program(Tom)
20380 Timezone fixes(Tom)
20382 Use implicit type coercion for matching DEFAULT values(Thomas)
20383 Add routines to help with single-byte (internal) character type(Thomas)
20384 Compilation of libpq for Windows fixes(Magnus)
20385 Upgrade to PyGreSQL 2.2(D'Arcy)
20393 <sect1 id="release-6-4">
20394 <title>Release 6.4</title>
20397 <title>Release date</title>
20398 <simpara>1998-10-30</simpara>
20402 There are <emphasis>many</emphasis> new features and improvements in this release.
20403 Thanks to our developers and maintainers, nearly every aspect of the system
20404 has received some attention since the previous release.
20405 Here is a brief, incomplete summary:
20410 Views and rules are now functional thanks to extensive new code in the
20411 rewrite rules system from Jan Wieck. He also wrote a chapter on it
20412 for the <citetitle>Programmer's Guide</citetitle>.
20417 Jan also contributed a second procedural language, <application>PL/pgSQL</application>, to go with the
20418 original <application>PL/pgTCL</application> procedural language he contributed last release.
20424 We have optional multiple-byte character set support from Tatsuo Ishii
20425 to complement our existing locale support.
20431 Client/server communications has been cleaned up, with better support for
20432 asynchronous messages and interrupts thanks to Tom Lane.
20438 The parser will now perform automatic type coercion to match arguments
20439 to available operators and functions, and to match columns and expressions
20440 with target columns. This uses a generic mechanism which supports
20441 the type extensibility features of <productname>PostgreSQL</productname>.
20442 There is a new chapter in the <citetitle>User's Guide</citetitle>
20443 which covers this topic.
20449 Three new data types have been added.
20450 Two types, <type>inet</type> and <type>cidr</type>, support various forms
20451 of IP network, subnet, and machine addressing. There is now an 8-byte integer
20452 type available on some platforms. See the chapter on data types
20453 in the <citetitle>User's Guide</citetitle> for details.
20454 A fourth type, <type>serial</type>, is now supported by the parser as an
20455 amalgam of the <type>int4</type> type, a sequence, and a unique index.
20461 Several more <acronym>SQL92</acronym>-compatible syntax features have been
20462 added, including <command>INSERT DEFAULT VALUES</command>
20468 The automatic configuration and installation system has received some
20469 attention, and should be more robust for more platforms than it has ever
20478 <title>Migration to Version 6.4</title>
20481 A dump/restore using <application>pg_dump</application>
20482 or <application>pg_dumpall</application>
20483 is required for those wishing to migrate data from any
20484 previous release of <productname>PostgreSQL</productname>.
20489 <title>Changes</title>
20495 Fix for a tiny memory leak in PQsetdb/PQfinish(Bryan)
20496 Remove char2-16 data types, use char/varchar(Darren)
20497 Pqfn not handles a NOTICE message(Anders)
20498 Reduced busywaiting overhead for spinlocks with many backends (dg)
20499 Stuck spinlock detection (dg)
20500 Fix up "ISO-style" timespan decoding and encoding(Thomas)
20501 Fix problem with table drop after rollback of transaction(Vadim)
20502 Change error message and remove non-functional update message(Vadim)
20503 Fix for COPY array checking
20504 Fix for SELECT 1 UNION SELECT NULL
20505 Fix for buffer leaks in large object calls(Pascal)
20506 Change owner from oid to int4 type(Bruce)
20507 Fix a bug in the oracle compatibility functions btrim() ltrim() and rtrim()
20508 Fix for shared invalidation cache overflow(Massimo)
20509 Prevent file descriptor leaks in failed COPY's(Bruce)
20510 Fix memory leak in libpgtcl's pg_select(Constantin)
20511 Fix problems with username/passwords over 8 characters(Tom)
20512 Fix problems with handling of asynchronous NOTIFY in backend(Tom)
20513 Fix of many bad system table entries(Tom)
20517 Upgrade ecpg and ecpglib,see src/interfaces/ecpc/ChangeLog(Michael)
20518 Show the index used in an EXPLAIN(Zeugswetter)
20519 EXPLAIN invokes rule system and shows plan(s) for rewritten queries(Jan)
20520 Multibyte awareness of many data types and functions, via configure(Tatsuo)
20521 New configure --with-mb option(Tatsuo)
20522 New initdb --pgencoding option(Tatsuo)
20523 New createdb -E multibyte option(Tatsuo)
20524 Select version(); now returns PostgreSQL version(Jeroen)
20525 libpq now allows asynchronous clients(Tom)
20526 Allow cancel from client of backend query(Tom)
20527 psql now cancels query with Control-C(Tom)
20528 libpq users need not issue dummy queries to get NOTIFY messages(Tom)
20529 NOTIFY now sends sender's PID, so you can tell whether it was your own(Tom)
20530 PGresult struct now includes associated error message, if any(Tom)
20531 Define "tz_hour" and "tz_minute" arguments to date_part()(Thomas)
20532 Add routines to convert between varchar and bpchar(Thomas)
20533 Add routines to allow sizing of varchar and bpchar into target columns(Thomas)
20534 Add bit flags to support timezonehour and minute in data retrieval(Thomas)
20535 Allow more variations on valid floating point numbers (e.g. ".1", "1e6")(Thomas)
20536 Fixes for unary minus parsing with leading spaces(Thomas)
20537 Implement TIMEZONE_HOUR, TIMEZONE_MINUTE per SQL92 specs(Thomas)
20538 Check for and properly ignore FOREIGN KEY column constraints(Thomas)
20539 Define USER as synonym for CURRENT_USER per SQL92 specs(Thomas)
20540 Enable HAVING clause but no fixes elsewhere yet.
20541 Make "char" type a synonym for "char(1)" (actually implemented as bpchar)(Thomas)
20542 Save string type if specified for DEFAULT clause handling(Thomas)
20543 Coerce operations involving different data types(Thomas)
20544 Allow some index use for columns of different types(Thomas)
20545 Add capabilities for automatic type conversion(Thomas)
20546 Cleanups for large objects, so file is truncated on open(Peter)
20547 Readline cleanups(Tom)
20548 Allow psql \f \ to make spaces as delimiter(Bruce)
20549 Pass pg_attribute.atttypmod to the frontend for column field lengths(Tom,Bruce)
20550 Msql compatibility library in /contrib(Aldrin)
20551 Remove the requirement that ORDER/GROUP BY clause identifiers be
20552 included in the target list(David)
20553 Convert columns to match columns in UNION clauses(Thomas)
20554 Remove fork()/exec() and only do fork()(Bruce)
20555 Jdbc cleanups(Peter)
20556 Show backend status on ps command line(only works on some platforms)(Bruce)
20557 Pg_hba.conf now has a sameuser option in the database field
20558 Make lo_unlink take oid param, not int4
20559 New DISABLE_COMPLEX_MACRO for compilers that cannot handle our macros(Bruce)
20560 Libpgtcl now handles NOTIFY as a Tcl event, need not send dummy queries(Tom)
20561 libpgtcl cleanups(Tom)
20562 Add -error option to libpgtcl's pg_result command(Tom)
20563 New locale patch, see docs/README/locale(Oleg)
20564 Fix for pg_dump so CONSTRAINT and CHECK syntax is correct(ccb)
20565 New contrib/lo code for large object orphan removal(Peter)
20566 New psql command "SET CLIENT_ENCODING TO 'encoding'" for multibytes
20567 feature, see /doc/README.mb(Tatsuo)
20568 contrib/noupdate code to revoke update permission on a column
20569 libpq can now be compiled on Windows(Magnus)
20570 Add PQsetdbLogin() in libpq
20571 New 8-byte integer type, checked by configure for OS support(Thomas)
20572 Better support for quoted table/column names(Thomas)
20573 Surround table and column names with double-quotes in pg_dump(Thomas)
20574 PQreset() now works with passwords(Tom)
20575 Handle case of GROUP BY target list column number out of range(David)
20576 Allow UNION in subselects
20577 Add auto-size to screen to \d? commands(Bruce)
20578 Use UNION to show all \d? results in one query(Bruce)
20579 Add \d? field search feature(Bruce)
20580 Pg_dump issues fewer \connect requests(Tom)
20581 Make pg_dump -z flag work better, document it in manual page(Tom)
20582 Add HAVING clause with full support for subselects and unions(Stephan)
20583 Full text indexing routines in contrib/fulltextindex(Maarten)
20584 Transaction ids now stored in shared memory(Vadim)
20585 New PGCLIENTENCODING when issuing COPY command(Tatsuo)
20586 Support for SQL92 syntax "SET NAMES"(Tatsuo)
20587 Support for LATIN2-5(Tatsuo)
20588 Add UNICODE regression test case(Tatsuo)
20589 Lock manager cleanup, new locking modes for LLL(Vadim)
20590 Allow index use with OR clauses(Bruce)
20591 Allows "SELECT NULL ORDER BY 1;"
20592 Explain VERBOSE prints the plan, and now pretty-prints the plan to
20593 the postmaster log file(Bruce)
20594 Add indexes display to \d command(Bruce)
20595 Allow GROUP BY on functions(David)
20596 New pg_class.relkind for large objects(Bruce)
20597 New way to send libpq NOTICE messages to a different location(Tom)
20598 New \w write command to psql(Bruce)
20599 New /contrib/findoidjoins scans oid columns to find join relationships(Bruce)
20600 Allow binary-compatible indexes to be considered when checking for valid
20601 Indexes for restriction clauses containing a constant(Thomas)
20602 New ISBN/ISSN code in /contrib/isbn_issn
20603 Allow NOT LIKE, IN, NOT IN, BETWEEN, and NOT BETWEEN constraint(Thomas)
20604 New rewrite system fixes many problems with rules and views(Jan)
20605 * Rules on relations work
20606 * Event qualifications on insert/update/delete work
20607 * New OLD variable to reference CURRENT, CURRENT will be remove in future
20608 * Update rules can reference NEW and OLD in rule qualifications/actions
20609 * Insert/update/delete rules on views work
20610 * Multiple rule actions are now supported, surrounded by parentheses
20611 * Regular users can create views/rules on tables they have RULE permits
20612 * Rules and views inherit the privileges of the creator
20613 * No rules at the column level
20614 * No UPDATE NEW/OLD rules
20615 * New pg_tables, pg_indexes, pg_rules and pg_views system views
20616 * Only a single action on SELECT rules
20617 * Total rewrite overhaul, perhaps for 6.5
20618 * handle subselects
20619 * handle aggregates on views
20620 * handle insert into select from view works
20621 System indexes are now multikey(Bruce)
20622 Oidint2, oidint4, and oidname types are removed(Bruce)
20623 Use system cache for more system table lookups(Bruce)
20624 New backend programming language PL/pgSQL in backend/pl(Jan)
20625 New SERIAL data type, auto-creates sequence/index(Thomas)
20626 Enable assert checking without a recompile(Massimo)
20627 User lock enhancements(Massimo)
20628 New setval() command to set sequence value(Massimo)
20629 Auto-remove unix socket file on start-up if no postmaster running(Massimo)
20630 Conditional trace package(Massimo)
20631 New UNLISTEN command(Massimo)
20632 psql and libpq now compile under Windows using win32.mak(Magnus)
20633 Lo_read no longer stores trailing NULL(Bruce)
20634 Identifiers are now truncated to 31 characters internally(Bruce)
20635 Createuser options now available on the command line
20636 Code for 64-bit integer supported added, configure tested, int8 type(Thomas)
20637 Prevent file descriptor leaf from failed COPY(Bruce)
20638 New pg_upgrade command(Bruce)
20639 Updated /contrib directories(Massimo)
20640 New CREATE TABLE DEFAULT VALUES statement available(Thomas)
20641 New INSERT INTO TABLE DEFAULT VALUES statement available(Thomas)
20642 New DECLARE and FETCH feature(Thomas)
20643 libpq's internal structures now not exported(Tom)
20644 Allow up to 8 key indexes(Bruce)
20645 Remove ARCHIVE key word, that is no longer used(Thomas)
20646 pg_dump -n flag to suppress quotes around indentifiers
20647 disable system columns for views(Jan)
20648 new INET and CIDR types for network addresses(TomH, Paul)
20649 no more double quotes in psql output
20650 pg_dump now dumps views(Terry)
20651 new SET QUERY_LIMIT(Tatsuo,Jan)
20653 Source Tree Changes
20654 -------------------
20655 /contrib cleanup(Jun)
