6 Release date: 2003-11-17
7 _________________________________________________________________
11 Major changes in this release:
13 IN/NOT IN subqueries are now much more efficient
14 In previous releases, IN/NOT IN subqueries were joined to the
15 upper query by sequentially scanning the subquery looking for a
16 match. The 7.4 code uses the same sophisticated techniques used
17 by ordinary joins and so is much faster. An IN will now usually
18 be as fast as or faster than an equivalent EXISTS subquery;
19 this reverses the conventional wisdom that applied to previous
22 Improved GROUP BY processing by using hash buckets
23 In previous releases, rows to be grouped had to be sorted
24 first. The 7.4 code can do GROUP BY without sorting, by
25 accumulating results into a hash table with one entry per
26 group. It will still use the sort technique, however, if the
27 hash table is estimated to be too large to fit in sort_mem.
29 New multikey hash join capability
30 In previous releases, hash joins could only occur on single
31 keys. This release allows multicolumn hash joins.
33 Queries using the explicit JOIN syntax are now better optimized
34 Prior releases evaluated queries using the explicit JOIN syntax
35 only in the order implied by the syntax. 7.4 allows full
36 optimization of these queries, meaning the optimizer considers
37 all possible join orderings and chooses the most efficient.
38 Outer joins, however, must still follow the declared ordering.
40 Faster and more powerful regular expression code
41 The entire regular expression module has been replaced with a
42 new version by Henry Spencer, originally written for Tcl. The
43 code greatly improves performance and supports several flavors
44 of regular expressions.
46 Function-inlining for simple SQL functions
47 Simple SQL functions can now be inlined by including their SQL
48 in the main query. This improves performance by eliminating
49 per-call overhead. That means, simple SQL functions now behave
52 Full support for IPv6 connections and IPv6 address data types
53 Previous releases allowed only IPv4 connections, and the IP
54 data types only supported IPv4 addresses. This release adds
55 full IPv6 support in both of these areas.
57 Major improvements in SSL performance and reliability
58 Several people very familiar with the SSL API have overhauled
59 our SSL code to improve SSL key negotiation and error recovery.
61 Make free space map to efficiently reuse empty index pages, and other
62 free space management improvements
63 In previous releases, B-tree index pages that were left empty
64 because of deleted rows could only be reused by rows with index
65 values similar to the original rows indexed on that page. In
66 7.4, "VACUUM" records empty index pages and allows them to be
67 reused for any future index rows.
69 SQL-standard information schema
70 The information schema provides a standardized and stable way
71 to access information about the schema objects defined in a
74 Cursors conform more closely with the SQL standard
75 The commands "FETCH" and "MOVE" have been overhauled to conform
76 more closely to the SQL standard.
78 Cursors can exist outside transactions
79 These cursors are also called holdable cursors
81 New client-to-server protocol
82 The new protocol adds error codes, more status information,
83 faster startup, better support for binary data transmission,
84 parameter values separated from SQL commands, prepared
85 statements available at the protocol level, and cleaner
86 recovery from "COPY" failures. The older protocol is still
87 supported by both server and clients.
89 libpq and ECPG applications are now fully thread-safe
90 While previous libpq releases already supported threads, this
91 release improves thread safety by fixing some non-thread-safe
92 code that was used during database connection startup. The
93 "configure" option "--enable-thread-safety" must be used to
96 New version of full-text indexing
97 A new full-text indexing suite is available in
101 The new autovacuum tool in "contrib/autovacuum" monitors the
102 database statistics tables for "INSERT"/"UPDATE"/"DELETE"
103 activity and automatically vacuums tables when needed.
105 Array handling has been improved and moved into the server core
106 Many array limitations have been removed, and arrays behave
107 more like fully-supported data types.
108 _________________________________________________________________
110 Migration to version 7.4
112 A dump/restore using pg_dump is required for those wishing to migrate
113 data from any previous release.
115 Observe the following incompatibilities:
117 * The server-side autocommit setting was removed and reimplemented
118 in client applications and languages. Server-side autocommit was
119 causing too many problems with languages and applications that
120 wanted to control their own autocommit behavior, so autocommit was
121 removed from the server and added to individual client APIs as
123 * Error message wording has changed substantially in this release.
124 Significant effort was invested to make the messages more
125 consistent and user-oriented. If your applications try to detect
126 different error conditions by parsing the error message, you are
127 strongly encourage to use the new error code facility.
128 * Inner joins using the explicit JOIN syntax may behave differently
129 because they are now better optimized.
130 * A number of server configuration parameters have been renamed for
131 clarity, primarily those related to logging.
132 * FETCH 0 or MOVE 0 now does nothing. In prior releases, FETCH 0
133 would fetch all remaining rows, and MOVE 0 would move to the end
135 * "FETCH" and "MOVE" now return the actual number of rows
136 fetched/moved, or zero if at the beginning/end of the cursor.
137 Prior releases would return the row count passed to the command,
138 not the number of rows actually fetched or moved.
139 * "COPY" now can process files that use carriage-return or
140 carriage-return/line-feed end-of-line sequences. Literal
141 carriage-returns and line-feeds are no longer accepted in data
142 values; use \r and \n instead.
143 * Trailing spaces are now trimmed when converting from type char(n)
144 to varchar(n) or text. This is what most people always expected to
146 * The data type float(p) now measures "p" in binary digits, not
147 decimal digits. The new behavior follows the SQL standard.
148 * Ambiguous date values now must match the ordering specified by the
149 datestyle setting. In prior releases, a date specification of
150 10/20/03 was interpreted as a date in October even if datestyle
151 specified that the day should be first. 7.4 will throw an error if
152 a date specification is invalid for the current setting of
154 * The functions oidrand, oidsrand, and userfntest have been removed.
155 These functions were determined to be no longer useful.
156 * String literals specifying time-varying date/time values, such as
157 'now' or 'today' will no longer work as expected in column default
158 expressions; they now cause the time of the table creation to be
159 the default, not the time of the insertion. Functions such as
160 now(), current_timestamp, or current_date should be used instead.
161 In previous releases, there was special code so that strings such
162 as 'now' were interpreted at "INSERT" time and not at table
163 creation time, but this work around didn't cover all cases.
164 Release 7.4 now requires that defaults be defined properly using
165 functions such as now() or current_timestamp. These will work in
167 * The dollar sign ($) is no longer allowed in operator names. It can
168 instead be a non-first character in identifiers. This was done to
169 improve compatibility with other database systems, and to avoid
170 syntax problems when parameter placeholders ($n) are written
171 adjacent to operators.
172 _________________________________________________________________
176 Below you will find a detailed account of the changes between release
177 7.4 and the previous major release.
178 _________________________________________________________________
180 Server Operation Changes
182 * Allow IPv6 server connections (Nigel Kukard, Johan Jordaan, Bruce,
183 Tom, Kurt Roeckx, Andrew Dunstan)
184 * Fix SSL to handle errors cleanly (Nathan Mueller)
185 In prior releases, certain SSL API error reports were not handled
186 correctly. This release fixes those problems.
187 * SSL protocol security and performance improvements (Sean
189 SSL key renegotiation was happening too frequently, causing poor
190 SSL performance. Also, initial key handling was improved.
191 * Print lock information when a deadlock is detected (Tom)
192 This allows easier debugging of deadlock situations.
193 * Update "/tmp" socket modification times regularly to avoid their
195 This should help prevent "/tmp" directory cleaner administration
196 scripts from removing server socket files.
197 * Enable PAM for Mac OS X (Aaron Hillegass)
198 * Make B-tree indexes fully WAL-safe (Tom)
199 In prior releases, under certain rare cases, a server crash could
200 cause B-tree indexes to become corrupt. This release removes those
202 * Allow B-tree index compaction and empty page reuse (Tom)
203 * Fix inconsistent index lookups during split of first root page
205 In prior releases, when a single-page index split into two pages,
206 there was a brief period when another database session could miss
207 seeing an index entry. This release fixes that rare failure case.
208 * Improve free space map allocation logic (Tom)
209 * Preserve free space information between server restarts (Tom)
210 In prior releases, the free space map was not saved when the
211 postmaster was stopped, so newly started servers had no free space
212 information. This release saves the free space map, and reloads it
213 when the server is restarted.
214 * Add start time to pg_stat_activity (Neil)
215 * New code to detect corrupt disk pages; erase with
216 zero_damaged_pages (Tom)
217 * New client/server protocol: faster, no username length limit,
218 allow clean exit from "COPY" (Tom)
219 * Add transaction status, table ID, column ID to client/server
221 * Add binary I/O to client/server protocol (Tom)
222 * Remove autocommit server setting; move to client applications
224 * New error message wording, error codes, and three levels of error
225 detail (Tom, Joe, Peter)
226 _________________________________________________________________
228 Performance Improvements
230 * Add hashing for GROUP BY aggregates (Tom)
231 * Make nested-loop joins be smarter about multicolumn indexes (Tom)
232 * Allow multikey hash joins (Tom)
233 * Improve constant folding (Tom)
234 * Add ability to inline simple SQL functions (Tom)
235 * Reduce memory usage for queries using complex functions (Tom)
236 In prior releases, functions returning allocated memory would not
237 free it until the query completed. This release allows the freeing
238 of function-allocated memory when the function call completes,
239 reducing the total memory used by functions.
240 * Improve GEQO optimizer performance (Tom)
241 There were several inefficiencies in the way the GEQO optimizer
242 managed potential query paths. This release fixes this.
243 * Allow IN/NOT IN to be handled via hash tables (Tom)
244 * Improve NOT IN (subquery) performance (Tom)
245 * Allow most IN subqueries to be processed as joins (Tom)
246 * Pattern matching operations can use indexes regardless of locale
248 There is no way for non-ASCII locales to use the standard indexes
249 for LIKE comparisons. This release adds a way to create a special
251 * Allow the postmaster to preload libraries using preload_libraries
253 For shared libraries that require a long time to load, this option
254 is available so the library can be preloaded in the postmaster and
255 inherited by all database sessions.
256 * Improve optimizer cost computations, particularly for subqueries
258 * Avoid sort when subquery ORDER BY matches upper query (Tom)
259 * Deduce that WHERE a.x = b.y AND b.y = 42 also means a.x = 42 (Tom)
260 * Allow hash/merge joins on complex joins (Tom)
261 * Allow hash joins for more data types (Tom)
262 * Allow join optimization of explicit inner joins, disable with
263 join_collapse_limit (Tom)
264 * Add parameter from_collapse_limit to control conversion of
265 subqueries to joins (Tom)
266 * Use faster and more powerful regular expression code from Tcl
268 * Use bit-mapped relation sets in the optimizer (Tom)
269 * Improve connection startup time (Tom)
270 The new client/server protocol requires fewer network packets to
271 start a database session.
272 * Improve trigger/constraint performance (Stephan)
273 * Improve speed of col IN (const, const, const, ...) (Tom)
274 * Fix hash indexes which were broken in rare cases (Tom)
275 * Improve hash index concurrency and speed (Tom)
276 Prior releases suffered from poor hash index performance,
277 particularly for high concurrency situations. This release fixes
278 that, and the development group is interested in reports comparing
279 B-tree and hash index performance.
280 * Align shared buffers on 32-byte boundary for copy speed
281 improvement (Manfred Spraul)
282 Certain CPU's perform faster data copies when addresses are
284 * Data type numeric reimplemented for better performance (Tom)
285 numeric used to be stored in base 100. The new code uses base
286 10000, for significantly better performance.
287 _________________________________________________________________
289 Server Configuration Changes
291 * Rename server parameter server_min_messages to log_min_messages
293 This was done so most parameters that control the server logs
295 * Rename show_*_stats to log_*_stats (Bruce)
296 * Rename show_source_port to log_source_port (Bruce)
297 * Rename hostname_lookup to log_hostname (Bruce)
298 * Add checkpoint_warning to warn of excessive checkpointing (Bruce)
299 In prior releases, it was difficult to determine if checkpoint was
300 happening too frequently. This feature adds a warning to the
301 server logs when excessive checkpointing happens.
302 * New read-only server parameters for localization (Tom)
303 * Change debug server log messages to output as DEBUG rather than
305 * Prevent server log variables from being turned off by
306 non-superusers (Bruce)
307 This is a security feature so non-superusers cannot disable
308 logging that was enabled by the administrator.