20656 Inline some small functions called for every row(Bruce)
20658 HP-UX cleanups(Tom)
20659 Multibyte regression tests(Soonmyung.)
20660 Remove --disabled options from configure
20661 Define PGDOC to use POSTGRESDIR by default
20662 Make regression optional
20663 Remove extra braces code to pgindent(Bruce)
20664 Add bsdi shared library support(Bruce)
20665 New --without-CXX support configure option(Brook)
20667 Update backend flowchart in tools/backend(Bruce)
20668 Change atttypmod from int16 to int32(Bruce, Tom)
20669 Getrusage() fix for platforms that do not have it(Tom)
20670 Add PQconnectdb, PGUSER, PGPASSWORD to libpq man page
20671 NS32K platform fixes(Phil Nelson, John Buller)
20672 SCO 7/UnixWare 2.x fixes(Billy,others)
20673 Sparc/Solaris 2.5 fixes(Ryan)
20674 Pgbuiltin.3 is obsolete, move to doc files(Thomas)
20675 Even more documentation(Thomas)
20676 Nextstep support(Jacek)
20678 pginterface manual page(Bruce)
20679 shared libraries all have version numbers
20680 merged all OS-specific shared library defines into one file
20681 smarter TCL/TK configuration checking(Billy)
20682 smarter perl configuration(Brook)
20683 configure uses supplied install-sh if no install script found(Tom)
20684 new Makefile.shlib for shared library configuration(Tom)
20690 <sect1 id="release-6-3-2">
20691 <title>Release 6.3.2</title>
20694 <title>Release date</title>
20695 <simpara>1998-04-07</simpara>
20699 This is a bug-fix release for 6.3.x.
20700 Refer to the release notes for version 6.3 for a more complete summary of new features.
20708 Repairs automatic configuration support for some platforms, including Linux,
20709 from breakage inadvertently introduced in version 6.3.1.
20715 Correctly handles function calls on the left side of BETWEEN and LIKE clauses.
20722 A dump/restore is NOT required for those running 6.3 or 6.3.1. A
20723 <literal>make distclean</>, <literal>make</>, and <literal>make install</> is all that is required.
20724 This last step should be performed while the postmaster is not running.
20725 You should re-link any custom applications that use <productname>PostgreSQL</productname> libraries.
20728 For upgrades from pre-6.3 installations,
20729 refer to the installation and migration instructions for version 6.3.
20733 <title>Changes</title>
20737 Configure detection improvements for tcl/tk(Brook Milligan, Alvin)
20738 Manual page improvements(Bruce)
20739 BETWEEN and LIKE fix(Thomas)
20740 fix for psql \connect used by pg_dump(Oliver Elphick)
20742 pgaccess, version 0.86
20743 qsort removed, now uses libc version, cleanups(Jeroen)
20744 fix for buffer over-runs detected(Maurice Gittens)
20745 fix for buffer overrun in libpgtcl(Randy Kunkee)
20746 fix for UNION with DISTINCT or ORDER BY(Bruce)
20747 gettimeofday configure check(Doug Winterburn)
20748 Fix "indexes not used" bug(Vadim)
20749 docs additions(Thomas)
20750 Fix for backend memory leak(Bruce)
20751 libreadline cleanup(Erwan MAS)
20752 Remove DISTDIR(Bruce)
20753 Makefile dependency cleanup(Jeroen van Vianen)
20754 ASSERT fixes(Bruce)
20760 <sect1 id="release-6-3-1">
20761 <title>Release 6.3.1</title>
20764 <title>Release date</title>
20765 <simpara>1998-03-23</simpara>
20774 Additional support for multibyte character sets.
20780 Repair byte ordering for mixed-endian clients and servers.
20786 Minor updates to allowed SQL syntax.
20792 Improvements to the configuration autodetection for installation.
20799 A dump/restore is NOT required for those running 6.3. A
20800 <literal>make distclean</>, <literal>make</>, and <literal>make install</> is all that is required.
20801 This last step should be performed while the postmaster is not running.
20802 You should re-link any custom applications that use <productname>PostgreSQL</productname> libraries.
20805 For upgrades from pre-6.3 installations,
20806 refer to the installation and migration instructions for version 6.3.
20810 <title>Changes</title>
20814 ecpg cleanup/fixes, now version 1.1(Michael Meskes)
20815 pg_user cleanup(Bruce)
20816 large object fix for pg_dump and tclsh (alvin)
20817 LIKE fix for multiple adjacent underscores
20818 fix for redefining builtin functions(Thomas)
20820 upgrade to pg_access 0.83
20821 updated CLUSTER manual page
20822 multibyte character set support, see doc/README.mb(Tatsuo)
20823 configure --with-pgport fix
20825 big-endian fix for backend communications(Kataoka)
20826 SUBSTR() and substring() fix(Jan)
20827 several jdbc fixes(Peter)
20828 libpgtcl improvements, see libptcl/README(Randy Kunkee)
20829 Fix for "Datasize = 0" error(Vadim)
20830 Prevent \do from wrapping(Bruce)
20831 Remove duplicate Russian character set entries
20833 Allow optional TABLE key word in LOCK and SELECT INTO(Thomas)
20834 CREATE SEQUENCE options to allow a negative integer(Thomas)
20835 Add "PASSWORD" as an allowed column identifier(Thomas)
20836 Add checks for UNION target fields(Bruce)
20837 Fix Alpha port(Dwayne Bailey)
20838 Fix for text arrays containing quotes(Doug Gibson)
20839 Solaris compile fix(Albert Chin-A-Young)
20840 Better identify tcl and tk libs and includes(Bruce)
20846 <sect1 id="release-6-3">
20847 <title>Release 6.3</title>
20850 <title>Release date</title>
20851 <simpara>1998-03-01</simpara>
20855 There are <emphasis>many</emphasis> new features and improvements in this release.
20856 Here is a brief, incomplete summary:
20861 Many new SQL features, including
20862 full <acronym>SQL92</acronym> subselect capability
20863 (everything is here but target-list subselects).
20869 Support for client-side environment variables to specify time zone and date style.
20875 Socket interface for client/server connection. This is the default now
20876 so you might need to start <application>postmaster</application> with the
20877 <option>-i</option> flag.
20883 Better password authorization mechanisms. Default table privileges have changed.
20889 Old-style <firstterm>time travel</firstterm>
20890 has been removed. Performance has been improved.
20898 Bruce Momjian wrote the following notes to introduce the new release.
20903 There are some general 6.3 issues that I want to mention. These are
20904 only the big items that cannot be described in one sentence. A review
20905 of the detailed changes list is still needed.
20908 First, we now have subselects. Now that we have them, I would like to
20909 mention that without subselects, SQL is a very limited language.
20910 Subselects are a major feature, and you should review your code for
20911 places where subselects provide a better solution for your queries. I
20912 think you will find that there are more uses for subselects than you might
20913 think. Vadim has put us on the big SQL map with subselects, and fully
20914 functional ones too. The only thing you cannot do with subselects is to
20915 use them in the target list.
20918 Second, 6.3 uses Unix domain sockets rather than TCP/IP by default. To
20919 enable connections from other machines, you have to use the new
20920 postmaster -i option, and of course edit <filename>pg_hba.conf</filename>. Also, for this
20921 reason, the format of <filename>pg_hba.conf</filename> has changed.