309 * log_min_messages/client_min_messages now controls debug_* output
311 This centralizes client debug information so all debug output can
312 be sent to either the client or server logs.
313 * Add Mac OS X Rendezvous server support (Chris Campbell)
314 This allows Mac OS X hosts to query the network for available
316 * Add ability to print only slow statements using
317 log_min_duration_statement (Christopher)
318 This is an often requested debugging feature that allows
319 administrators to see only slow queries in their server logs.
320 * Allow "pg_hba.conf" to accept netmasks in CIDR format (Andrew
322 This allows administrators to merge the host IP address and
323 netmask fields into a single CIDR field in "pg_hba.conf".
324 * New read-only parameter is_superuser (Tom)
325 * New parameter log_error_verbosity to control error detail (Tom)
326 This works with the new error reporting feature to supply
327 additional error information like hints, file names and line
329 * postgres --describe-config now dumps server config variables
331 This option is useful for administration tools that need to know
332 the configuration variable names and their minimums, maximums,
333 defaults, and descriptions.
334 * Add new columns in pg_settings: context, type, source, min_val,
336 * Make default shared_buffers 1000 and max_connections 100, if
338 Prior versions defaulted to 64 shared buffers so PostgreSQL would
339 start on even very old systems. This release tests the amount of
340 shared memory allowed by the platform and selects more reasonable
341 default values if possible. Of course, users are still encouraged
342 to evaluate their resource load and size shared_buffers
344 * New "pg_hba.conf" record type hostnossl to prevent SSL connections
346 In prior releases, there was no way to prevent SSL connections if
347 both the client and server supported SSL. This option allows that
349 * Remove parameter geqo_random_seed (Tom)
350 * Add server parameter regex_flavor to control regular expression
352 * Make "pg_ctl" better handle nonstandard ports (Greg)
353 _________________________________________________________________
357 * New SQL-standard information schema (Peter)
358 * Add read-only transactions (Peter)
359 * Print key name and value in foreign-key violation messages (Dmitry
361 * Allow users to see their own queries in pg_stat_activity (Kevin
363 In prior releases, only the superuser could see query strings
364 using pg_stat_activity. Now ordinary users can see their own query
366 * Fix aggregates in subqueries to match SQL standard (Tom)
367 The SQL standard says that an aggregate function appearing within
368 a nested subquery belongs to the outer query if its argument
369 contains only outer-query variables. Prior PostgreSQL releases did
370 not handle this fine point correctly.
371 * Add option to prevent auto-addition of tables referenced in query
373 By default, tables mentioned in the query are automatically added
374 to the FROM clause if they are not already there. This is
375 compatible with historic POSTGRES behavior but is contrary to the
376 SQL standard. This option allows selecting standard-compatible
378 * Allow UPDATE ... SET col = DEFAULT (Rod)
379 This allows "UPDATE" to set a column to its declared default
381 * Allow expressions to be used in LIMIT/OFFSET (Tom)
382 In prior releases, LIMIT/OFFSET could only use constants, not
384 * Implement CREATE TABLE AS EXECUTE (Neil, Peter)
385 _________________________________________________________________
387 Object Manipulation Changes
389 * Make "CREATE SEQUENCE" grammar more conforming to SQL 2003 (Neil)
390 * Add statement-level triggers (Neil)
391 While this allows a trigger to fire at the end of a statement, it
392 does not allow the trigger to access all rows modified by the
393 statement. This capability is planned for a future release.
394 * Add check constraints for domains (Rod)
395 This greatly increases the usefulness of domains by allowing them
396 to use check constraints.
397 * Add "ALTER DOMAIN" (Rod)
398 This allows manipulation of existing domains.
399 * Fix several zero-column table bugs (Tom)
400 PostgreSQL supports zero-column tables. This fixes various bugs
401 that occur when using such tables.
402 * Have ALTER TABLE ... ADD PRIMARY KEY add not-null constraint (Rod)
403 In prior releases, ALTER TABLE ... ADD PRIMARY would add a unique
404 index, but not a not-null constraint. That is fixed in this
406 * Add ALTER TABLE ... WITHOUT OIDS (Rod)
407 This allows control over whether new and updated rows will have an
408 OID column. This is most useful for saving storage space.
409 * Add ALTER SEQUENCE to modify minimum, maximum, increment, cache,
411 * Add ALTER TABLE ... CLUSTER ON (Alvaro Herrera)
412 This command is used by "pg_dump" to record the cluster column for
413 each table previously clustered. This information is used by
414 database-wide cluster to cluster all previously clustered tables.
415 * Improve automatic type casting for domains (Rod, Tom)
416 * Allow dollar signs in identifiers, except as first character (Tom)
417 * Disallow dollar signs in operator names, so x=$1 works (Tom)
418 * Allow copying table schema using LIKE subtable, also SQL 2003
419 feature INCLUDING DEFAULTS (Rod)
420 * Add WITH GRANT OPTION clause to "GRANT" (Peter)
421 This enabled "GRANT" to give other users the ability to grant
422 privileges on a object.
423 _________________________________________________________________
425 Utility Command Changes
427 * Add ON COMMIT clause to "CREATE TABLE" for temporary tables
429 This adds the ability for a table to be dropped or all rows
430 deleted on transaction commit.
431 * Allow cursors outside transactions using WITH HOLD (Neil)
432 In previous releases, cursors were removed at the end of the
433 transaction. Using WITH HOLD, the current release allows cursors
434 to remain readable after the creating transaction.
435 * FETCH 0 and MOVE 0 now do nothing (Bruce)
436 In previous releases, FETCH 0 fetched all remaining rows, and MOVE
437 0 moved to the end of the cursor.
438 * Cause "FETCH" and "MOVE" to return the number of rows
439 fetched/moved, or zero if at the beginning/end of cursor, per SQL
441 In prior releases, the row count returned by "FETCH" and "MOVE"
442 did not accurately reflect the number of rows processed.
443 * Properly handle SCROLL with cursors, or report an error (Neil)
444 Certain cursors can not be fetched backwards optimally. By
445 specifying SCROLL, extra work will be performed to guarantee that
446 the cursor can be fetched in reverse or random order.
447 * Implement SQL-compatible option FIRST, LAST, ABSOLUTE n, RELATIVE
448 n for "FETCH" and "MOVE" (Tom)
449 * Allow "EXPLAIN" on "DECLARE CURSOR" (Tom)
450 * Allow "CLUSTER" to use index marked as pre-clustered by default
452 * Allow "CLUSTER" to cluster all tables (Alvaro Herrera)
453 This allows all previously clustered tables in a database to be
454 reclustered with a single command.
455 * Prevent "CLUSTER" on partial indexes (Tom)
456 * Allow DOS and Mac line-endings in "COPY" files (Bruce)
457 * Disallow literal carriage return as a data value,
458 backslash-carriage-return and \r are still allowed (Bruce)
459 * "COPY" changes (binary, \.) (Tom)
460 * Recover from "COPY" failure cleanly (Tom)
461 * Prevent possible memory leaks in "COPY" (Tom)
462 * Make "TRUNCATE" transaction-safe (Rod)
463 "TRUNCATE" can now be used inside a transaction, and rolled back
464 if the transaction aborts.
465 * Allow prepare/bind of utility commands like "FETCH" and "EXPLAIN"
467 * Add "EXPLAIN EXECUTE" (Neil)
468 * Improve "VACUUM" performance on indexes by reducing WAL traffic
470 * Functional indexes have been generalized into indexes on
472 In prior releases, functional indexes only supported a simple
473 function applied to one or more column names. This release allows
474 any type of scalar expression.
475 * Have "SHOW TRANSACTION ISOLATION" match input to "SET TRANSACTION
477 * Have "COMMENT ON DATABASE" on nonlocal database generate a warning
479 Database comments are stored in database-local tables so comments
480 on a database have to be stored in each database.
481 * Improve reliability of "LISTEN"/"NOTIFY" (Tom)
482 * Allow "REINDEX" to reliably reindex nonshared system catalog
484 This allows system tables to be reindexed without the requirement
485 of a standalone session, which was necessary in previous releases.
486 The only tables that now require a standalone session for
487 reindexing are the global system tables pg_database, pg_shadow,
489 _________________________________________________________________
491 Data Type and Function Changes
493 * New server parameter extra_float_digits to control precision
494 display of floating-point numbers (Pedro Ferreira, Tom)
495 This controls output precision which was causing regression
497 * Allow +1300 as a numeric time-zone specifier, for FJST (Tom)
498 * Remove rarely used functions oidrand, oidsrand, and userfntest
500 * Add md5() function to main server, already in "contrib/pgcrypto"
502 An MD5 function was frequently requested. For more complex
503 encryption capabilities, use "contrib/pgcrypto".
504 * Increase date range of timestamp (John Cochran)
505 * Change EXTRACT(EPOCH FROM timestamp) so timestamp without time
506 zone is assumed to be in local time, not GMT (Tom)
507 * Trap division by zero in case the operating system doesn't prevent
509 * Change the numeric data type internally to base 10000 (Tom)
510 * New hostmask() function (Greg Wickham)
511 * Fixes for to_char() and to_timestamp() (Karel)
512 * Allow functions that can take any argument data type and return
513 any data type, using anyelement and anyarray (Joe)
514 This allows the creation of functions that can work with any data
516 * Arrays may now be specified as ARRAY[1,2,3],
517 ARRAY[['a','b'],['c','d']], or ARRAY[ARRAY[ARRAY[2]]] (Joe)
518 * Allow proper comparisons for arrays, including ORDER BY and
519 DISTINCT support (Joe)
520 * Allow indexes on array columns (Joe)
521 * Allow array concatenation with || (Joe)
522 * Allow WHERE qualification expr op ANY/SOME/ALL (array_expr) (Joe)
523 This allows arrays to behave like a list of values, for purposes
524 like SELECT * FROM tab WHERE col IN (array_val).
525 * New array functions array_append, array_cat, array_lower,
526 array_prepend, array_to_string, array_upper, string_to_array (Joe)
527 * Allow user defined aggregates to use polymorphic functions (Joe)
528 * Allow assignments to empty arrays (Joe)
529 * Allow 60 in seconds fields of time, timestamp, and interval input
531 Sixty-second values are needed for leap seconds.
532 * Allow cidr data type to be cast to text (Tom)
533 * Disallow invalid time zone names (Tom)
534 * Trim trailing spaces when char is cast to varchar or text (Tom)
535 * Make float(p) measure the precision "p" in binary digits, not
537 * Add IPv6 support to the inet and cidr data types (Michael Graff)
538 * Add family() function to report whether address is IPv4 or IPv6
540 * Have SHOW datestyle generate output similar to that used by SET
542 * Make EXTRACT(TIMEZONE) and SET/SHOW TIME ZONE follow the SQL
543 convention for the sign of time zone offsets, i.e., positive is
545 * Fix date_trunc('quarter', ...) (B?jthe Zolt?n)
546 Prior releases returned an incorrect value for this function call.
547 * Make initcap() more compatible with Oracle (Mike Nolan)
548 initcap() now uppercases a letter appearing after any
549 non-alphanumeric character, rather than only after whitespace.
550 * Allow only datestyle field order for date values not in ISO-8601
552 * Add new datestyle values MDY, DMY, and YMD to set input field
553 order; honor US and European for backward compatibility (Tom)
554 * String literals like 'now' or 'today' will no longer work as a
555 column default. Use functions such as now(), current_timestamp
556 instead. (change required for prepared statements) (Tom)
557 * Treat NaN as larger than any other value in min()/max() (Tom)
558 NaN was already sorted after ordinary numeric values for most
559 purposes, but min() and max() didn't get this right.
560 * Prevent interval from suppressing :00 seconds display
561 * New function pg_get_triggerdef(prettyprint) and
562 pg_constraint_is_visible()
563 * Allow time to be specified as 040506 or 0405 (Tom)
564 * Input date order must now be YYYY-MM-DD (with 4-digit year) or
566 * Make pg_get_constraintdef to support unique, primary-key, and
567 check constraints (Christopher)
568 _________________________________________________________________
570 Server-Side Language Changes
572 * Prevent PL/pgSQL crash when RETURN NEXT is used on a zero-row
573 record variable (Tom)
574 * Make PL/Python's spi_execute interface handle null values properly
576 * Allow PL/pgSQL to declare variables of composite types without
578 * Fix PL/Python's _quote() function to handle big integers
579 * Make PL/Python an untrusted language, now called plpythonu (Kevin
581 The Python language no longer supports a restricted execution
582 environment, so the trusted version of PL/Python was removed. If
583 this situation changes, a version of PL/python that can be used by
584 non-superusers will be readded.