20924 Third, <type>char()</type> fields will now allow faster access than <type>varchar()</type> or
20925 <type>text</type>. Specifically, the <type>text</> and <type>varchar()</type> have a penalty for access to
20926 any columns after the first column of this type. <type>char()</type> used to also
20927 have this access penalty, but it no longer does. This might suggest that
20928 you redesign some of your tables, especially if you have short character
20929 columns that you have defined as <type>varchar()</type> or <type>text</type>. This and other
20930 changes make 6.3 even faster than earlier releases.
20933 We now have passwords definable independent of any Unix file. There are
20934 new SQL USER commands.
20935 See the <citetitle>Administrator's Guide</citetitle> for more
20936 information. There is a new table, pg_shadow, which is used to store
20937 user information and user passwords, and it by default only SELECT-able
20938 by the <systemitem>postgres</systemitem> super-user. pg_user is now a view of pg_shadow, and is
20939 SELECT-able by PUBLIC. You should keep using pg_user in your
20940 application without changes.
20943 User-created tables now no longer have SELECT privilege to PUBLIC by
20944 default. This was done because the ANSI standard requires it. You can
20945 of course GRANT any privileges you want after the table is created.
20946 System tables continue to be SELECT-able by PUBLIC.
20949 We also have real deadlock detection code. No more sixty-second
20950 timeouts. And the new locking code implements a <acronym>FIFO</acronym> better, so there
20951 should be less resource starvation during heavy use.
20954 Many complaints have been made about inadequate documentation in previous
20955 releases. Thomas has put much effort into many new manuals for this
20956 release. Check out the doc/ directory.
20959 For performance reasons, time travel is gone, but can be implemented
20960 using triggers (see <filename>pgsql/contrib/spi/README</filename>). Please check out the new
20961 \d command for types, operators, etc. Also, views have their own
20962 privileges now, not based on the underlying tables, so privileges on
20963 them have to be set separately. Check <filename>/pgsql/interfaces</filename> for some new
20964 ways to talk to <productname>PostgreSQL</productname>.
20967 This is the first release that really required an explanation for
20968 existing users. In many ways, this was necessary because the new
20969 release removes many limitations, and the work-arounds people were using
20970 are no longer needed.
20974 <title>Migration to Version 6.3</title>
20977 A dump/restore using <application>pg_dump</application>
20978 or <application>pg_dumpall</application>
20979 is required for those wishing to migrate data from any
20980 previous release of <productname>PostgreSQL</productname>.
20985 <title>Changes</title>
20991 Fix binary cursors broken by MOVE implementation(Vadim)
20992 Fix for tcl library crash(Jan)
20993 Fix for array handling, from Gerhard Hintermayer
20994 Fix acl error, and remove duplicate pqtrace(Bruce)
20995 Fix psql \e for empty file(Bruce)
20996 Fix for textcat on varchar() fields(Bruce)
20997 Fix for DBT Sendproc (Zeugswetter Andres)
20998 Fix vacuum analyze syntax problem(Bruce)
20999 Fix for international identifiers(Tatsuo)
21000 Fix aggregates on inherited tables(Bruce)
21001 Fix substr() for out-of-bounds data
21002 Fix for select 1=1 or 2=2, select 1=1 and 2=2, and select sum(2+2)(Bruce)
21003 Fix notty output to show status result. -q option still turns it off(Bruce)
21004 Fix for count(*), aggs with views and multiple tables and sum(3)(Bruce)
21006 Fix for PQtrace start/stop several times(Bruce)
21007 Fix a variety of locking problems like newer lock waiters getting
21008 lock before older waiters, and having readlock people not share
21009 locks if a writer is waiting for a lock, and waiting writers not
21010 getting priority over waiting readers(Bruce)
21011 Fix crashes in psql when executing queries from external files(James)
21012 Fix problem with multiple order by columns, with the first one having
21013 NULL values(Jeroen)
21014 Use correct hash table support functions for float8 and int4(Thomas)
21015 Re-enable JOIN= option in CREATE OPERATOR statement (Thomas)
21016 Change precedence for boolean operators to match expected behavior(Thomas)
21017 Generate elog(ERROR) on over-large integer(Bruce)
21018 Allow multiple-argument functions in constraint clauses(Thomas)
21019 Check boolean input literals for 'true','false','yes','no','1','0'
21020 and throw elog(ERROR) if unrecognized(Thomas)
21021 Major large objects fix
21022 Fix for GROUP BY showing duplicates(Vadim)
21023 Fix for index scans in MergeJoin(Vadim)
21027 Subselects with EXISTS, IN, ALL, ANY key words (Vadim, Bruce, Thomas)
21028 New User Manual(Thomas, others)
21029 Speedup by inlining some frequently-called functions
21030 Real deadlock detection, no more timeouts(Bruce)
21031 Add SQL92 "constants" CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP,
21032 CURRENT_USER(Thomas)
21033 Modify constraint syntax to be SQL92-compliant(Thomas)
21034 Implement SQL92 PRIMARY KEY and UNIQUE clauses using indexes(Thomas)
21035 Recognize SQL92 syntax for FOREIGN KEY. Throw elog notice(Thomas)
21036 Allow NOT NULL UNIQUE constraint clause (each allowed separately before)(Thomas)
21037 Allow PostgreSQL-style casting ("::") of non-constants(Thomas)
21038 Add support for SQL3 TRUE and FALSE boolean constants(Thomas)
21039 Support SQL92 syntax for IS TRUE/IS FALSE/IS NOT TRUE/IS NOT FALSE(Thomas)
21040 Allow shorter strings for boolean literals (e.g. "t", "tr", "tru")(Thomas)
21041 Allow SQL92 delimited identifiers(Thomas)
21042 Implement SQL92 binary and hexadecimal string decoding (b'10' and x'1F')(Thomas)
21043 Support SQL92 syntax for type coercion of literal strings
21044 (e.g. "DATETIME 'now'")(Thomas)
21045 Add conversions for int2, int4, and OID types to and from text(Thomas)
21046 Use shared lock when building indexes(Vadim)
21047 Free memory allocated for an user query inside transaction block after
21048 this query is done, was turned off in <= 6.2.1(Vadim)
21049 New SQL statement CREATE PROCEDURAL LANGUAGE(Jan)
21050 New <productname>PostgreSQL</productname> Procedural Language (PL) backend interface(Jan)
21051 Rename pg_dump -H option to -h(Bruce)
21052 Add Java support for passwords, European dates(Peter)
21053 Use indexes for LIKE and ~, !~ operations(Bruce)
21054 Add hash functions for datetime and timespan(Thomas)
21055 Time Travel removed(Vadim, Bruce)
21056 Add paging for \d and \z, and fix \i(Bruce)
21057 Add Unix domain socket support to backend and to frontend library(Goran)
21058 Implement CREATE DATABASE/WITH LOCATION and initlocation utility(Thomas)
21059 Allow more SQL92 and/or <productname>PostgreSQL</productname> reserved words as column identifiers(Thomas)
21060 Augment support for SQL92 SET TIME ZONE...(Thomas)
21061 SET/SHOW/RESET TIME ZONE uses TZ backend environment variable(Thomas)
21062 Implement SET keyword = DEFAULT and SET TIME ZONE DEFAULT(Thomas)
21063 Enable SET TIME ZONE using TZ environment variable(Thomas)
21064 Add PGDATESTYLE environment variable to frontend and backend initialization(Thomas)
21065 Add PGTZ, PGCOSTHEAP, PGCOSTINDEX, PGRPLANS, PGGEQO
21066 frontend library initialization environment variables(Thomas)
21067 Regression tests time zone automatically set with "setenv PGTZ PST8PDT"(Thomas)
21068 Add pg_description table for info on tables, columns, operators, types, and
21070 Increase 16 char limit on system table/index names to 32 characters(Bruce)
21071 Rename system indexes(Bruce)
21072 Add 'GERMAN' option to SET DATESTYLE(Thomas)
21073 Define an "ISO-style" timespan output format with "hh:mm:ss" fields(Thomas)
21074 Allow fractional values for delta times (e.g. '2.5 days')(Thomas)
21075 Validate numeric input more carefully for delta times(Thomas)
21076 Implement day of year as possible input to date_part()(Thomas)
21077 Define timespan_finite() and text_timespan() functions(Thomas)
21078 Remove archive stuff(Bruce)
21079 Allow for a pg_password authentication database that is separate from
21080 the system password file(Todd)
21081 Dump ACLs, GRANT, REVOKE privileges(Matt)
21082 Define text, varchar, and bpchar string length functions(Thomas)
21083 Fix Query handling for inheritance, and cost computations(Bruce)
21084 Implement CREATE TABLE/AS SELECT (alternative to SELECT/INTO)(Thomas)
21085 Allow NOT, IS NULL, IS NOT NULL in constraints(Thomas)
21086 Implement UNIONs for SELECT(Bruce)
21087 Add UNION, GROUP, DISTINCT to INSERT(Bruce)
21088 varchar() stores only necessary bytes on disk(Bruce)
21089 Fix for BLOBs(Peter)
21090 Mega-Patch for JDBC...see README_6.3 for list of changes(Peter)
21091 Remove unused "option" from PQconnectdb()
21092 New LOCK command and lock manual page describing deadlocks(Bruce)
21093 Add new psql \da, \dd, \df, \do, \dS, and \dT commands(Bruce)
21094 Enhance psql \z to show sequences(Bruce)
21095 Show NOT NULL and DEFAULT in psql \d table(Bruce)
21096 New psql .psqlrc file start-up(Andrew)
21097 Modify sample start-up script in contrib/linux to show syslog(Thomas)
21098 New types for IP and MAC addresses in contrib/ip_and_mac(TomH)
21099 Unix system time conversions with date/time types in contrib/unixdate(Thomas)
21100 Update of contrib stuff(Massimo)
21101 Add Unix socket support to DBD::Pg(Goran)
21102 New python interface (PyGreSQL 2.0)(D'Arcy)
21103 New frontend/backend protocol has a version number, network byte order(Phil)
21104 Security features in pg_hba.conf enhanced and documented, many cleanups(Phil)
21105 CHAR() now faster access than VARCHAR() or TEXT
21106 ecpg embedded SQL preprocessor
21107 Reduce system column overhead(Vadmin)
21108 Remove pg_time table(Vadim)
21109 Add pg_type attribute to identify types that need length (bpchar, varchar)
21110 Add report of offending line when COPY command fails
21111 Allow VIEW privileges to be set separately from the underlying tables.
21112 For security, use GRANT/REVOKE on views as appropriate(Jan)
21113 Tables now have no default GRANT SELECT TO PUBLIC. You must
21114 explicitly grant such privileges.