585 * Allow polymorphic PL/pgSQL functions (Tom, Joe)
586 * Allow polymorphic SQL functions (Joe)
587 * Improved compiled function caching mechanism in PL/pgSQL with full
588 support for polymorphism (Joe)
589 * Add new parameter $0 in PL/pgSQL representing the function's
590 actual return type (Joe)
591 * Allow PL/Tcl and PL/Python to use the same trigger on multiple
593 * Fixed PL/Tcl's spi_prepare to accept fully qualified type names in
594 the parameter type list (Jan)
595 _________________________________________________________________
599 * Add \pset pager always to always use pager (Greg)
600 This forces the pager to be used even if the number of rows is
601 less than the screen height. This is valuable for rows that wrap
602 across several screen rows.
603 * Improve tab completion (Rod, Ross Reedstrom, Ian Barwick)
604 * Reorder \? help into groupings (Harald Armin Massa, Bruce)
605 * Add backslash commands for listing schemas, casts, and conversions
607 * "\encoding" now changes based on the server parameter
608 client_encoding server (Tom)
609 In previous versions, "\encoding" was not aware of encoding
610 changes made using SET client_encoding.
611 * Save editor buffer into readline history (Ross)
612 When "\e" is used to edit a query, the result is saved in the
613 readline history for retrieval using the up arrow.
614 * Improve "\d" display (Christopher)
615 * Enhance HTML mode to be more standards-conforming (Greg)
616 * New "\set AUTOCOMMIT off" capability (Tom)
617 This takes the place of the removed server parameter autocommit.
618 * New "\set VERBOSITY" to control error detail (Tom)
619 This controls the new error reporting details.
620 * New prompt escape sequence %x to show transaction status (Tom)
621 * Long options for psql are now available on all platforms
622 _________________________________________________________________
626 * Multiple pg_dump fixes, including tar format and large objects
627 * Allow pg_dump to dump specific schemas (Neil)
628 * Make pg_dump preserve column storage characteristics (Christopher)
629 This preserves ALTER TABLE ... SET STORAGE information.
630 * Make pg_dump preserve "CLUSTER" characteristics (Christopher)
631 * Have pg_dumpall use "GRANT"/"REVOKE" to dump database-level
633 * Allow pg_dumpall to support the options "-a", "-s", "-x" of
635 * Prevent pg_dump from lowercasing identifiers specified on the
637 * pg_dump options "--use-set-session-authorization" and
638 "--no-reconnect" now do nothing, all dumps use "SET SESSION
640 pg_dump no longer reconnects to switch users, but instead always
641 uses "SET SESSION AUTHORIZATION". This will reduce password
642 prompting during restores.
643 * Long options for pg_dump are now available on all platforms
644 PostgreSQL now includes its own long-option processing routines.
645 _________________________________________________________________
649 * Add function PQfreemem for freeing memory on Windows, suggested
651 Windows requires that memory allocated in a library be freed by a
652 function in the same library, hence free() doesn't work for
653 freeing memory allocated by libpq. PQfreemem is the proper way to
654 free libpq memory, especially on Windows, and is recommended for
655 other platforms as well.
656 * Document service capability, and add sample file (Bruce)
657 This allows clients to look up connection information in a central
658 file on the client machine.
659 * Make PQsetdbLogin have the same defaults as PQconnectdb (Tom)
660 * Allow libpq to cleanly fail when result sets are too large (Tom)
661 * Improve performance of function PGunescapeBytea (Ben Lamb)
662 * Allow thread-safe libpq with "configure" option
663 "--enable-thread-safety" (Lee Kindness, Philip Yarra)
664 * Allow function pqInternalNotice to accept a format string and
665 arguments instead of just a preformatted message (Tom, Sean
667 * Control SSL negotiation with sslmode values disable, allow,
668 prefer, and require (Jon Jensen)
669 * Allow new error codes and levels of text (Tom)
670 * Allow access to the underlying table and column of a query result
672 This is helpful for query-builder applications that want to know
673 the underlying table and column names associated with a specific
675 * Allow access to the current transaction status (Tom)
676 * Add ability to pass binary data directly to the server (Tom)
677 * Add function PQexecPrepared and PQsendQueryPrepared functions
678 which perform bind/execute of previously prepared statements (Tom)
679 _________________________________________________________________
683 * Allow setNull on updateable result sets
684 * Allow executeBatch on a prepared statement (Barry)
685 * Support SSL connections (Barry)
686 * Handle schema names in result sets (Paul Sorenson)
687 * Add refcursor support (Nic Ferrier)
688 _________________________________________________________________
690 Miscellaneous Interface Changes
692 * Prevent possible memory leak or core dump during libpgtcl shutdown
694 * Add Informix compatibility to ECPG (Michael)
695 This allows ECPG to process embedded C programs that were written
696 using certain Informix extensions.
697 * Add type decimal to ECPG that is fixed length, for Informix
699 * Allow thread-safe embedded SQL programs with "configure" option
700 "--enable-thread-safety" (Lee Kindness, Bruce)
701 This allows multiple threads to access the database at the same
703 * Moved Python client PyGreSQL to http://www.pygresql.org (Marc)
704 _________________________________________________________________
708 * Prevent need for separate platform geometry regression result
710 * Improved PPC locking primitive (Reinhard Max)
711 * New function palloc0 to allocate and clear memory (Bruce)
712 * Fix locking code for s390x CPU (64-bit) (Tom)
713 * Allow OpenBSD to use local ident credentials (William Ahern)
714 * Make query plan trees read-only to executor (Tom)
715 * Add Darwin startup scripts (David Wheeler)
716 * Allow libpq to compile with Borland C++ compiler (Lester Godwin,
718 * Use our own version of getopt_long() if needed (Peter)
719 * Convert administration scripts to C (Peter)
720 * Bison >= 1.85 is now required to build the PostgreSQL grammar, if
722 * Merge documentation into one book (Peter)
723 * Add Windows compatibility functions (Bruce)
724 * Allow client interfaces to compile under MinGW (Bruce)
725 * New ereport() function for error reporting (Tom)
726 * Support Intel compiler on Linux (Peter)
727 * Improve Linux startup scripts (Slawomir Sudnik, Darko Prenosil)
728 * Add support for AMD Opteron and Itanium (Jeffrey W. Baker, Bruce)
729 * Remove "--enable-recode" option from "configure"
730 This was no longer needed now that we have "CREATE CONVERSION".
731 * Generate a compile error if spinlock code is not found (Bruce)
732 Platforms without spinlock code will now fail to compile, rather
733 than silently using semaphores. This failure can be disabled with
734 a new "configure" option.
735 _________________________________________________________________
739 * Change dbmirror license to BSD
740 * Improve earthdistance (Bruno Wolff III)
741 * Portability improvements to pgcrypto (Marko Kreen)
742 * Prevent crash in xml (John Gray, Michael Richards)
745 * Update cube (Bruno Wolff III)
746 * Update earthdistance to use cube (Bruno Wolff III)
747 * Update btree_gist (Oleg)
748 * New tsearch2 full-text search module (Oleg, Teodor)
749 * Add hash-based crosstab function to tablefuncs (Joe)
750 * Add serial column to order connectby() siblings in tablefuncs
752 * Add named persistent connections to dblink (Shridhar Daithanka)
753 * New pg_autovacuum allows automatic "VACUUM" (Matthew T. O'Connor)
754 * Make pgbench honor environment variables PGHOST, PGPORT, PGUSER
756 * Improve intarray (Teodor Sigaev)
757 * Improve pgstattuple (Rod)
758 * Fix bug in metaphone() in fuzzystrmatch
759 * Improve adddepend (Rod)
760 * Update spi/timetravel (B?jthe Zolt?n)
761 * Fix dbase "-s" option and improve non-ASCII handling (Thomas
762 Behr,M?rcio Smiderle)
763 * Remove array module because features now included by default (Joe)
764 _________________________________________________________________
768 Release date: 2003-07-24
770 This has a variety of fixes from 7.3.3.
771 _________________________________________________________________
773 Migration to version 7.3.4
775 A dump/restore is *not* required for those running 7.3.*.
776 _________________________________________________________________
780 * Repair breakage in timestamp-to-date conversion for dates before
782 * Prevent rare possibility of server startup failure (Tom)
783 * Fix bugs in interval-to-time conversion (Tom)
784 * Add constraint names in a few places in pg_dump (Rod)
785 * Improve performance of functions with many parameters (Tom)
786 * Fix to_ascii() buffer overruns (Tom)
787 * Prevent restore of database comments from throwing an error (Tom)
788 * Work around buggy strxfrm() present in some Solaris releases (Tom)
789 * Properly escape jdbc setObject() strings to improve security
791 _________________________________________________________________
795 Release date: 2003-05-22
797 This release contains of variety of fixes for version 7.3.2.
798 _________________________________________________________________
800 Migration to version 7.3.3
802 A dump/restore is *not* required for those running version 7.3.*.
803 _________________________________________________________________
807 * Repair sometimes-incorrect computation of StartUpID after a crash
808 * Avoid slowness with lots of deferred triggers in one transaction
810 * Don't lock referenced row when "UPDATE" doesn't change foreign
812 * Use "-fPIC" not "-fpic" on Sparc (Tom Callaway)
813 * Repair lack of schema-awareness in contrib/reindexdb
814 * Fix contrib/intarray error for zero-element result array (Teodor)
815 * Ensure createuser script will exit on control-C (Oliver)
816 * Fix errors when the type of a dropped column has itself been
818 * "CHECKPOINT" does not cause database panic on failure in
820 * Accept 60 in seconds fields of timestamp, time, interval input
822 * Issue notice, not error, if TIMESTAMP, TIME, or INTERVAL precision
824 * Fix abstime-to-time cast function (fix is not applied unless you
826 * Fix pg_proc entry for timestampt_izone (fix is not applied unless
828 * Make EXTRACT(EPOCH FROM timestamp without time zone) treat input
830 * "'now'::timestamptz" gave wrong answer if timezone changed earlier
832 * HAVE_INT64_TIMESTAMP code for time with timezone overwrote its
834 * Accept "GLOBAL TEMP/TEMPORARY" as a synonym for "TEMPORARY"
835 * Avoid improper schema-privilege-check failure in foreign-key
837 * Fix bugs in foreign-key triggers for "SET DEFAULT" action
838 * Fix incorrect time-qual check in row fetch for "UPDATE" and
840 * Foreign-key clauses were parsed but ignored in "ALTER TABLE ADD
842 * Fix createlang script breakage for case where handler function
844 * Fix misbehavior on zero-column tables in pg_dump, COPY, ANALYZE,
846 * Fix misbehavior of func_error() on type names containing '%'
847 * Fix misbehavior of replace() on strings containing '%'
848 * Regular-expression patterns containing certain multibyte
850 * Account correctly for "NULL"s in more cases in join size
852 * Avoid conflict with system definition of isblank() function or
854 * Fix failure to convert large code point values in EUC_TW
856 * Fix error recovery for SSL_read/SSL_write calls
857 * Don't do early constant-folding of type coercion expressions
858 * Validate page header fields immediately after reading in any page
859 * Repair incorrect check for ungrouped variables in unnamed joins
860 * Fix buffer overrun in to_ascii (Guido Notari)
861 * contrib/ltree fixes (Teodor)
862 * Fix core dump in deadlock detection on machines where char is
864 * Avoid running out of buffers in many-way indexscan (bug introduced
866 * Fix planner's selectivity estimation functions to handle domains
868 * Fix dbmirror memory-allocation bug (Steven Singer)
869 * Prevent infinite loop in ln(numeric) due to roundoff error
870 * "GROUP BY" got confused if there were multiple equal GROUP BY
872 * Fix bad plan when inherited "UPDATE"/"DELETE" references another
874 * Prevent clustering on incomplete (partial or non-NULL-storing)
876 * Service shutdown request at proper time if it arrives while still
878 * Fix left-links in temporary indexes (could make backwards scans
880 * Fix incorrect handling of client_encoding setting in
881 postgresql.conf (Tatsuo)
882 * Fix failure to respond to "pg_ctl stop -m fast" after
883 Async_NotifyHandler runs
884 * Fix SPI for case where rule contains multiple statements of the
886 * Fix problem with checking for wrong type of access privilege in
888 * Fix problem with "EXCEPT" in "CREATE RULE"
889 * Prevent problem with dropping temp tables having serial columns
890 * Fix replace_vars_with_subplan_refs failure in complex views
891 * Fix regexp slowness in single-byte encodings (Tatsuo)
892 * Allow qualified type names in "CREATE CAST" and " DROP CAST"
893 * Accept SETOF type[], which formerly had to be written SETOF _type
894 * Fix pg_dump core dump in some cases with procedural languages
895 * Force ISO datestyle in pg_dump output, for portability (Oliver)
896 * pg_dump failed to handle error return from lo_read (Oleg Drokin)
897 * pg_dumpall failed with groups having no members (Nick Eskelinen)
898 * pg_dumpall failed to recognize --globals-only switch
899 * pg_restore failed to restore blobs if -X disable-triggers is
901 * Repair intrafunction memory leak in plpgsql
902 * pltcl's "elog" command dumped core if given wrong parameters (Ian
904 * plpython used wrong value of atttypmod (Brad McLean)
905 * Fix improper quoting of boolean values in Python interface
907 * Added addDataType() method to PGConnection interface for JDBC
908 * Fixed various problems with updateable ResultSets for JDBC (Shawn
910 * Fixed various problems with DatabaseMetaData for JDBC (Kris Jurka,
912 * Fixed problem with parsing table ACLs in JDBC
913 * Better error message for character set conversion problems in JDBC
914 _________________________________________________________________
918 Release date: 2003-02-04
920 This release contains a variety of fixes for version 7.3.1.
921 _________________________________________________________________
923 Migration to version 7.3.2
925 A dump/restore is *not* required for those running version 7.3.*.
926 _________________________________________________________________
930 * Restore creation of OID column in CREATE TABLE AS / SELECT INTO
931 * Fix pg_dump core dump when dumping views having comments
932 * Dump DEFERRABLE/INITIALLY DEFERRED constraints properly
933 * Fix UPDATE when child table's column numbering differs from parent
934 * Increase default value of max_fsm_relations
935 * Fix problem when fetching backwards in a cursor for a single-row
937 * Make backward fetch work properly with cursor on SELECT DISTINCT
939 * Fix problems with loading pg_dump files containing contrib/lo
941 * Fix problem with all-numeric user names
942 * Fix possible memory leak and core dump during disconnect in
944 * Make plpython's spi_execute command handle nulls properly (Andrew
946 * Adjust plpython error reporting so that its regression test passes
948 * Work with bison 1.875
949 * Handle mixed-case names properly in plpgsql's %type (Neil)
950 * Fix core dump in pltcl when executing a query rewritten by a rule
951 * Repair array subscript overruns (per report from Yichen Xie)
952 * Reduce MAX_TIME_PRECISION from 13 to 10 in floating-point case
953 * Correctly case-fold variable names in per-database and per-user
955 * Fix coredump in plpgsql's RETURN NEXT when SELECT into record
957 * Fix outdated use of pg_type.typprtlen in python client interface
958 * Correctly handle fractional seconds in timestamps in JDBC driver
959 * Improve performance of getImportedKeys() in JDBC
960 * Make shared-library symlinks work standardly on HPUX (Giles)
961 * Repair inconsistent rounding behavior for timestamp, time,
963 * SSL negotiation fixes (Nathan Mueller)
964 * Make libpq's ~/.pgpass feature work when connecting with
966 * Update my2pg, ora2pg
967 * Translation updates
968 * Add casts between types lo and oid in contrib/lo
969 * fastpath code now checks for privilege to call function
970 _________________________________________________________________
974 Release date: 2002-12-18
976 This release contains a variety of fixes for version 7.3.
977 _________________________________________________________________
979 Migration to version 7.3.1
981 A dump/restore is *not* required for those running version 7.3.
982 However, it should be noted that the main PostgreSQL interface
983 library, libpq, has a new major version number for this release, which
984 may require recompilation of client code in certain cases.
985 _________________________________________________________________
989 * Fix a core dump of COPY TO when client/server encodings don't
991 * Allow pg_dump to work with pre-7.2 servers (Philip)
992 * contrib/adddepend fixes (Tom)
993 * Fix problem with deletion of per-user/per-database config settings
995 * contrib/vacuumlo fix (Tom)
996 * Allow 'password' encryption even when pg_shadow contains MD5
998 * contrib/dbmirror fix (Steven Singer)
999 * Optimizer fixes (Tom)
1000 * contrib/tsearch fixes (Teodor Sigaev, Magnus)
1001 * Allow locale names to be mixed case (Nicolai Tufar)
1002 * Increment libpq library's major version number (Bruce)
1003 * pg_hba.conf error reporting fixes (Bruce, Neil)
1004 * Add SCO Openserver 5.0.4 as a supported platform (Bruce)
1005 * Prevent EXPLAIN from crashing server (Tom)
1006 * SSL fixes (Nathan Mueller)
1007 * Prevent composite column creation via ALTER TABLE (Tom)
1008 _________________________________________________________________
1012 Release date: 2002-11-27
1013 _________________________________________________________________
1017 Major changes in this release:
1020 Schemas allow users to create objects in separate namespaces,
1021 so two people or applications can have tables with the same
1022 name. There is also a public schema for shared tables.
1023 Table/index creation can be restricted by removing privileges
1024 on the public schema.
1027 PostgreSQL now supports the ALTER TABLE ... DROP COLUMN
1031 Functions returning multiple rows and/or multiple columns are
1032 now much easier to use than before. You can call such a "table
1033 function" in the SELECT FROM clause, treating its output like a
1034 table. Also, PL/pgSQL functions can now return sets.
1037 PostgreSQL now supports prepared queries, for improved
1041 PostgreSQL now records object dependencies, which allows
1042 improvements in many areas. "DROP" statements now take either
1043 CASCADE or RESTRICT to control whether dependent objects are
1047 Functions and procedural languages now have privileges, and
1048 functions can be defined to run with the privileges of their
1051 Internationalization
1052 Both multibyte and locale support are now always enabled.
1055 A variety of logging options have been enhanced.
1058 A large number of interfaces have been moved to
1059 http://gborg.postgresql.org where they can be developed and
1060 released independently.
1062 Functions/Identifiers
1063 By default, functions can now take up to 32 parameters, and
1064 identifiers can be up to 63 bytes long. Also, OPAQUE is now
1065 deprecated: there are specific "pseudo-datatypes" to represent
1066 each of the former meanings of OPAQUE in function argument and
1068 _________________________________________________________________
1070 Migration to version 7.3
1072 A dump/restore using pg_dump is required for those wishing to migrate
1073 data from any previous release. If your application examines the
1074 system catalogs, additional changes will be required due to the
1075 introduction of schemas in 7.3; for more information, see:
1076 http://developer.postgresql.org/~momjian/upgrade_tips_7.3.
1078 Observe the following incompatibilities:
1080 * Pre-6.3 clients are no longer supported.
1081 * "pg_hba.conf" now has a column for the user name and additional
1082 features. Existing files need to be adjusted.
1083 * Several "postgresql.conf" logging parameters have been renamed.
1084 * LIMIT #,# has been disabled; use LIMIT # OFFSET #.
1085 * "INSERT" statements with column lists must specify a value for
1086 each specified column. For example, INSERT INTO tab (col1, col2)
1087 VALUES ('val1') is now invalid. It's still allowed to supply fewer
1088 columns than expected if the "INSERT" does not have a column list.
1089 * serial columns are no longer automatically UNIQUE; thus, an index
1090 will not automatically be created.
1091 * A "SET" command inside an aborted transaction is now rolled back.
1092 * "COPY" no longer considers missing trailing columns to be null.
1093 All columns need to be specified. (However, one may achieve a
1094 similar effect by specifying a column list in the "COPY" command.)
1095 * The data type timestamp is now equivalent to timestamp without
1096 time zone, instead of timestamp with time zone.
1097 * Pre-7.3 databases loaded into 7.3 will not have the new object
1098 dependencies for serial columns, unique constraints, and foreign
1099 keys. See the directory "contrib/adddepend/" for a detailed
1100 description and a script that will add such dependencies.
1101 * An empty string ('') is no longer allowed as the input into an
1102 integer field. Formerly, it was silently interpreted as 0.