21115 Clean up tutorial examples(Darren)
21117 Source Tree Changes
21118 -------------------
21119 Add new html development tools, and flow chart in /tools/backend
21120 Fix for SCO compiles
21121 Stratus computer port Robert Gillies
21122 Added support for shlib for BSD44_derived & i386_solaris
21123 Make configure more automated(Brook)
21124 Add script to check regression test results
21125 Break parser functions into smaller files, group together(Bruce)
21126 Rename heap_create to heap_create_and_catalog, rename heap_creatr
21127 to heap_create()(Bruce)
21128 Sparc/Linux patch for locking(TomS)
21129 Remove PORTNAME and reorganize port-specific stuff(Marc)
21130 Add optimizer README file(Bruce)
21131 Remove some recursion in optimizer and clean up some code there(Bruce)
21132 Fix for NetBSD locking(Henry)
21133 Fix for libptcl make(Tatsuo)
21135 Change IS TRUE, IS FALSE, ... to expressions using "=" rather than
21136 function calls to istrue() or isfalse() to allow optimization(Thomas)
21137 Various fixes NetBSD/Sparc related(TomH)
21138 Alpha linux locking(Travis,Ryan)
21139 Change elog(WARN) to elog(ERROR)(Bruce)
21140 FAQ for FreeBSD(Marc)
21141 Bring in the PostODBC source tree as part of our standard distribution(Marc)
21142 A minor patch for HP/UX 10 vs 9(Stan)
21143 New pg_attribute.atttypmod for type-specific info like varchar length(Bruce)
21144 UnixWare patches(Billy)
21145 New i386 'lock' for spinlock asm(Billy)
21146 Support for multiplexed backends is removed
21147 Start an OpenBSD port
21149 Start a Cygnus port
21150 Add string functions to regression suite(Thomas)
21151 Expand a few function names formerly truncated to 16 characters(Thomas)
21152 Remove un-needed malloc() calls and replace with palloc()(Bruce)
21158 <sect1 id="release-6-2-1">
21159 <title>Release 6.2.1</title>
21162 <title>Release date</title>
21163 <simpara>1997-10-17</simpara>
21167 6.2.1 is a bug-fix and usability release on 6.2.
21175 Allow strings to span lines, per <acronym>SQL92</acronym>.
21181 Include example trigger function for inserting user names on table updates.
21188 This is a minor bug-fix release on 6.2.
21189 For upgrades from pre-6.2 systems, a full dump/reload is required.
21190 Refer to the 6.2 release notes for instructions.
21194 <title>Migration from version 6.2 to version 6.2.1</title>
21197 This is a minor bug-fix release. A dump/reload is not required from version 6.2,
21198 but is required from any release prior to 6.2.
21201 In upgrading from version 6.2, if you choose to dump/reload you will find that
21202 avg(money) is now calculated correctly. All other bug fixes take effect
21203 upon updating the executables.
21206 Another way to avoid dump/reload is to use the following SQL command
21207 from <command>psql</command> to update the existing system table:
21210 update pg_aggregate set aggfinalfn = 'cash_div_flt8'
21211 where aggname = 'avg' and aggbasetype = 790;
21215 This will need to be done to every existing database, including template1.
21220 <title>Changes</title>
21224 Allow TIME and TYPE column names(Thomas)
21225 Allow larger range of true/false as boolean values(Thomas)
21226 Support output of "now" and "current"(Thomas)
21227 Handle DEFAULT with INSERT of NULL properly(Vadim)
21228 Fix for relation reference counts problem in buffer manager(Vadim)
21229 Allow strings to span lines, like ANSI(Thomas)
21230 Fix for backward cursor with ORDER BY(Vadim)
21231 Fix avg(cash) computation(Thomas)
21232 Fix for specifying a column twice in ORDER/GROUP BY(Vadim)
21233 Documented new libpq function to return affected rows, PQcmdTuples(Bruce)
21234 Trigger function for inserting user names for INSERT/UPDATE(Brook Milligan)
21240 <sect1 id="release-6-2">
21241 <title>Release 6.2</title>
21244 <title>Release date</title>
21245 <simpara>1997-10-02</simpara>
21249 A dump/restore is required for those wishing to migrate data from
21250 previous releases of <productname>PostgreSQL</productname>.
21254 <title>Migration from version 6.1 to version 6.2</title>
21257 This migration requires a complete dump of the 6.1 database and a
21258 restore of the database in 6.2.
21261 Note that the <command>pg_dump</command> and <command>pg_dumpall</command> utility from 6.2 should be used
21262 to dump the 6.1 database.
21267 <title>Migration from version 1.<replaceable>x</> to version 6.2</title>
21270 Those migrating from earlier 1.* releases should first upgrade to 1.09
21271 because the COPY output format was improved from the 1.02 release.
21276 <title>Changes</title>
21282 Fix problems with pg_dump for inheritance, sequences, archive tables(Bruce)
21283 Fix compile errors on overflow due to shifts, unsigned, and bad prototypes
21284 from Solaris(Diab Jerius)
21285 Fix bugs in geometric line arithmetic (bad intersection calculations)(Thomas)
21286 Check for geometric intersections at endpoints to avoid rounding ugliness(Thomas)
21287 Catch non-functional delete attempts(Vadim)
21288 Change time function names to be more consistent(Michael Reifenberg)
21289 Check for zero divides(Michael Reifenberg)
21290 Fix very old bug which made rows changed/inserted by a command
21291 visible to the command itself (so we had multiple update of
21292 updated rows, etc.)(Vadim)
21293 Fix for SELECT null, 'fail' FROM pg_am (Patrick)
21294 SELECT NULL as EMPTY_FIELD now allowed(Patrick)
21295 Remove un-needed signal stuff from contrib/pginterface
21296 Fix OR (where x != 1 or x isnull didn't return rows with x NULL) (Vadim)
21297 Fix time_cmp function (Vadim)
21298 Fix handling of functions with non-attribute first argument in
21299 WHERE clauses (Vadim)
21300 Fix GROUP BY when order of entries is different from order
21301 in target list (Vadim)
21302 Fix pg_dump for aggregates without sfunc1 (Vadim)
21306 Default genetic optimizer GEQO parameter is now 8(Bruce)
21307 Allow use parameters in target list having aggregates in functions(Vadim)
21308 Added JDBC driver as an interface(Adrian & Peter)
21309 pg_password utility
21310 Return number of rows inserted/affected by INSERT/UPDATE/DELETE etc.(Vadim)
21311 Triggers implemented with CREATE TRIGGER (SQL3)(Vadim)
21312 SPI (Server Programming Interface) allows execution of queries inside
21313 C-functions (Vadim)
21314 NOT NULL implemented (SQL92)(Robson Paniago de Miranda)
21315 Include reserved words for string handling, outer joins, and unions(Thomas)
21316 Implement extended comments ("/* ... */") using exclusive states(Thomas)
21317 Add "//" single-line comments(Bruce)
21318 Remove some restrictions on characters in operator names(Thomas)
21319 DEFAULT and CONSTRAINT for tables implemented (SQL92)(Vadim & Thomas)
21320 Add text concatenation operator and function (SQL92)(Thomas)
21321 Support WITH TIME ZONE syntax (SQL92)(Thomas)
21322 Support INTERVAL unit TO unit syntax (SQL92)(Thomas)
21323 Define types DOUBLE PRECISION, INTERVAL, CHARACTER,
21324 and CHARACTER VARYING (SQL92)(Thomas)
21325 Define type FLOAT(p) and rudimentary DECIMAL(p,s), NUMERIC(p,s) (SQL92)(Thomas)
21326 Define EXTRACT(), POSITION(), SUBSTRING(), and TRIM() (SQL92)(Thomas)
21327 Define CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP (SQL92)(Thomas)
21328 Add syntax and warnings for UNION, HAVING, INNER and OUTER JOIN (SQL92)(Thomas)
21329 Add more reserved words, mostly for SQL92 compliance(Thomas)
21330 Allow hh:mm:ss time entry for timespan/reltime types(Thomas)
21331 Add center() routines for lseg, path, polygon(Thomas)
21332 Add distance() routines for circle-polygon, polygon-polygon(Thomas)
21333 Check explicitly for points and polygons contained within polygons
21334 using an axis-crossing algorithm(Thomas)
21335 Add routine to convert circle-box(Thomas)
21336 Merge conflicting operators for different geometric data types(Thomas)
21337 Replace distance operator "<===>" with "<->"(Thomas)
21338 Replace "above" operator "!^" with ">^" and "below" operator "!|" with "<^"(Thomas)
21339 Add routines for text trimming on both ends, substring, and string position(Thomas)
21340 Added conversion routines circle(box) and poly(circle)(Thomas)
21341 Allow internal sorts to be stored in memory rather than in files(Bruce & Vadim)
21342 Allow functions and operators on internally-identical types to succeed(Bruce)
21343 Speed up backend start-up after profiling analysis(Bruce)
21344 Inline frequently called functions for performance(Bruce)
21345 Reduce open() calls(Bruce)
21346 psql: Add PAGER for \h and \?,\C fix
21347 Fix for psql pager when no tty(Bruce)
21348 New entab utility(Bruce)
21349 General trigger functions for referential integrity (Vadim)
21350 General trigger functions for time travel (Vadim)
21351 General trigger functions for AUTOINCREMENT/IDENTITY feature (Vadim)
21352 MOVE implementation (Vadim)
21354 Source Tree Changes
21355 -------------------
21356 HP-UX 10 patches (Vladimir Turin)
21357 Added SCO support, (Daniel Harris)
21358 MkLinux patches (Tatsuo Ishii)
21359 Change geometric box terminology from "length" to "width"(Thomas)
21360 Deprecate temporary unstored slope fields in geometric code(Thomas)
21361 Remove restart instructions from INSTALL(Bruce)
21362 Look in /usr/ucb first for install(Bruce)
21363 Fix c++ copy example code(Thomas)
21364 Add -o to psql manual page(Bruce)
21365 Prevent relname unallocated string length from being copied into database(Bruce)
21366 Cleanup for NAMEDATALEN use(Bruce)
21367 Fix pg_proc names over 15 chars in output(Bruce)
21368 Add strNcpy() function(Bruce)
21369 remove some (void) casts that are unnecessary(Bruce)
21370 new interfaces directory(Marc)
21371 Replace fopen() calls with calls to fd.c functions(Bruce)
21372 Make functions static where possible(Bruce)
21373 enclose unused functions in #ifdef NOT_USED(Bruce)
21374 Remove call to difftime() in timestamp support to fix SunOS(Bruce & Thomas)
21375 Changes for Digital Unix
21376 Portability fix for pg_dumpall(Bruce)
21377 Rename pg_attribute.attnvals to attdispersion(Bruce)
21378 "intro/unix" manual page now "pgintro"(Bruce)
21379 "built-in" manual page now "pgbuiltin"(Bruce)
21380 "drop" manual page now "drop_table"(Bruce)
21381 Add "create_trigger", "drop_trigger" manual pages(Thomas)
21382 Add constraints regression test(Vadim & Thomas)
21383 Add comments syntax regression test(Thomas)
21384 Add PGINDENT and support program(Bruce)
21385 Massive commit to run PGINDENT on all *.c and *.h files(Bruce)
21386 Files moved to /src/tools directory(Bruce)
21387 SPI and Trigger programming guides (Vadim & D'Arcy)
21393 <sect1 id="release-6-1-1">
21394 <title>Release 6.1.1</title>
21397 <title>Release date</title>
21398 <simpara>1997-07-22</simpara>
21402 <title>Migration from version 6.1 to version 6.1.1</title>
21405 This is a minor bug-fix release. A dump/reload is not required from version 6.1,
21406 but is required from any release prior to 6.1.