1103 _________________________________________________________________
1109 * Add pg_locks view to show locks (Neil)
1110 * Security fixes for password negotiation memory allocation (Neil)
1111 * Remove support for version 0 FE/BE protocol (PostgreSQL 6.2 and
1113 * Reserve the last few backend slots for superusers, add parameter
1114 superuser_reserved_connections to control this (Nigel J. Andrews)
1115 _________________________________________________________________
1119 * Improve startup by calling localtime() only once (Tom)
1120 * Cache system catalog information in flat files for faster startup
1122 * Improve caching of index information (Tom)
1123 * Optimizer improvements (Tom, Fernando Nasser)
1124 * Catalog caches now store failed lookups (Tom)
1125 * Hash function improvements (Neil)
1126 * Improve performance of query tokenization and network handling
1128 * Speed improvement for large object restore (Mario Weilguni)
1129 * Mark expired index entries on first lookup, saving later heap
1131 * Avoid excessive NULL bitmap padding (Manfred Koizar)
1132 * Add BSD-licensed qsort() for Solaris, for performance (Bruce)
1133 * Reduce per-row overhead by four bytes (Manfred Koizar)
1134 * Fix GEQO optimizer bug (Neil Conway)
1135 * Make WITHOUT OID actually save four bytes per row (Manfred Koizar)
1136 * Add default_statistics_target variable to specify ANALYZE buckets
1138 * Use local buffer cache for temporary tables so no WAL overhead
1140 * Improve free space map performance on large tables (Stephen
1142 * Improved WAL write concurrency (Tom)
1143 _________________________________________________________________
1147 * Add privileges on functions and procedural languages (Peter)
1148 * Add OWNER to CREATE DATABASE so superusers can create databases on
1149 behalf of unprivileged users (Gavin Sherry, Tom)
1150 * Add new object privilege bits EXECUTE and USAGE (Tom)
1151 * Add SET SESSION AUTHORIZATION DEFAULT and RESET SESSION
1153 * Allow functions to be executed with the privilege of the function
1155 _________________________________________________________________
1157 Server Configuration
1159 * Server log messages now tagged with LOG, not DEBUG (Bruce)
1160 * Add user column to pg_hba.conf (Bruce)
1161 * Have log_connections output two lines in log file (Tom)
1162 * Remove debug_level from postgresql.conf, now server_min_messages
1164 * New ALTER DATABASE/USER ... SET command for per-user/database
1165 initialization (Peter)
1166 * New parameters server_min_messages and client_min_messages to
1167 control which messages are sent to the server logs or client
1168 applications (Bruce)
1169 * Allow pg_hba.conf to specify lists of users/databases separated by
1170 commas, group names prepended with +, and file names prepended
1172 * Remove secondary password file capability and pg_password utility
1174 * Add variable db_user_namespace for database-local user names
1176 * SSL improvements (Bear Giles)
1177 * Make encryption of stored passwords the default (Bruce)
1178 * Allow pg_statistics to be reset by calling pg_stat_reset()
1180 * Add log_duration parameter (Bruce)
1181 * Rename debug_print_query to log_statement (Bruce)
1182 * Rename show_query_stats to show_statement_stats (Bruce)
1183 * Add param log_min_error_statement to print commands to logs on
1185 _________________________________________________________________
1189 * Make cursors insensitive, meaning their contents do not change
1191 * Disable LIMIT #,# syntax; now only LIMIT # OFFSET # supported
1193 * Increase identifier length to 63 (Neil, Bruce)
1194 * UNION fixes for merging >= 3 columns of different lengths (Tom)
1195 * Add DEFAULT key word to INSERT, e.g., INSERT ... (..., DEFAULT,
1197 * Allow views to have default values using ALTER COLUMN ... SET
1199 * Fail on INSERTs with column lists that don't supply all column
1200 values, e.g., INSERT INTO tab (col1, col2) VALUES ('val1'); (Rod)
1201 * Fix for join aliases (Tom)
1202 * Fix for FULL OUTER JOINs (Tom)
1203 * Improve reporting of invalid identifier and location (Tom, Gavin)
1204 * Fix OPEN cursor(args) (Tom)
1205 * Allow 'ctid' to be used in a view and currtid(viewname) (Hiroshi)
1206 * Fix for CREATE TABLE AS with UNION (Tom)
1207 * SQL99 syntax improvements (Thomas)
1208 * Add statement_timeout variable to cancel queries (Bruce)
1209 * Allow prepared queries with PREPARE/EXECUTE (Neil)
1210 * Allow FOR UPDATE to appear after LIMIT/OFFSET (Bruce)
1211 * Add variable autocommit (Tom, David Van Wie)
1212 _________________________________________________________________
1216 * Make equals signs optional in CREATE DATABASE (Gavin Sherry)
1217 * Make ALTER TABLE OWNER change index ownership too (Neil)
1218 * New ALTER TABLE tabname ALTER COLUMN colname SET STORAGE controls
1219 TOAST storage, compression (John Gray)
1220 * Add schema support, CREATE/DROP SCHEMA (Tom)
1221 * Create schema for temporary tables (Tom)
1222 * Add variable search_path for schema search (Tom)
1223 * Add ALTER TABLE SET/DROP NOT NULL (Christopher)
1224 * New CREATE FUNCTION volatility levels (Tom)
1225 * Make rule names unique only per table (Tom)
1226 * Add 'ON tablename' clause to DROP RULE and COMMENT ON RULE (Tom)
1227 * Add ALTER TRIGGER RENAME (Joe)
1228 * New current_schema() and current_schemas() inquiry functions (Tom)
1229 * Allow functions to return multiple rows (table functions) (Joe)
1230 * Make WITH optional in CREATE DATABASE, for consistency (Bruce)
1231 * Add object dependency tracking (Rod, Tom)
1232 * Add RESTRICT/CASCADE to DROP commands (Rod)
1233 * Add ALTER TABLE DROP for non-CHECK CONSTRAINT (Rod)
1234 * Autodestroy sequence on DROP of table with SERIAL (Rod)
1235 * Prevent column dropping if column is used by foreign key (Rod)
1236 * Automatically drop constraints/functions when object is dropped
1238 * Add CREATE/DROP OPERATOR CLASS (Bill Studenmund, Tom)
1239 * Add ALTER TABLE DROP COLUMN (Christopher, Tom, Hiroshi)
1240 * Prevent inherited columns from being removed or renamed (Alvaro
1242 * Fix foreign key constraints to not error on intermediate database
1244 * Propagate column or table renaming to foreign key constraints
1245 * Add CREATE OR REPLACE VIEW (Gavin, Neil, Tom)
1246 * Add CREATE OR REPLACE RULE (Gavin, Neil, Tom)
1247 * Have rules execute alphabetically, returning more predictable
1249 * Triggers are now fired in alphabetical order (Tom)
1250 * Add /contrib/adddepend to handle pre-7.3 object dependencies (Rod)
1251 * Allow better casting when inserting/updating values (Tom)
1252 _________________________________________________________________
1256 * Have COPY TO output embedded carriage returns and newlines as \r
1258 * Allow DELIMITER in COPY FROM to be 8-bit clean (Tatsuo)
1259 * Make pg_dump use ALTER TABLE ADD PRIMARY KEY, for performance
1261 * Disable brackets in multistatement rules (Bruce)
1262 * Disable VACUUM from being called inside a function (Bruce)
1263 * Allow dropdb and other scripts to use identifiers with spaces
1265 * Restrict database comment changes to the current database
1266 * Allow comments on operators, independent of the underlying
1268 * Rollback SET commands in aborted transactions (Tom)
1269 * EXPLAIN now outputs as a query (Tom)
1270 * Display condition expressions and sort keys in EXPLAIN (Tom)
1271 * Add 'SET LOCAL var = value' to set configuration variables for a
1272 single transaction (Tom)
1273 * Allow ANALYZE to run in a transaction (Bruce)
1274 * Improve COPY syntax using new WITH clauses, keep backward
1275 compatibility (Bruce)
1276 * Fix pg_dump to consistently output tags in non-ASCII dumps (Bruce)
1277 * Make foreign key constraints clearer in dump file (Rod)
1278 * Add COMMENT ON CONSTRAINT (Rod)
1279 * Allow COPY TO/FROM to specify column names (Brent Verner)
1280 * Dump UNIQUE and PRIMARY KEY constraints as ALTER TABLE (Rod)
1281 * Have SHOW output a query result (Joe)
1282 * Generate failure on short COPY lines rather than pad NULLs (Neil)
1283 * Fix CLUSTER to preserve all table attributes (Alvaro Herrera)
1284 * New pg_settings table to view/modify GUC settings (Joe)
1285 * Add smart quoting, portability improvements to pg_dump output
1287 * Dump serial columns out as SERIAL (Tom)
1288 * Enable large file support, >2G for pg_dump (Peter, Philip Warner,
1290 * Disallow TRUNCATE on tables that are involved in referential
1292 * Have TRUNCATE also auto-truncate the toast table of the relation
1294 * Add clusterdb utility that will auto-cluster an entire database
1295 based on previous CLUSTER operations (Alvaro Herrera)
1296 * Overhaul pg_dumpall (Peter)
1297 * Allow REINDEX of TOAST tables (Tom)
1298 * Implemented START TRANSACTION, per SQL99 (Neil)
1299 * Fix rare index corruption when a page split affects bulk delete
1301 * Fix ALTER TABLE ... ADD COLUMN for inheritance (Alvaro Herrera)
1302 _________________________________________________________________
1304 Data Types and Functions
1306 * Fix factorial(0) to return 1 (Bruce)
1307 * Date/time/timezone improvements (Thomas)
1308 * Fix for array slice extraction (Tom)
1309 * Fix extract/date_part to report proper microseconds for timestamp
1311 * Allow text_substr() and bytea_substr() to read TOAST values more
1312 efficiently (John Gray)
1313 * Add domain support (Rod)
1314 * Make WITHOUT TIME ZONE the default for TIMESTAMP and TIME data
1316 * Allow alternate storage scheme of 64-bit integers for date/time
1317 types using --enable-integer-datetimes in configure (Thomas)
1318 * Make timezone(timestamptz) return timestamp rather than a string
1320 * Allow fractional seconds in date/time types for dates prior to 1BC
1322 * Limit timestamp data types to 6 decimal places of precision
1324 * Change timezone conversion functions from timetz() to timezone()
1326 * Add configuration variables datestyle and timezone (Tom)
1327 * Add OVERLAY(), which allows substitution of a substring in a
1329 * Add SIMILAR TO (Thomas, Tom)
1330 * Add regular expression SUBSTRING(string FROM pat FOR escape)
1332 * Add LOCALTIME and LOCALTIMESTAMP functions (Thomas)
1333 * Add named composite types using CREATE TYPE typename AS (column)
1335 * Allow composite type definition in the table alias clause (Joe)
1336 * Add new API to simplify creation of C language table functions
1338 * Remove ODBC-compatible empty parentheses from calls to SQL99
1339 functions for which these parentheses do not match the standard
1341 * Allow macaddr data type to accept 12 hex digits with no separators
1343 * Add CREATE/DROP CAST (Peter)
1344 * Add IS DISTINCT FROM operator (Thomas)
1345 * Add SQL99 TREAT() function, synonym for CAST() (Thomas)
1346 * Add pg_backend_pid() to output backend pid (Bruce)
1347 * Add IS OF / IS NOT OF type predicate (Thomas)
1348 * Allow bit string constants without fully-specified length (Thomas)
1349 * Allow conversion between 8-byte integers and bit strings (Thomas)
1350 * Implement hex literal conversion to bit string literal (Thomas)
1351 * Allow table functions to appear in the FROM clause (Joe)
1352 * Increase maximum number of function parameters to 32 (Bruce)
1353 * No longer automatically create index for SERIAL column (Tom)
1354 * Add current_database() (Rod)
1355 * Fix cash_words() to not overflow buffer (Tom)
1356 * Add functions replace(), split_part(), to_hex() (Joe)
1357 * Fix LIKE for bytea as a right-hand argument (Joe)
1358 * Prevent crashes caused by SELECT cash_out(2) (Tom)
1359 * Fix to_char(1,'FM999.99') to return a period (Karel)
1360 * Fix trigger/type/language functions returning OPAQUE to return
1362 _________________________________________________________________
1364 Internationalization
1366 * Add additional encodings: Korean (JOHAB), Thai (WIN874),
1367 Vietnamese (TCVN), Arabic (WIN1256), Simplified Chinese (GBK),
1368 Korean (UHC) (Eiji Tokuya)
1369 * Enable locale support by default (Peter)
1370 * Add locale variables (Peter)
1371 * Escape byes >= 0x7f for multibyte in PQescapeBytea/PQunescapeBytea
1373 * Add locale awareness to regular expression character classes
1374 * Enable multibyte support by default (Tatsuo)
1375 * Add GB18030 multibyte support (Bill Huang)
1376 * Add CREATE/DROP CONVERSION, allowing loadable encodings (Tatsuo,
1378 * Add pg_conversion table (Tatsuo)
1379 * Add SQL99 CONVERT() function (Tatsuo)
1380 * pg_dumpall, pg_controldata, and pg_resetxlog now national-language
1382 * New and updated translations
1383 _________________________________________________________________
1385 Server-side Languages
1387 * Allow recursive SQL function (Peter)
1388 * Change PL/Tcl build to use configured compiler and Makefile.shlib
1390 * Overhaul the PL/pgSQL FOUND variable to be more Oracle-compatible
1392 * Allow PL/pgSQL to handle quoted identifiers (Tom)
1393 * Allow set-returning PL/pgSQL functions (Neil)
1394 * Make PL/pgSQL schema-aware (Joe)
1395 * Remove some memory leaks (Nigel J. Andrews, Tom)
1396 _________________________________________________________________
1400 * Don't lowercase psql \connect database name for 7.2.0
1402 * Add psql \timing to time user queries (Greg Sabino Mullane)
1403 * Have psql \d show index information (Greg Sabino Mullane)
1404 * New psql \dD shows domains (Jonathan Eisler)
1405 * Allow psql to show rules on views (Paul ?)