21407 Refer to the release notes for 6.1 for more details.
21412 <title>Changes</title>
21416 fix for SET with options (Thomas)
21417 allow pg_dump/pg_dumpall to preserve ownership of all tables/objects(Bruce)
21418 new psql \connect option allows changing usernames without changing databases
21419 fix for initdb --debug option(Yoshihiko Ichikawa))
21420 lextest cleanup(Bruce)
21422 fix date/time month boundary arithmetic(Thomas)
21423 fix timezone daylight handling for some ports(Thomas, Bruce, Tatsuo)
21424 timestamp overhauled to use standard functions(Thomas)
21425 other code cleanup in date/time routines(Thomas)
21426 psql's \d now case-insensitive(Bruce)
21427 psql's backslash commands can now have trailing semicolon(Bruce)
21428 fix memory leak in psql when using \g(Bruce)
21429 major fix for endian handling of communication to server(Thomas, Tatsuo)
21430 Fix for Solaris assembler and include files(Yoshihiko Ichikawa)
21431 allow underscores in usernames(Bruce)
21432 pg_dumpall now returns proper status, portability fix(Bruce)
21438 <sect1 id="release-6-1">
21439 <title>Release 6.1</title>
21442 <title>Release date</title>
21443 <simpara>1997-06-08</simpara>
21447 The regression tests have been adapted and extensively modified for the
21448 6.1 release of <productname>PostgreSQL</productname>.
21452 Three new data types (<type>datetime</type>, <type>timespan</type>, and <type>circle</type>) have been added to
21453 the native set of <productname>PostgreSQL</productname> types. Points, boxes, paths, and polygons
21454 have had their output formats made consistent across the data types.
21455 The polygon output in misc.out has only been spot-checked for correctness
21456 relative to the original regression output.
21460 <productname>PostgreSQL</productname> 6.1 introduces a new, alternate
21461 optimizer which uses <firstterm>genetic</firstterm>
21462 algorithms. These algorithms introduce a random behavior in the ordering
21463 of query results when the query contains multiple qualifiers or multiple
21464 tables (giving the optimizer a choice on order of evaluation). Several
21465 regression tests have been modified to explicitly order the results, and
21466 hence are insensitive to optimizer choices. A few regression tests are
21467 for data types which are inherently unordered (e.g. points and time
21468 intervals) and tests involving those types are explicitly bracketed with
21469 <command>set geqo to 'off'</command> and <command>reset geqo</command>.
21473 The interpretation of array specifiers (the curly braces around atomic
21474 values) appears to have changed sometime after the original regression
21475 tests were generated. The current <filename>./expected/*.out</filename> files reflect this
21476 new interpretation, which might not be correct!
21480 The float8 regression test fails on at least some platforms. This is due
21481 to differences in implementations of <function>pow()</function> and <function>exp()</function> and the signaling
21482 mechanisms used for overflow and underflow conditions.
21486 The <quote>random</> results in the random test should cause the
21487 <quote>random</quote> test to be <quote>failed</quote>, since the
21488 regression tests are evaluated using a simple diff. However,
21489 <quote>random</> does not seem to produce random results on my test
21490 machine (Linux/<application>gcc</>/i686).
21494 <title>Migration to Version 6.1</title>
21497 This migration requires a complete dump of the 6.0 database and a
21498 restore of the database in 6.1.
21501 Those migrating from earlier 1.* releases should first upgrade to 1.09
21502 because the COPY output format was improved from the 1.02 release.
21507 <title>Changes</title>
21513 packet length checking in library routines
21514 lock manager priority patch
21515 check for under/over flow of float8(Bruce)
21516 multitable join fix(Vadim)
21517 SIGPIPE crash fix(Darren)
21518 large object fixes(Sven)
21519 allow btree indexes to handle NULLs(Vadim)
21520 timezone fixes(D'Arcy)
21521 select SUM(x) can return NULL on no rows(Thomas)
21522 internal optimizer, executor bug fixes(Vadim)
21523 fix problem where inner loop in < or <= has no rows(Vadim)
21524 prevent re-commuting join index clauses(Vadim)
21525 fix join clauses for multiple tables(Vadim)
21526 fix hash, hashjoin for arrays(Vadim)
21527 fix btree for abstime type(Vadim)
21528 large object fixes(Raymond)
21529 fix buffer leak in hash indexes (Vadim)
21530 fix rtree for use in inner scan (Vadim)
21531 fix gist for use in inner scan, cleanups (Vadim, Andrea)
21532 avoid unnecessary local buffers allocation (Vadim, Massimo)
21533 fix local buffers leak in transaction aborts (Vadim)
21534 fix file manager memmory leaks, cleanups (Vadim, Massimo)
21535 fix storage manager memmory leaks (Vadim)
21536 fix btree duplicates handling (Vadim)
21537 fix deleted rows reincarnation caused by vacuum (Vadim)
21538 fix SELECT varchar()/char() INTO TABLE made zero-length fields(Bruce)
21539 many psql, pg_dump, and libpq memory leaks fixed using Purify (Igor)
21543 attribute optimization statistics(Bruce)
21544 much faster new btree bulk load code(Paul)
21545 BTREE UNIQUE added to bulk load code(Vadim)
21546 new lock debug code(Massimo)
21547 massive changes to libpg++(Leo)
21548 new GEQO optimizer speeds table multitable optimization(Martin)
21549 new WARN message for non-unique insert into unique key(Marc)
21550 update x=-3, no spaces, now valid(Bruce)
21551 remove case-sensitive identifier handling(Bruce,Thomas,Dan)
21552 debug backend now pretty-prints tree(Darren)
21553 new Oracle character functions(Edmund)
21554 new plaintext password functions(Dan)
21555 no such class or insufficient privilege changed to distinct messages(Dan)
21556 new ANSI timestamp function(Dan)
21557 new ANSI Time and Date types (Thomas)
21558 move large chunks of data in backend(Martin)
21559 multicolumn btree indexes(Vadim)
21560 new SET var TO value command(Martin)
21561 update transaction status on reads(Dan)
21562 new locale settings for character types(Oleg)
21563 new SEQUENCE serial number generator(Vadim)
21564 GROUP BY function now possible(Vadim)
21565 re-organize regression test(Thomas,Marc)
21566 new optimizer operation weights(Vadim)
21567 new psql \z grant/permit option(Marc)
21568 new MONEY data type(D'Arcy,Thomas)
21569 tcp socket communication speed improved(Vadim)
21570 new VACUUM option for attribute statistics, and for certain columns (Vadim)
21571 many geometric type improvements(Thomas,Keith)
21572 additional regression tests(Thomas)
21573 new datestyle variable(Thomas,Vadim,Martin)
21574 more comparison operators for sorting types(Thomas)
21575 new conversion functions(Thomas)
21576 new more compact btree format(Vadim)
21577 allow pg_dumpall to preserve database ownership(Bruce)
21578 new SET GEQO=# and R_PLANS variable(Vadim)
21579 old (!GEQO) optimizer can use right-sided plans (Vadim)
21580 typechecking improvement in SQL parser(Bruce)
21581 new SET, SHOW, RESET commands(Thomas,Vadim)
21582 new \connect database USER option
21583 new destroydb -i option (Igor)
21584 new \dt and \di psql commands (Darren)
21585 SELECT "\n" now escapes newline (A. Duursma)
21586 new geometry conversion functions from old format (Thomas)
21588 Source tree changes
21589 -------------------
21590 new configuration script(Marc)
21591 readline configuration option added(Marc)
21592 OS-specific configuration options removed(Marc)
21593 new OS-specific template files(Marc)
21594 no more need to edit Makefile.global(Marc)
21595 re-arrange include files(Marc)
21596 nextstep patches (Gregor Hoffleit)
21597 removed Windows-specific code(Bruce)
21598 removed postmaster -e option, now only postgres -e option (Bruce)
21599 merge duplicate library code in front/backends(Martin)
21600 now works with eBones, international Kerberos(Jun)
21601 more shared library support
21602 c++ include file cleanup(Bruce)
21603 warn about buggy flex(Bruce)
21604 DG/UX, Ultrix, IRIX, AIX portability fixes
21610 <sect1 id="release-6-0">
21611 <title>Release 6.0</title>
21614 <title>Release date</title>
21615 <simpara>1997-01-29</simpara>
21619 A dump/restore is required for those wishing to migrate data from
21620 previous releases of <productname>PostgreSQL</productname>.
21624 <title>Migration from version 1.09 to version 6.0</title>
21627 This migration requires a complete dump of the 1.09 database and a
21628 restore of the database in 6.0.
21633 <title>Migration from pre-1.09 to version 6.0</title>
21636 Those migrating from earlier 1.* releases should first upgrade to 1.09
21637 because the COPY output format was improved from the 1.02 release.
21642 <title>Changes</title>
21648 ALTER TABLE bug - running postgres process needs to re-read table definition
21649 Allow vacuum to be run on one table or entire database(Bruce)
21651 Fix array over-runs of memory writes(Kurt)
21652 Fix elusive btree range/non-range bug(Dan)
21653 Fix for hash indexes on some types like time and date
21654 Fix for pg_log size explosion
21655 Fix permissions on lo_export()(Bruce)
21656 Fix uninitialized reads of memory(Kurt)
21657 Fixed ALTER TABLE ... char(3) bug(Bruce)
21658 Fixed a few small memory leaks
21659 Fixed EXPLAIN handling of options and changed full_path option name
21660 Fixed output of group acl privileges
21661 Memory leaks (hunt and destroy with tools like Purify(Kurt)
21662 Minor improvements to rules system
21664 New asserts for run-checking
21665 Overhauled parser/analyze code to properly report errors and increase speed
21666 Pg_dump -d now handles NULL's properly(Bruce)
21667 Prevent SELECT NULL from crashing server (Bruce)
21668 Properly report errors when INSERT ... SELECT columns did not match
21669 Properly report errors when insert column names were not correct
21670 psql \g filename now works(Bruce)
21671 psql fixed problem with multiple statements on one line with multiple outputs
21672 Removed duplicate system OIDs
21673 SELECT * INTO TABLE . GROUP/ORDER BY gives unlink error if table exists(Bruce)
21674 Several fixes for queries that crashed the backend
21675 Starting quote in insert string errors(Bruce)
21676 Submitting an empty query now returns empty status, not just " " query(Bruce)
21680 Add EXPLAIN manual page(Bruce)
21681 Add UNIQUE index capability(Dan)
21682 Add hostname/user level access control rather than just hostname and user
21683 Add synonym of != for <>(Bruce)
21684 Allow "select oid,* from table"
21685 Allow BY,ORDER BY to specify columns by number, or by non-alias table.column(Bruce)
21686 Allow COPY from the frontend(Bryan)
21687 Allow GROUP BY to use alias column name(Bruce)
21688 Allow actual compression, not just reuse on the same page(Vadim)
21689 Allow installation-configuration option to auto-add all local users(Bryan)
21690 Allow libpq to distinguish between text value '' and null(Bruce)
21691 Allow non-postgres users with createdb privs to destroydb's
21692 Allow restriction on who can create C functions(Bryan)
21693 Allow restriction on who can do backend COPY(Bryan)
21694 Can shrink tables, pg_time and pg_log(Vadim & Erich)
21695 Change debug level 2 to print queries only, changed debug heading layout(Bruce)
21696 Change default decimal constant representation from float4 to float8(Bruce)
21697 European date format now set when postmaster is started
21698 Execute lowercase function names if not found with exact case
21699 Fixes for aggregate/GROUP processing, allow 'select sum(func(x),sum(x+y) from z'
21700 Gist now included in the distribution(Marc)
21701 Idend authentication of local users(Bryan)
21702 Implement BETWEEN qualifier(Bruce)
21703 Implement IN qualifier(Bruce)
21704 libpq has PQgetisnull()(Bruce)
21705 libpq++ improvements
21706 New options to initdb(Bryan)
21707 Pg_dump allow dump of OIDs(Bruce)
21708 Pg_dump create indexes after tables are loaded for speed(Bruce)
21709 Pg_dumpall dumps all databases, and the user table
21710 Pginterface additions for NULL values(Bruce)
21711 Prevent postmaster from being run as root
21712 psql \h and \? is now readable(Bruce)
21713 psql allow backslashed, semicolons anywhere on the line(Bruce)
21714 psql changed command prompt for lines in query or in quotes(Bruce)
21715 psql char(3) now displays as (bp)char in \d output(Bruce)
21716 psql return code now more accurate(Bryan?)
21717 psql updated help syntax(Bruce)
21718 Re-visit and fix vacuum(Vadim)
21719 Reduce size of regression diffs, remove timezone name difference(Bruce)
21720 Remove compile-time parameters to enable binary distributions(Bryan)
21721 Reverse meaning of HBA masks(Bryan)
21722 Secure Authentication of local users(Bryan)
21723 Speed up vacuum(Vadim)
21724 Vacuum now had VERBOSE option(Bruce)
21726 Source tree changes
21727 -------------------
21728 All functions now have prototypes that are compared against the calls
21729 Allow asserts to be disabled easily from Makefile.global(Bruce)
21730 Change oid constants used in code to #define names
21731 Decoupled sparc and solaris defines(Kurt)
21732 Gcc -Wall compiles cleanly with warnings only from unfixable constructs
21733 Major include file reorganization/reduction(Marc)
21734 Make now stops on compile failure(Bryan)
21735 Makefile restructuring(Bryan, Marc)
21736 Merge bsdi_2_1 to bsdi(Bruce)
21737 Monitor program removed
21738 Name change from Postgres95 to PostgreSQL
21739 New config.h file(Marc, Bryan)
21740 PG_VERSION now set to 6.0 and used by postmaster
21741 Portability additions, including Ultrix, DG/UX, AIX, and Solaris
21742 Reduced the number of #define's, centralized #define's
21743 Remove duplicate OIDS in system tables(Dan)
21744 Remove duplicate system catalog info or report mismatches(Dan)
21745 Removed many os-specific #define's
21746 Restructured object file generation/location(Bryan, Marc)
21747 Restructured port-specific file locations(Bryan, Marc)
21748 Unused/uninitialized variables corrected
21754 <sect1 id="release-1-09">
21755 <title>Release 1.09</title>
21758 <title>Release date</title>
21759 <simpara>1996-11-04</simpara>
21763 Sorry, we didn't keep track of changes from 1.02 to 1.09. Some of
21764 the changes listed in 6.0 were actually included in the 1.02.1 to 1.09
21769 <sect1 id="release-1-02">
21770 <title>Release 1.02</title>
21773 <title>Release date</title>
21774 <simpara>1996-08-01</simpara>
21778 <title>Migration from version 1.02 to version 1.02.1</title>
21781 Here is a new migration file for 1.02.1. It includes the 'copy' change
21782 and a script to convert old <acronym>ASCII</acronym> files.
21786 The following notes are for the benefit of users who want to migrate
21787 databases from <productname>Postgres95</> 1.01 and 1.02 to <productname>Postgres95</> 1.02.1.
21790 If you are starting afresh with <productname>Postgres95</> 1.02.1 and do not need
21791 to migrate old databases, you do not need to read any further.
21796 In order to upgrade older <productname>Postgres95</> version 1.01 or 1.02 databases to
21797 version 1.02.1, the following steps are required:
21802 Start up a new 1.02.1 postmaster
21807 Add the new built-in functions and operators of 1.02.1 to 1.01 or 1.02
21808 databases. This is done by running the new 1.02.1 server against
21809 your own 1.01 or 1.02 database and applying the queries attached at
21810 the end of the file. This can be done easily through <command>psql</>. If your
21811 1.01 or 1.02 database is named <literal>testdb</literal> and you have cut the commands
21812 from the end of this file and saved them in <filename>addfunc.sql</filename>:
21814 % psql testdb -f addfunc.sql
21817 Those upgrading 1.02 databases will get a warning when executing the
21818 last two statements in the file because they are already present in 1.02. This is
21819 not a cause for concern.
21826 <title>Dump/Reload Procedure</title>
21829 If you are trying to reload a pg_dump or text-mode, <literal>copy tablename to
21830 stdout</literal> generated with a previous version, you will need to run the
21831 attached <command>sed</command> script on the ASCII file before loading it into the
21832 database. The old format used '.' as end-of-data, while '\.' is now the
21833 end-of-data marker. Also, empty strings are now loaded in as '' rather
21834 than NULL. See the copy manual page for full details.
21837 sed 's/^\.$/\\./g' <in_file >out_file
21841 If you are loading an older binary copy or non-<systemitem>stdout</> copy, there is no
21842 end-of-data character, and hence no conversion necessary.
21845 -- following lines added by agc to reflect the case-insensitive
21846 -- regexp searching for varchar (in 1.02), and bpchar (in 1.02.1)
21847 create operator ~* (leftarg = bpchar, rightarg = text, procedure = texticregexeq);
21848 create operator !~* (leftarg = bpchar, rightarg = text, procedure = texticregexne);
21849 create operator ~* (leftarg = varchar, rightarg = text, procedure = texticregexeq);
21850 create operator !~* (leftarg = varchar, rightarg = text, procedure = texticregexne);
21856 <title>Changes</title>
21860 Source code maintenance and development
21861 * worldwide team of volunteers
21862 * the source tree now in CVS at ftp.ki.net
21865 * psql (and underlying libpq library) now has many more options for
21866 formatting output, including HTML
21867 * pg_dump now output the schema and/or the data, with many fixes to
21868 enhance completeness.
21869 * psql used in place of monitor in administration shell scripts.
21870 monitor to be deprecated in next release.