1406 * Fix for psql variable substitution (Tom)
1407 * Allow psql \d to show temporary table structure (Tom)
1408 * Allow psql \d to show foreign keys (Rod)
1409 * Fix \? to honor \pset pager (Bruce)
1410 * Have psql reports its version number on startup (Tom)
1411 * Allow \copy to specify column names (Tom)
1412 _________________________________________________________________
1416 * Add $HOME/.pgpass to store host/user password combinations (Alvaro
1418 * Add PQunescapeBytea() function to libpq (Patrick Welche)
1419 * Fix for sending large queries over non-blocking connections
1421 * Fix for libpq using timers on Win9X (David Ford)
1422 * Allow libpq notify to handle servers with different-length
1424 * Add libpq PQescapeString() and PQescapeBytea() to Windows (Bruce)
1425 * Fix for SSL with non-blocking connections (Jack Bates)
1426 * Add libpq connection timeout parameter (Denis A Ustimenko)
1427 _________________________________________________________________
1431 * Allow JDBC to compile with JDK 1.4 (Dave)
1432 * Add JDBC 3 support (Barry)
1433 * Allows JDBC to set loglevel by adding ?loglevel=X to the
1434 connection URL (Barry)
1435 * Add Driver.info() message that prints out the version number
1437 * Add updateable result sets (Raghu Nidagal, Dave)
1438 * Add support for callable statements (Paul Bethe)
1439 * Add query cancel capability
1440 * Add refresh row (Dave)
1441 * Fix MD5 encryption handling for multibyte servers (Jun Kawai)
1442 * Add support for prepared statements (Barry)
1443 _________________________________________________________________
1445 Miscellaneous Interfaces
1447 * Fixed ECPG bug concerning octal numbers in single quotes (Michael)
1448 * Move src/interfaces/libpgeasy to http://gborg.postgresql.org
1450 * Improve Python interface (Elliot Lee, Andrew Johnson, Greg
1452 * Add libpgtcl connection close event (Gerhard Hintermayer)
1453 * Move src/interfaces/libpq++ to http://gborg.postgresql.org (Marc,
1455 * Move src/interfaces/odbc to http://gborg.postgresql.org (Marc)
1456 * Move src/interfaces/libpgeasy to http://gborg.postgresql.org
1458 * Move src/interfaces/perl5 to http://gborg.postgresql.org (Marc,
1460 * Remove src/bin/pgaccess from main tree, now at
1461 http://www.pgaccess.org (Bruce)
1462 * Add pg_on_connection_loss command to libpgtcl (Gerhard
1464 _________________________________________________________________
1468 * Fix for parallel make (Peter)
1469 * AIX fixes for linking Tcl (Andreas Zeugswetter)
1470 * Allow PL/Perl to build under Cygwin (Jason Tishler)
1471 * Improve MIPS compiles (Peter, Oliver Elphick)
1472 * Require Autoconf version 2.53 (Peter)
1473 * Require readline and zlib by default in configure (Peter)
1474 * Allow Solaris to use Intimate Shared Memory (ISM), for performance
1475 (Scott Brunza, P.J. Josh Rovero)
1476 * Always enable syslog in compile, remove --enable-syslog option
1478 * Always enable multibyte in compile, remove --enable-multibyte
1480 * Always enable locale in compile, remove --enable-locale option
1482 * Fix for Win9x DLL creation (Magnus Naeslund)
1483 * Fix for link() usage by WAL code on Windows, BeOS (Jason Tishler)
1484 * Add sys/types.h to c.h, remove from main files (Peter, Bruce)
1485 * Fix AIX hang on SMP machines (Tomoyuki Niijima)
1486 * AIX SMP hang fix (Tomoyuki Niijima)
1487 * Fix pre-1970 date handling on newer glibc libraries (Tom)
1488 * Fix PowerPC SMP locking (Tom)
1489 * Prevent gcc -ffast-math from being used (Peter, Tom)
1490 * Bison >= 1.50 now required for developer builds
1491 * Kerberos 5 support now builds with Heimdal (Peter)
1492 * Add appendix in the User's Guide which lists SQL features (Thomas)
1493 * Improve loadable module linking to use RTLD_NOW (Tom)
1494 * New error levels WARNING, INFO, LOG, DEBUG[1-5] (Bruce)
1495 * New src/port directory holds replaced libc functions (Peter,
1497 * New pg_namespace system catalog for schemas (Tom)
1498 * Add pg_class.relnamespace for schemas (Tom)
1499 * Add pg_type.typnamespace for schemas (Tom)
1500 * Add pg_proc.pronamespace for schemas (Tom)
1501 * Restructure aggregates to have pg_proc entries (Tom)
1502 * System relations now have their own namespace, pg_* test not
1503 required (Fernando Nasser)
1504 * Rename TOAST index names to be *_index rather than *_idx (Neil)
1505 * Add namespaces for operators, opclasses (Tom)
1506 * Add additional checks to server control file (Thomas)
1507 * New Polish FAQ (Marcin Mazurek)
1508 * Add Posix semaphore support (Tom)
1509 * Document need for reindex (Bruce)
1510 * Rename some internal identifiers to simplify Windows compile (Jan,
1512 * Add documentation on computing disk space (Bruce)
1513 * Remove KSQO from GUC (Bruce)
1514 * Fix memory leak in rtree (Kenneth Been)
1515 * Modify a few error messages for consistency (Bruce)
1516 * Remove unused system table columns (Peter)
1517 * Make system columns NOT NULL where appropriate (Tom)
1518 * Clean up use of sprintf in favor of snprintf() (Neil, Jukka
1520 * Remove OPAQUE and create specific subtypes (Tom)
1521 * Cleanups in array internal handling (Joe, Tom)
1522 * Disallow pg_atoi('') (Bruce)
1523 * Remove parameter wal_files because WAL files are now recycled
1525 * Add version numbers to heap pages (Tom)
1526 _________________________________________________________________
1530 * Allow inet arrays in /contrib/array (Neil)
1531 * GiST fixes (Teodor Sigaev, Neil)
1532 * Upgrade /contrib/mysql
1533 * Add /contrib/dbsize which shows table sizes without vacuum (Peter)
1534 * Add /contrib/intagg, integer aggregator routines (mlw)
1535 * Improve /contrib/oid2name (Neil, Bruce)
1536 * Improve /contrib/tsearch (Oleg, Teodor Sigaev)
1537 * Cleanups of /contrib/rserver (Alexey V. Borzov)
1538 * Update /contrib/oracle conversion utility (Gilles Darold)
1539 * Update /contrib/dblink (Joe)
1540 * Improve options supported by /contrib/vacuumlo (Mario Weilguni)
1541 * Improvements to /contrib/intarray (Oleg, Teodor Sigaev, Andrey
1543 * Add /contrib/reindexdb utility (Shaun Thomas)
1544 * Add indexing to /contrib/isbn_issn (Dan Weston)
1545 * Add /contrib/dbmirror (Steven Singer)
1546 * Improve /contrib/pgbench (Neil)
1547 * Add /contrib/tablefunc table function examples (Joe)
1548 * Add /contrib/ltree data type for tree structures (Teodor Sigaev,
1550 * Move /contrib/pg_controldata, pg_resetxlog into main tree (Bruce)
1551 * Fixes to /contrib/cube (Bruno Wolff)
1552 * Improve /contrib/fulltextindex (Christopher)
1553 _________________________________________________________________
1557 Release date: 2003-01-30
1559 This release contains a variety of fixes for version 7.2.3, including
1560 fixes to prevent possible data loss.
1561 _________________________________________________________________
1563 Migration to version 7.2.4
1565 A dump/restore is *not* required for those running version 7.2.*.
1566 _________________________________________________________________
1570 * Fix some additional cases of VACUUM "No one parent tuple was
1572 * Prevent VACUUM from being called inside a function (Bruce)
1573 * Ensure pg_clog updates are sync'd to disk before marking
1575 * Avoid integer overflow during large hash joins
1576 * Make GROUP commands work when pg_group.grolist is large enough to
1578 * Fix errors in datetime tables; some timezone names weren't being
1580 * Fix integer overflows in circle_poly(), path_encode(), path_add()
1582 * Repair long-standing logic errors in lseg_eq(), lseg_ne(),
1584 _________________________________________________________________
1588 Release date: 2002-10-01
1590 This release contains a variety of fixes for version 7.2.2, including
1591 fixes to prevent possible data loss.
1592 _________________________________________________________________
1594 Migration to version 7.2.3
1596 A dump/restore is *not* required for those running version 7.2.*.
1597 _________________________________________________________________
1601 * Prevent possible compressed transaction log loss (Tom)
1602 * Prevent non-superuser from increasing most recent vacuum info
1604 * Handle pre-1970 date values in newer versions of glibc (Tom)
1605 * Fix possible hang during server shutdown
1606 * Prevent spinlock hangs on SMP PPC machines (Tomoyuki Niijima)
1607 * Fix pg_dump to properly dump FULL JOIN USING (Tom)
1608 _________________________________________________________________
1612 Release date: 2002-08-23
1614 This release contains a variety of fixes for version 7.2.1.
1615 _________________________________________________________________
1617 Migration to version 7.2.2
1619 A dump/restore is *not* required for those running version 7.2.*.
1620 _________________________________________________________________
1624 * Allow EXECUTE of "CREATE TABLE AS ... SELECT" in PL/pgSQL (Tom)
1625 * Fix for compressed transaction log id wraparound (Tom)
1626 * Fix PQescapeBytea/PQunescapeBytea so that they handle bytes > 0x7f
1628 * Fix for psql and pg_dump crashing when invoked with non-existent
1629 long options (Tatsuo)
1630 * Fix crash when invoking geometric operators (Tom)
1631 * Allow OPEN cursor(args) (Tom)
1632 * Fix for rtree_gist index build (Teodor)
1633 * Fix for dumping user-defined aggregates (Tom)
1634 * contrib/intarray fixes (Oleg)
1635 * Fix for complex UNION/EXCEPT/INTERSECT queries using parens (Tom)
1636 * Fix to pg_convert (Tatsuo)
1637 * Fix for crash with long DATA strings (Thomas, Neil)
1638 * Fix for repeat(), lpad(), rpad() and long strings (Neil)
1639 _________________________________________________________________
1643 Release date: 2002-03-21
1645 This release contains a variety of fixes for version 7.2.
1646 _________________________________________________________________
1648 Migration to version 7.2.1
1650 A dump/restore is *not* required for those running version 7.2.
1651 _________________________________________________________________
1655 * Ensure that sequence counters do not go backwards after a crash
1657 * Fix pgaccess kanji-conversion key binding (Tatsuo)
1658 * Optimizer improvements (Tom)
1659 * Cash I/O improvements (Tom)
1661 * Compile fix for missing AuthBlockSig (Heiko)
1662 * Additional time zones and time zone fixes (Thomas)
1663 * Allow psql \connect to handle mixed case database and user names
1665 * Return proper OID on command completion even with ON INSERT rules
1667 * Allow COPY FROM to use 8-bit DELIMITERS (Tatsuo)
1668 * Fix bug in extract/date_part for milliseconds/microseconds
1670 * Improve handling of multiple UNIONs with different lengths (Tom)
1671 * contrib/btree_gist improvements (Teodor Sigaev)
1672 * contrib/tsearch dictionary improvements, see README.tsearch for an
1673 additional installation step (Thomas T. Thai, Teodor Sigaev)
1674 * Fix for array subscripts handling (Tom)
1675 * Allow EXECUTE of "CREATE TABLE AS ... SELECT" in PL/pgSQL (Tom)
1676 _________________________________________________________________
1680 Release date: 2002-02-04
1681 _________________________________________________________________
1685 This release improves PostgreSQL for use in high-volume applications.
1687 Major changes in this release:
1690 Vacuuming no longer locks tables, thus allowing normal user
1691 access during the vacuum. A new "VACUUM FULL" command does
1692 old-style vacuum by locking the table and shrinking the on-disk
1696 There is no longer a problem with installations that exceed
1697 four billion transactions.
1700 OIDs are now optional. Users can now create tables without OIDs
1701 for cases where OID usage is excessive.
1704 The system now computes histogram column statistics during
1705 "ANALYZE", allowing much better optimizer choices.
1708 A new MD5 encryption option allows more secure storage and
1709 transfer of passwords. A new Unix-domain socket authentication
1710 option is available on Linux and BSD systems.
1713 Administrators can use the new table access statistics module
1714 to get fine-grained information about table and index usage.
1716 Internationalization
1717 Program and library messages can now be displayed in several
1719 _________________________________________________________________
1721 Migration to version 7.2
1723 A dump/restore using "pg_dump" is required for those wishing to
1724 migrate data from any previous release.
1726 Observe the following incompatibilities:
1728 * The semantics of the "VACUUM" command have changed in this
1729 release. You may wish to update your maintenance procedures
1731 * In this release, comparisons using = NULL will always return false
1732 (or NULL, more precisely). Previous releases automatically
1733 transformed this syntax to IS NULL. The old behavior can be
1734 re-enabled using a "postgresql.conf" parameter.
1735 * The "pg_hba.conf" and "pg_ident.conf" configuration is now only
1736 reloaded after receiving a SIGHUP signal, not with each
1738 * The function "octet_length()" now returns the uncompressed data
1740 * The date/time value 'current' is no longer available. You will
1741 need to rewrite your applications.
1742 * The timestamp(), time(), and interval() functions are no longer
1743 available. Instead of timestamp(), use timestamp 'string' or CAST.
1745 The SELECT ... LIMIT #,# syntax will be removed in the next release.
1746 You should change your queries to use separate LIMIT and OFFSET
1747 clauses, e.g. LIMIT 10 OFFSET 20.