21871 * date/time functions enhanced
21872 * NULL insert/update/comparison fixed/enhanced
21873 * TCL/TK lib and shell fixed to work with both tck7.4/tk4.0 and tcl7.5/tk4.1
21875 Bug Fixes (almost too numerous to mention)
21877 * storage management
21878 * check for NULL pointer before dereferencing
21882 * added SolarisX86 port
21883 * added BSD/OS 2.1 port
21888 Contributors (apologies to any missed)
21889 * Kurt J. Lidl <lidl@va.pubnix.com>
21890 (missed in first run, but no less important)
21891 * Erich Stamberger <eberger@gewi.kfunigraz.ac.at>
21892 * Jason Wright <jason@shiloh.vnet.net>
21893 * Cees de Groot <C.deGroot@inter.NL.net>
21894 * ernst.molitor@uni-bonn.de
21895 * michael.siebenborn@ae3.Hypo.DE (Michael Siebenborn (6929))
21896 * Brian E. Gallew <geek+@cmu.edu>
21897 * Vadim B. Mikheev <vadim@sable.krasnoyarsk.su>
21898 * Adam Sussman <myddryn@vidya.com>
21899 * Chris Dunlop <chris@onthe.net.au>
21900 * Marc G. Fournier <scrappy@ki.net>
21901 * Dan McGuirk <mcguirk@indirect.com>
21902 * Dr_George_D_Detlefsen <drgeorge@ilt.com>
21903 * Erich Stamberger <eberger@gewi.kfunigraz.ac.at>
21904 * Massimo Dal Zotto <dz@cs.unitn.it>
21905 * Randy Kunkee <kunkee@Starbase.NeoSoft.COM>
21906 * Rick Weldon <rick@wisetech.com>
21907 * Thomas van Reimersdahl <reimersd@dali.techinfo.rwth-aachen.de>
21908 * david bennett <dave@bensoft.com>
21909 * ernst.molitor@uni-bonn.de
21910 * Julian Assange <proff@suburbia.net>
21911 * Bruce Momjian <pgman@candle.pha.pa.us>
21912 * Paul "Shag" Walmsley <ccshag@cclabs.missouri.edu>
21913 * "Alistair G. Crooks" <azcb0@sde.uts.amdahl.com>
21918 <sect1 id="release-1-01">
21919 <title>Release 1.01</title>
21922 <title>Release date</title>
21923 <simpara>1996-02-23</simpara>
21928 <title>Migration from version 1.0 to version 1.01</title>
21931 The following notes are for the benefit of users who want to migrate
21932 databases from <productname>Postgres95</> 1.0 to <productname>Postgres95</> 1.01.
21935 If you are starting afresh with <productname>Postgres95</> 1.01 and do not need
21936 to migrate old databases, you do not need to read any further.
21939 In order to <productname>Postgres95</> version 1.01 with databases created with
21940 <productname>Postgres95</> version 1.0, the following steps are required:
21945 Set the definition of <symbol>NAMEDATALEN</symbol> in <filename>src/Makefile.global</filename> to 16
21946 and <symbol>OIDNAMELEN</symbol> to 20.
21951 Decide whether you want to use Host based authentication.
21956 If you do, you must create a file name <literal>pg_hba</literal> in your top-level data
21957 directory (typically the value of your <envar>$PGDATA</envar>). <filename>src/libpq/pg_hba</filename>
21958 shows an example syntax.
21963 If you do not want host-based authentication, you can comment out
21968 in <filename>src/Makefile.global</filename>
21971 Note that host-based authentication is turned on by default, and if
21972 you do not take steps A or B above, the out-of-the-box 1.01 will
21973 not allow you to connect to 1.0 databases.
21981 Compile and install 1.01, but DO NOT do the <command>initdb</command> step.
21986 Before doing anything else, terminate your 1.0 postmaster, and
21987 backup your existing <envar>$PGDATA</envar> directory.
21992 Set your <envar>PGDATA</envar> environment variable to your 1.0 databases, but set up
21993 path up so that 1.01 binaries are being used.
21998 Modify the file <filename><envar>$PGDATA</envar>/PG_VERSION</filename> from 5.0 to 5.1
22003 Start up a new 1.01 postmaster
22008 Add the new built-in functions and operators of 1.01 to 1.0
22009 databases. This is done by running the new 1.01 server against
22010 your own 1.0 database and applying the queries attached and saving
22011 in the file 1.0_to_1.01.sql. This can be done easily through <command>psql</command>.
22012 If your 1.0 database is name <literal>testdb</literal>:
22015 % psql testdb -f 1.0_to_1.01.sql
22018 and then execute the following commands (cut and paste from here):
22021 -- add builtin functions that are new to 1.01
22023 create function int4eqoid (int4, oid) returns bool as 'foo'
22024 language 'internal';
22025 create function oideqint4 (oid, int4) returns bool as 'foo'
22026 language 'internal';
22027 create function char2icregexeq (char2, text) returns bool as 'foo'
22028 language 'internal';
22029 create function char2icregexne (char2, text) returns bool as 'foo'
22030 language 'internal';
22031 create function char4icregexeq (char4, text) returns bool as 'foo'
22032 language 'internal';
22033 create function char4icregexne (char4, text) returns bool as 'foo'
22034 language 'internal';
22035 create function char8icregexeq (char8, text) returns bool as 'foo'
22036 language 'internal';
22037 create function char8icregexne (char8, text) returns bool as 'foo'
22038 language 'internal';
22039 create function char16icregexeq (char16, text) returns bool as 'foo'
22040 language 'internal';
22041 create function char16icregexne (char16, text) returns bool as 'foo'
22042 language 'internal';
22043 create function texticregexeq (text, text) returns bool as 'foo'
22044 language 'internal';
22045 create function texticregexne (text, text) returns bool as 'foo'
22046 language 'internal';
22048 -- add builtin functions that are new to 1.01
22050 create operator = (leftarg = int4, rightarg = oid, procedure = int4eqoid);
22051 create operator = (leftarg = oid, rightarg = int4, procedure = oideqint4);
22052 create operator ~* (leftarg = char2, rightarg = text, procedure = char2icregexeq);
22053 create operator !~* (leftarg = char2, rightarg = text, procedure = char2icregexne);
22054 create operator ~* (leftarg = char4, rightarg = text, procedure = char4icregexeq);
22055 create operator !~* (leftarg = char4, rightarg = text, procedure = char4icregexne);
22056 create operator ~* (leftarg = char8, rightarg = text, procedure = char8icregexeq);
22057 create operator !~* (leftarg = char8, rightarg = text, procedure = char8icregexne);
22058 create operator ~* (leftarg = char16, rightarg = text, procedure = char16icregexeq);
22059 create operator !~* (leftarg = char16, rightarg = text, procedure = char16icregexne);
22060 create operator ~* (leftarg = text, rightarg = text, procedure = texticregexeq);
22061 create operator !~* (leftarg = text, rightarg = text, procedure = texticregexne);
22069 <title>Changes</title>
22074 * 1.01 is backwards compatible with 1.0 database provided the user
22075 follow the steps outlined in the MIGRATION_from_1.0_to_1.01 file.
22076 If those steps are not taken, 1.01 is not compatible with 1.0 database.
22079 * added PQdisplayTuples() to libpq and changed monitor and psql to use it
22080 * added NeXT port (requires SysVIPC implementation)
22081 * added CAST .. AS ... syntax
22082 * added ASC and DESC key words
22083 * added 'internal' as a possible language for CREATE FUNCTION
22084 internal functions are C functions which have been statically linked
22085 into the postgres backend.
22086 * a new type "name" has been added for system identifiers (table names,
22087 attribute names, etc.) This replaces the old char16 type. The
22088 of name is set by the NAMEDATALEN #define in src/Makefile.global
22089 * a readable reference manual that describes the query language.
22090 * added host-based access control. A configuration file ($PGDATA/pg_hba)
22091 is used to hold the configuration data. If host-based access control
22092 is not desired, comment out HBA=1 in src/Makefile.global.
22093 * changed regex handling to be uniform use of Henry Spencer's regex code
22094 regardless of platform. The regex code is included in the distribution
22095 * added functions and operators for case-insensitive regular expressions.
22096 The operators are ~* and !~*.
22097 * pg_dump uses COPY instead of SELECT loop for better performance
22100 * fixed an optimizer bug that was causing core dumps when
22101 functions calls were used in comparisons in the WHERE clause
22102 * changed all uses of getuid to geteuid so that effective uids are used
22103 * psql now returns non-zero status on errors when using -c
22104 * applied public patches 1-14
22110 <sect1 id="release-1-0">
22111 <title>Release 1.0</title>
22114 <title>Release date</title>
22115 <simpara>1995-09-05</simpara>
22119 <title>Changes</title>
22124 * The copyright of <productname>Postgres</productname> 1.0 has been loosened to be freely modifiable
22125 and modifiable for any purpose. Please read the COPYRIGHT file.
22126 Thanks to Professor Michael Stonebraker for making this possible.
22129 * date formats have to be MM-DD-YYYY (or DD-MM-YYYY if you're using
22130 EUROPEAN STYLE). This follows SQL-92 specs.
22131 * "delimiters" is now a key word
22134 * sql LIKE syntax has been added
22135 * copy command now takes an optional USING DELIMITER specification.
22136 delimiters can be any single-character string.
22137 * IRIX 5.3 port has been added.
22138 Thanks to Paul Walmsley and others.
22139 * updated pg_dump to work with new libpq
22140 * \d has been added psql
22141 Thanks to Keith Parks
22142 * regexp performance for architectures that use POSIX regex has been
22143 improved due to caching of precompiled patterns.
22144 Thanks to Alistair Crooks
22145 * a new version of libpq++
22146 Thanks to William Wanders
22149 * arbitrary userids can be specified in the createuser script
22150 * \c to connect to other databases in psql now works.
22151 * bad pg_proc entry for float4inc() is fixed
22152 * users with usecreatedb field set can now create databases without
22153 having to be usesuper
22154 * remove access control entries when the entry no longer has any
22156 * fixed non-portable datetimes implementation
22157 * added kerberos flags to the src/backend/Makefile
22158 * libpq now works with kerberos
22159 * typographic errors in the user manual have been corrected.