1748 _________________________________________________________________
1754 * Create temporary files in a separate directory (Bruce)
1755 * Delete orphaned temporary files on postmaster startup (Bruce)
1756 * Added unique indexes to some system tables (Tom)
1757 * System table operator reorganization (Oleg Bartunov, Teodor
1759 * Renamed pg_log to pg_clog (Tom)
1760 * Enable SIGTERM, SIGQUIT to kill backends (Jan)
1761 * Removed compile-time limit on number of backends (Tom)
1762 * Better cleanup for semaphore resource failure (Tatsuo, Tom)
1763 * Allow safe transaction ID wraparound (Tom)
1764 * Removed OIDs from some system tables (Tom)
1765 * Removed "triggered data change violation" error check (Tom)
1766 * SPI portal creation of prepared/saved plans (Jan)
1767 * Allow SPI column functions to work for system columns (Tom)
1768 * Long value compression improvement (Tom)
1769 * Statistics collector for table, index access (Jan)
1770 * Truncate extra-long sequence names to a reasonable value (Tom)
1771 * Measure transaction times in milliseconds (Thomas)
1772 * Fix TID sequential scans (Hiroshi)
1773 * Superuser ID now fixed at 1 (Peter E)
1774 * New pg_ctl "reload" option (Tom)
1775 _________________________________________________________________
1779 * Optimizer improvements (Tom)
1780 * New histogram column statistics for optimizer (Tom)
1781 * Reuse write-ahead log files rather than discarding them (Tom)
1782 * Cache improvements (Tom)
1783 * IS NULL, IS NOT NULL optimizer improvement (Tom)
1784 * Improve lock manager to reduce lock contention (Tom)
1785 * Keep relcache entries for index access support functions (Tom)
1786 * Allow better selectivity with NaN and infinities in NUMERIC (Tom)
1787 * R-tree performance improvements (Kenneth Been)
1788 * B-tree splits more efficient (Tom)
1789 _________________________________________________________________
1793 * Change UPDATE, DELETE privileges to be distinct (Peter E)
1794 * New REFERENCES, TRIGGER privileges (Peter E)
1795 * Allow GRANT/REVOKE to/from more than one user at a time (Peter E)
1796 * New has_table_privilege() function (Joe Conway)
1797 * Allow non-superuser to vacuum database (Tom)
1798 * New SET SESSION AUTHORIZATION command (Peter E)
1799 * Fix bug in privilege modifications on newly created tables (Tom)
1800 * Disallow access to pg_statistic for non-superuser, add
1801 user-accessible views (Tom)
1802 _________________________________________________________________
1804 Client Authentication
1806 * Fork postmaster before doing authentication to prevent hangs
1808 * Add ident authentication over Unix domain sockets on Linux, *BSD
1809 (Helge Bahmann, Oliver Elphick, Teodor Sigaev, Bruce)
1810 * Add a password authentication method that uses MD5 encryption
1812 * Allow encryption of stored passwords using MD5 (Bruce)
1813 * PAM authentication (Dominic J. Eidson)
1814 * Load pg_hba.conf and pg_ident.conf only on startup and SIGHUP
1816 _________________________________________________________________
1818 Server Configuration
1820 * Interpretation of some time zone abbreviations as Australian
1821 rather than North American now settable at run time (Bruce)
1822 * New parameter to set default transaction isolation level (Peter E)
1823 * New parameter to enable conversion of "expr = NULL" into "expr IS
1824 NULL", off by default (Peter E)
1825 * New parameter to control memory usage by VACUUM (Tom)
1826 * New parameter to set client authentication timeout (Tom)
1827 * New parameter to set maximum number of open files (Tom)
1828 _________________________________________________________________
1832 * Statements added by INSERT rules now execute after the INSERT
1834 * Prevent unadorned relation names in target list (Bruce)
1835 * NULLs now sort after all normal values in ORDER BY (Tom)
1836 * New IS UNKNOWN, IS NOT UNKNOWN Boolean tests (Tom)
1837 * New SHARE UPDATE EXCLUSIVE lock mode (Tom)
1838 * New EXPLAIN ANALYZE command that shows run times and row counts
1839 (Martijn van Oosterhout)
1840 * Fix problem with LIMIT and subqueries (Tom)
1841 * Fix for LIMIT, DISTINCT ON pushed into subqueries (Tom)
1842 * Fix nested EXCEPT/INTERSECT (Tom)
1843 _________________________________________________________________
1847 * Fix SERIAL in temporary tables (Bruce)
1848 * Allow temporary sequences (Bruce)
1849 * Sequences now use int8 internally (Tom)
1850 * New SERIAL8 creates int8 columns with sequences, default still
1852 * Make OIDs optional using WITHOUT OIDS (Tom)
1853 * Add %TYPE syntax to CREATE TYPE (Ian Lance Taylor)
1854 * Add ALTER TABLE / DROP CONSTRAINT for CHECK constraints
1855 (Christopher Kings-Lynne)
1856 * New CREATE OR REPLACE FUNCTION to alter existing function
1857 (preserving the function OID) (Gavin Sherry)
1858 * Add ALTER TABLE / ADD [ UNIQUE | PRIMARY ] (Christopher
1860 * Allow column renaming in views
1861 * Make ALTER TABLE / RENAME COLUMN update column names of indexes
1863 * Fix for ALTER TABLE / ADD CONSTRAINT ... CHECK with inherited
1864 tables (Stephan Szabo)
1865 * ALTER TABLE RENAME update foreign-key trigger arguments correctly
1867 * DROP AGGREGATE and COMMENT ON AGGREGATE now accept an aggtype
1869 * Add automatic return type data casting for SQL functions (Tom)
1870 * Allow GiST indexes to handle NULLs and multikey indexes (Oleg
1871 Bartunov, Teodor Sigaev, Tom)
1872 * Enable partial indexes (Martijn van Oosterhout)
1873 _________________________________________________________________
1877 * Add RESET ALL, SHOW ALL (Marko Kreen)
1878 * CREATE/ALTER USER/GROUP now allow options in any order (Vince)
1879 * Add LOCK A, B, C functionality (Neil Padgett)
1880 * New ENCRYPTED/UNENCRYPTED option to CREATE/ALTER USER (Bruce)
1881 * New light-weight VACUUM does not lock table; old semantics are
1882 available as VACUUM FULL (Tom)
1883 * Disable COPY TO/FROM on views (Bruce)
1884 * COPY DELIMITERS string must be exactly one character (Tom)
1885 * VACUUM warning about index tuples fewer than heap now only appears
1886 when appropriate (Martijn van Oosterhout)
1887 * Fix privilege checks for CREATE INDEX (Tom)
1888 * Disallow inappropriate use of CREATE/DROP INDEX/TRIGGER/VIEW (Tom)
1889 _________________________________________________________________
1891 Data Types and Functions
1893 * SUM(), AVG(), COUNT() now uses int8 internally for speed (Tom)
1894 * Add convert(), convert2() (Tatsuo)
1895 * New function bit_length() (Peter E)
1896 * Make the "n" in CHAR(n)/VARCHAR(n) represents letters, not bytes
1898 * CHAR(), VARCHAR() now reject strings that are too long (Peter E)
1899 * BIT VARYING now rejects bit strings that are too long (Peter E)
1900 * BIT now rejects bit strings that do not match declared size (Peter
1902 * INET, CIDR text conversion functions (Alex Pilosov)
1903 * INET, CIDR operators << and <<= indexable (Alex Pilosov)
1904 * Bytea \### now requires valid three digit octal number
1905 * Bytea comparison improvements, now supports =, <>, >, >=, <, and
1907 * Bytea now supports B-tree indexes
1908 * Bytea now supports LIKE, LIKE...ESCAPE, NOT LIKE, NOT
1910 * Bytea now supports concatenation
1911 * New bytea functions: position, substring, trim, btrim, and length
1912 * New encode() function mode, "escaped", converts minimally escaped
1914 * Add pg_database_encoding_max_length() (Tatsuo)
1915 * Add pg_client_encoding() function (Tatsuo)
1916 * now() returns time with millisecond precision (Thomas)
1917 * New TIMESTAMP WITHOUT TIMEZONE data type (Thomas)
1918 * Add ISO date/time specification with "T", yyyy-mm-ddThh:mm:ss
1920 * New xid/int comparison functions (Hiroshi)
1921 * Add precision to TIME, TIMESTAMP, and INTERVAL data types (Thomas)
1922 * Modify type coercion logic to attempt binary-compatible functions
1924 * New encode() function installed by default (Marko Kreen)
1925 * Improved to_*() conversion functions (Karel Zak)
1926 * Optimize LIKE/ILIKE when using single-byte encodings (Tatsuo)
1927 * New functions in contrib/pgcrypto: crypt(), hmac(), encrypt(),
1928 gen_salt() (Marko Kreen)
1929 * Correct description of translate() function (Bruce)
1930 * Add INTERVAL argument for SET TIME ZONE (Thomas)
1931 * Add INTERVAL YEAR TO MONTH (etc.) syntax (Thomas)
1932 * Optimize length functions when using single-byte encodings
1934 * Fix path_inter, path_distance, path_length, dist_ppath to handle
1935 closed paths (Curtis Barrett, Tom)
1936 * octet_length(text) now returns non-compressed length (Tatsuo,
1938 * Handle "July" full name in date/time literals (Greg Sabino
1940 * Some datatype() function calls now evaluated differently
1941 * Add support for Julian and ISO time specifications (Thomas)