22160 * btrees with multiple index never worked, now we tell you they don't
22161 work when you try to use them
22167 <sect1 id="release-0-03">
22168 <title><productname>Postgres95</productname> Release 0.03</title>
22171 <title>Release date</title>
22172 <simpara>1995-07-21</simpara>
22176 <title>Changes</title>
22179 Incompatible changes:
22180 * BETA-0.3 IS INCOMPATIBLE WITH DATABASES CREATED WITH PREVIOUS VERSIONS
22181 (due to system catalog changes and indexing structure changes).
22182 * double-quote (") is deprecated as a quoting character for string literals;
22183 you need to convert them to single quotes ('). <!-- " -->
22184 * name of aggregates (eg. int4sum) are renamed in accordance with the
22185 SQL standard (eg. sum).
22186 * CHANGE ACL syntax is replaced by GRANT/REVOKE syntax.
22187 * float literals (eg. 3.14) are now of type float4 (instead of float8 in
22188 previous releases); you might have to do typecasting if you depend on it
22189 being of type float8. If you neglect to do the typecasting and you assign
22190 a float literal to a field of type float8, you might get incorrect values
22192 * LIBPQ has been totally revamped so that frontend applications
22193 can connect to multiple backends
22194 * the usesysid field in pg_user has been changed from int2 to int4 to
22195 allow wider range of Unix user ids.
22196 * the netbsd/freebsd/bsd o/s ports have been consolidated into a
22197 single BSD44_derived port. (thanks to Alistair Crooks)
22199 SQL standard-compliance (the following details changes that makes postgres95
22200 more compliant to the SQL-92 standard):
22201 * the following SQL types are now built-in: smallint, int(eger), float, real,
22202 char(N), varchar(N), date and time.
22204 The following are aliases to existing postgres types:
22205 smallint -> int2
22206 integer, int -> int4
22207 float, real -> float4
22208 char(N) and varchar(N) are implemented as truncated text types. In
22209 addition, char(N) does blank-padding.
22210 * single-quote (') is used for quoting string literals; '' (in addition to
22211 \') is supported as means of inserting a single quote in a string
22212 * SQL standard aggregate names (MAX, MIN, AVG, SUM, COUNT) are used
22213 (Also, aggregates can now be overloaded, i.e. you can define your
22214 own MAX aggregate to take in a user-defined type.)
22215 * CHANGE ACL removed. GRANT/REVOKE syntax added.
22216 - Privileges can be given to a group using the "GROUP" key word.
22218 GRANT SELECT ON foobar TO GROUP my_group;
22219 The key word 'PUBLIC' is also supported to mean all users.
22221 Privileges can only be granted or revoked to one user or group
22224 "WITH GRANT OPTION" is not supported. Only class owners can change
22226 - The default access control is to grant users readonly access.
22227 You must explicitly grant insert/update access to users. To change
22228 this, modify the line in
22229 src/backend/utils/acl.h
22230 that defines ACL_WORLD_DEFAULT
22233 * the bug where aggregates of empty tables were not run has been fixed. Now,
22234 aggregates run on empty tables will return the initial conditions of the
22235 aggregates. Thus, COUNT of an empty table will now properly return 0.
22236 MAX/MIN of an empty table will return a row of value NULL.
22237 * allow the use of \; inside the monitor
22238 * the LISTEN/NOTIFY asynchronous notification mechanism now work
22239 * NOTIFY in rule action bodies now work
22240 * hash indexes work, and access methods in general should perform better.
22241 creation of large btree indexes should be much faster. (thanks to Paul
22244 Other changes and enhancements:
22245 * addition of an EXPLAIN statement used for explaining the query execution
22246 plan (eg. "EXPLAIN SELECT * FROM EMP" prints out the execution plan for
22248 * WARN and NOTICE messages no longer have timestamps on them. To turn on
22249 timestamps of error messages, uncomment the line in
22250 src/backend/utils/elog.h:
22251 /* define ELOG_TIMESTAMPS */
22252 * On an access control violation, the message
22253 "Either no such class or insufficient privilege"
22254 will be given. This is the same message that is returned when
22255 a class is not found. This dissuades non-privileged users from
22256 guessing the existence of privileged classes.
22257 * some additional system catalog changes have been made that are not
22258 visible to the user.
22261 * The -oid option has been added to the "pg_result" tcl command.
22262 pg_result -oid returns oid of the last row inserted. If the
22263 last command was not an INSERT, then pg_result -oid returns "".
22264 * the large object interface is available as pg_lo* tcl commands:
22265 pg_lo_open, pg_lo_close, pg_lo_creat, etc.
22267 Portability enhancements and New Ports:
22268 * flex/lex problems have been cleared up. Now, you should be able to use
22269 flex instead of lex on any platforms. We no longer make assumptions of
22270 what lexer you use based on the platform you use.
22271 * The Linux-ELF port is now supported. Various configuration have been
22272 tested: The following configuration is known to work:
22273 kernel 1.2.10, gcc 2.6.3, libc 4.7.2, flex 2.5.2, bison 1.24
22274 with everything in ELF format,
22277 * ipcclean added to the distribution
22278 ipcclean usually does not need to be run, but if your backend crashes
22279 and leaves shared memory segments hanging around, ipcclean will
22280 clean them up for you.
22283 * the user manual has been revised and libpq documentation added.
22289 <sect1 id="release-0-02">
22290 <title><productname>Postgres95</productname> Release 0.02</title>
22293 <title>Release date</title>
22294 <simpara>1995-05-25</simpara>
22298 <title>Changes</title>
22302 Incompatible changes:
22303 * The SQL statement for creating a database is 'CREATE DATABASE' instead
22304 of 'CREATEDB'. Similarly, dropping a database is 'DROP DATABASE' instead
22305 of 'DESTROYDB'. However, the names of the executables 'createdb' and
22306 'destroydb' remain the same.
22309 * pgperl - a Perl (4.036) interface to Postgres95
22310 * pg_dump - a utility for dumping out a postgres database into a
22311 script file containing query commands. The script files are in a ASCII
22312 format and can be used to reconstruct the database, even on other
22313 machines and other architectures. (Also good for converting
22314 a Postgres 4.2 database to Postgres95 database.)
22316 The following ports have been incorporated into postgres95-beta-0.02:
22317 * the NetBSD port by Alistair Crooks
22318 * the AIX port by Mike Tung
22319 * the Windows NT port by Jon Forrest (more stuff but not done yet)
22320 * the Linux ELF port by Brian Gallew
22322 The following bugs have been fixed in postgres95-beta-0.02:
22323 * new lines not escaped in COPY OUT and problem with COPY OUT when first
22325 * cannot type return to use the default user id in createuser
22326 * SELECT DISTINCT on big tables crashes
22327 * Linux installation problems
22328 * monitor doesn't allow use of 'localhost' as PGHOST
22329 * psql core dumps when doing \c or \l
22330 * the "pgtclsh" target missing from src/bin/pgtclsh/Makefile
22331 * libpgtcl has a hard-wired default port number
22332 * SELECT DISTINCT INTO TABLE hangs
22333 * CREATE TYPE doesn't accept 'variable' as the internallength
22334 * wrong result using more than 1 aggregate in a SELECT
22340 <sect1 id="release-0-01">
22341 <title><productname>Postgres95</productname> Release 0.01</title>
22344 <title>Release date</title>
22345 <simpara>1995-05-01</simpara>
22354 <sect1 id="timing-results">
22355 <title>Timing Results</title>
22358 These timing results are from running the regression test with the commands
22361 % cd src/test/regress
22363 % time make runtest
22367 Timing under Linux 2.0.27 seems to have a roughly 5% variation from run
22368 to run, presumably due to the scheduling vagaries of multitasking systems.
22372 <title>Version 6.5</title>
22375 As has been the case for previous releases, timing between
22376 releases is not directly comparable since new regression tests
22377 have been added. In general, 6.5 is faster than previous
22382 Timing with <function>fsync()</function> disabled:
22386 02:00 Dual Pentium Pro 180, 224MB, UW-SCSI, Linux 2.0.36, gcc 2.7.2.3 -O2 -m486
22387 04:38 Sparc Ultra 1 143MHz, 64MB, Solaris 2.6
22392 Timing with <function>fsync()</function> enabled:
22396 04:21 Dual Pentium Pro 180, 224MB, UW-SCSI, Linux 2.0.36, gcc 2.7.2.3 -O2 -m486
22399 For the <systemitem class="osname">Linux</systemitem> system above, using <acronym>UW-SCSI</acronym> disks rather than (older) <acronym>IDE</acronym>
22400 disks leads to a 50% improvement in speed on the regression test.
22405 <title>Version 6.4beta</title>
22408 The times for this release are not directly comparable to those for previous releases
22409 since some additional regression tests have been included.
22410 In general, however, 6.4 should be slightly faster than the previous release (thanks, Bruce!).
22415 02:26 Dual Pentium Pro 180, 96MB, UW-SCSI, Linux 2.0.30, gcc 2.7.2.1 -O2 -m486
22421 <title>Version 6.3</title>
22424 The times for this release are not directly comparable to those for previous releases
22425 since some additional regression tests have been included and some obsolete tests involving
22426 time travel have been removed.
22427 In general, however, 6.3 is substantially faster than previous releases (thanks, Bruce!).
22432 02:30 Dual Pentium Pro 180, 96MB, UW-SCSI, Linux 2.0.30, gcc 2.7.2.1 -O2 -m486
22433 04:12 Dual Pentium Pro 180, 96MB, EIDE, Linux 2.0.30, gcc 2.7.2.1 -O2 -m486
22439 <title>Version 6.1</title>
22444 06:12 Pentium Pro 180, 32MB, EIDE, Linux 2.0.30, gcc 2.7.2 -O2 -m486
22445 12:06 P-100, 48MB, Linux 2.0.29, gcc
22446 39:58 Sparc IPC 32MB, Solaris 2.5, gcc 2.7.2.1 -O -g