1942 _________________________________________________________________
1944 Internationalization
1946 * National language support in psql, pg_dump, libpq, and server
1948 * Message translations in Chinese (simplified, traditional), Czech,
1949 French, German, Hungarian, Russian, Swedish (Peter E, Serguei A.
1950 Mokhov, Karel Zak, Weiping He, Zhenbang Wei, Kovacs Zoltan)
1951 * Make trim, ltrim, rtrim, btrim, lpad, rpad, translate multibyte
1953 * Add LATIN5,6,7,8,9,10 support (Tatsuo)
1954 * Add ISO 8859-5,6,7,8 support (Tatsuo)
1955 * Correct LATIN5 to mean ISO-8859-9, not ISO-8859-5 (Tatsuo)
1956 * Make mic2ascii() non-ASCII aware (Tatsuo)
1957 * Reject invalid multibyte character sequences (Tatsuo)
1958 _________________________________________________________________
1962 * Now uses portals for SELECT loops, allowing huge result sets (Jan)
1963 * CURSOR and REFCURSOR support (Jan)
1964 * Can now return open cursors (Jan)
1965 * Add ELSEIF (Klaus Reger)
1966 * Improve PL/pgSQL error reporting, including location of error
1968 * Allow IS or FOR key words in cursor declaration, for compatibility
1970 * Fix for SELECT ... FOR UPDATE (Tom)
1971 * Fix for PERFORM returning multiple rows (Tom)
1972 * Make PL/pgSQL use the server's type coercion code (Tom)
1973 * Memory leak fix (Jan, Tom)
1974 * Make trailing semicolon optional (Tom)
1975 _________________________________________________________________
1979 * New untrusted PL/Perl (Alex Pilosov)
1980 * PL/Perl is now built on some platforms even if libperl is not
1982 _________________________________________________________________
1986 * Now reports errorInfo (Vsevolod Lobko)
1987 * Add spi_lastoid function (bob@redivi.com)
1988 _________________________________________________________________
1992 * ...is new (Andrew Bosma)
1993 _________________________________________________________________
1997 * \d displays indexes in unique, primary groupings (Christopher
1999 * Allow trailing semicolons in backslash commands (Greg Sabino
2001 * Read password from /dev/tty if possible
2002 * Force new password prompt when changing user and database (Tatsuo,
2004 * Format the correct number of columns for Unicode (Patrice)
2005 _________________________________________________________________
2009 * New function PQescapeString() to escape quotes in command strings
2011 * New function PQescapeBytea() escapes binary strings for use as SQL
2013 _________________________________________________________________
2017 * Return OID of INSERT (Ken K)
2018 * Handle more data types (Ken K)
2019 * Handle single quotes and newlines in strings (Ken K)
2020 * Handle NULL variables (Ken K)
2021 * Fix for time zone handling (Barry Lind)
2022 * Improved Druid support
2023 * Allow eight-bit characters with non-multibyte server (Barry Lind)
2024 * Support BIT, BINARY types (Ned Wolpert)
2025 * Reduce memory usage (Michael Stephens, Dave Cramer)
2026 * Update DatabaseMetaData (Peter E)
2027 * Add DatabaseMetaData.getCatalogs() (Peter E)
2028 * Encoding fixes (Anders Bengtsson)
2029 * Get/setCatalog methods (Jason Davies)
2030 * DatabaseMetaData.getColumns() now returns column defaults (Jason
2032 * DatabaseMetaData.getColumns() performance improvement (Jeroen van
2034 * Some JDBC1 and JDBC2 merging (Anders Bengtsson)
2035 * Transaction performance improvements (Barry Lind)
2036 * Array fixes (Greg Zoller)
2037 * Serialize addition
2038 * Fix batch processing (Rene Pijlman)
2039 * ExecSQL method reorganization (Anders Bengtsson)
2040 * GetColumn() fixes (Jeroen van Vianen)
2041 * Fix isWriteable() function (Rene Pijlman)
2042 * Improved passage of JDBC2 conformance tests (Rene Pijlman)
2043 * Add bytea type capability (Barry Lind)
2044 * Add isNullable() (Rene Pijlman)
2045 * JDBC date/time test suite fixes (Liam Stewart)
2046 * Fix for SELECT 'id' AS xxx FROM table (Dave Cramer)
2047 * Fix DatabaseMetaData to show precision properly (Mark Lillywhite)
2048 * New getImported/getExported keys (Jason Davies)
2049 * MD5 password encryption support (Jeremy Wohl)
2050 * Fix to actually use type cache (Ned Wolpert)
2051 _________________________________________________________________
2055 * Remove query size limit (Hiroshi)
2056 * Remove text field size limit (Hiroshi)
2057 * Fix for SQLPrimaryKeys in multibyte mode (Hiroshi)
2058 * Allow ODBC procedure calls (Hiroshi)
2059 * Improve boolean handing (Aidan Mountford)
2060 * Most configuration options now settable via DSN (Hiroshi)
2061 * Multibyte, performance fixes (Hiroshi)
2062 * Allow driver to be used with iODBC or unixODBC (Peter E)
2063 * MD5 password encryption support (Bruce)
2064 * Add more compatibility functions to odbc.sql (Peter E)
2065 _________________________________________________________________
2069 * EXECUTE ... INTO implemented (Christof Petig)
2070 * Multiple row descriptor support (e.g. CARDINALITY) (Christof
2072 * Fix for GRANT parameters (Lee Kindness)
2073 * Fix INITIALLY DEFERRED bug
2074 * Various bug fixes (Michael, Christof Petig)
2075 * Auto allocation for indicator variable arrays (int *ind_p=NULL)
2076 * Auto allocation for string arrays (char **foo_pp=NULL)
2077 * ECPGfree_auto_mem fixed
2078 * All function names with external linkage are now prefixed by ECPG
2079 * Fixes for arrays of structures (Michael)
2080 _________________________________________________________________
2084 * Python fix fetchone() (Gerhard Haring)
2085 * Use UTF, Unicode in Tcl where appropriate (Vsevolod Lobko,
2087 * Add Tcl COPY TO/FROM (ljb)
2088 * Prevent output of default index op class in pg_dump (Tom)
2089 * Fix libpgeasy memory leak (Bruce)
2090 _________________________________________________________________
2094 * Configure, dynamic loader, and shared library fixes (Peter E)
2095 * Fixes in QNX 4 port (Bernd Tegge)
2096 * Fixes in Cygwin and Windows ports (Jason Tishler, Gerhard Haring,
2097 Dmitry Yurtaev, Darko Prenosil, Mikhail Terekhov)
2098 * Fix for Windows socket communication failures (Magnus, Mikhail
2100 * Hurd compile fix (Oliver Elphick)
2101 * BeOS fixes (Cyril Velter)
2102 * Remove configure --enable-unicode-conversion, now enabled by
2104 * AIX fixes (Tatsuo, Andreas)
2105 * Fix parallel make (Peter E)
2106 * Install SQL language manual pages into OS-specific directories
2108 * Rename config.h to pg_config.h (Peter E)
2109 * Reorganize installation layout of header files (Peter E)
2110 _________________________________________________________________
2114 * Remove SEP_CHAR (Bruce)
2115 * New GUC hooks (Tom)
2116 * Merge GUC and command line handling (Marko Kreen)
2117 * Remove EXTEND INDEX (Martijn van Oosterhout, Tom)
2118 * New pgjindent utility to indent java code (Bruce)
2119 * Remove define of true/false when compiling under C++ (Leandro
2121 * pgindent fixes (Bruce, Tom)
2122 * Replace strcasecmp() with strcmp() where appropriate (Peter E)
2123 * Dynahash portability improvements (Tom)
2124 * Add 'volatile' usage in spinlock structures
2125 * Improve signal handling logic (Tom)
2126 _________________________________________________________________
2130 * New contrib/rtree_gist (Oleg Bartunov, Teodor Sigaev)
2131 * New contrib/tsearch full-text indexing (Oleg, Teodor Sigaev)
2132 * Add contrib/dblink for remote database access (Joe Conway)
2133 * contrib/ora2pg Oracle conversion utility (Gilles Darold)
2134 * contrib/xml XML conversion utility (John Gray)
2135 * contrib/fulltextindex fixes (Christopher Kings-Lynne)
2136 * New contrib/fuzzystrmatch with levenshtein and metaphone, soundex
2138 * Add contrib/intarray boolean queries, binary search, fixes (Oleg
2140 * New pg_upgrade utility (Bruce)
2141 * Add new pg_resetxlog options (Bruce, Tom)
2142 _________________________________________________________________
2146 Release date: 2001-08-15
2147 _________________________________________________________________
2149 Migration to version 7.1.3
2151 A dump/restore is *not* required for those running 7.1.X.
2152 _________________________________________________________________
2156 Remove unused WAL segements of large transactions (Tom)
2157 Multiaction rule fix (Tom)
2158 PL/pgSQL memory allocation fix (Jan)
2159 VACUUM buffer fix (Tom)
2160 Regression test fixes (Tom)
2161 pg_dump fixes for GRANT/REVOKE/comments on views, user-defined types (Tom)
2162 Fix subselects with DISTINCT ON or LIMIT (Tom)
2164 Disable COPY TO/FROM a view (Tom)
2165 Cygwin build (Jason Tishler)
2167 _________________________________________________________________
2171 Release date: 2001-05-11
2173 This has one fix from 7.1.1.
2174 _________________________________________________________________
2176 Migration to version 7.1.2
2178 A dump/restore is *not* required for those running 7.1.X.
2179 _________________________________________________________________
2183 Fix PL/pgSQL SELECTs when returning no rows
2184 Fix for psql backslash core dump
2185 Referential integrity privilege fix
2189 _________________________________________________________________
2193 Release date: 2001-05-05
2195 This has a variety of fixes from 7.1.
2196 _________________________________________________________________
2198 Migration to version 7.1.1
2200 A dump/restore is *not* required for those running 7.1.
2201 _________________________________________________________________
2205 Fix for numeric MODULO operator (Tom)
2206 pg_dump fixes (Philip)
2207 pg_dump can dump 7.0 databases (Philip)
2208 readline 4.2 fixes (Peter E)
2210 AIX, MSWIN, VAX, N32K fixes (Tom)
2211 Multibytes fixes (Tom)
2212 Unicode fixes (Tatsuo)
2213 Optimizer improvements (Tom)
2214 Fix for whole rows in functions (Tom)
2215 Fix for pg_ctl and option strings with spaces (Peter E)
2216 ODBC fixes (Hiroshi)
2217 EXTRACT can now take string argument (Thomas)
2218 Python fixes (Darcy)
2220 _________________________________________________________________
2224 Release date: 2001-04-13
2226 This release focuses on removing limitations that have existed in the
2227 PostgreSQL code for many years.
2229 Major changes in this release:
2231 Write-ahead Log (WAL)
2232 To maintain database consistency in case of an operating system
2233 crash, previous releases of PostgreSQL have forced all data
2234 modifications to disk before each transaction commit. With WAL,
2235 only one log file must be flushed to disk, greatly improving
2236 performance. If you have been using -F in previous releases to
2237 disable disk flushes, you may want to consider discontinuing
2241 TOAST - Previous releases had a compiled-in row length limit,
2242 typically 8k - 32k. This limit made storage of long text fields
2243 difficult. With TOAST, long rows of any length can be stored
2244 with good performance.
2247 We now support outer joins. The UNION/NOT IN workaround for
2248 outer joins is no longer required. We use the SQL92 outer join
2252 The previous C function manager did not handle null values
2253 properly, nor did it support 64-bit CPU's (Alpha). The new
2254 function manager does. You can continue using your old custom
2255 functions, but you may want to rewrite them in the future to
2256 use the new function manager call interface.
2259 A large number of complex queries that were unsupported in
2260 previous releases now work. Many combinations of views,
2261 aggregates, UNION, LIMIT, cursors, subqueries, and inherited
2262 tables now work properly. Inherited tables are now accessed by
2263 default. Subqueries in FROM are now supported.
2264 _________________________________________________________________
2266 Migration to version 7.1
2268 A dump/restore using pg_dump is required for those wishing to migrate
2269 data from any previous release.
2270 _________________________________________________________________
2276 Many multibyte/Unicode/locale fixes (Tatsuo and others)
2277 More reliable ALTER TABLE RENAME (Tom)
2278 Kerberos V fixes (David Wragg)
2279 Fix for INSERT INTO...SELECT where targetlist has subqueries (Tom)
2280 Prompt username/password on standard error (Bruce)
2281 Large objects inv_read/inv_write fixes (Tom)
2282 Fixes for to_char(), to_date(), to_ascii(), and to_timestamp() (Karel,
2284 Prevent query expressions from leaking memory (Tom)
2285 Allow UPDATE of arrays elements (Tom)
2286 Wake up lock waiters during cancel (Hiroshi)
2287 Fix rare cursor crash when using hash join (Tom)
2288 Fix for DROP TABLE/INDEX in rolled-back transaction (Hiroshi)
2289 Fix psql crash from \l+ if MULTIBYTE enabled (Peter E)
2290 Fix truncation of rule names during CREATE VIEW (Ross Reedstrom)
2291 Fix PL/perl (Alex Kapranoff)
2292 Disallow LOCK on views (Mark Hollomon)
2293 Disallow INSERT/UPDATE/DELETE on views (Mark Hollomon)
2294 Disallow DROP RULE, CREATE INDEX, TRUNCATE on views (Mark Hollomon)
2295 Allow PL/pgSQL accept non-ASCII identifiers (Tatsuo)
2296 Allow views to proper handle GROUP BY, aggregates, DISTINCT (Tom)
2297 Fix rare failure with TRUNCATE command (Tom)
2298 Allow UNION/INTERSECT/EXCEPT to be used with ALL, subqueries, views,
2299 DISTINCT, ORDER BY, SELECT...INTO (Tom)
2300 Fix parser failures during aborted transactions (Tom)
2301 Allow temporary relations to properly clean up indexes (Bruce)
2302 Fix VACUUM problem with moving rows in same page (Tom)
2303 Modify pg_dump to better handle user-defined items in template1 (Philip)
2304 Allow LIMIT in VIEW (Tom)
2305 Require cursor FETCH to honor LIMIT (Tom)
2306 Allow PRIMARY/FOREIGN Key definitions on inherited columns (Stephan)
2307 Allow ORDER BY, LIMIT in subqueries (Tom)
2308 Allow UNION in CREATE RULE (Tom)
2309 Make ALTER/DROP TABLE rollback-able (Vadim, Tom)
2310 Store initdb collation in pg_control so collation cannot be changed (Tom)
2311 Fix INSERT...SELECT with rules (Tom)
2312 Fix FOR UPDATE inside views and subselects (Tom)
2313 Fix OVERLAPS operators conform to SQL92 spec regarding NULLs (Tom)
2314 Fix lpad() and rpad() to handle length less than input string (Tom)
2315 Fix use of NOTIFY in some rules (Tom)
2316 Overhaul btree code (Tom)
2317 Fix NOT NULL use in Pl/pgSQL variables (Tom)
2318 Overhaul GIST code (Oleg)
2319 Fix CLUSTER to preserve constraints and column default (Tom)
2320 Improved deadlock detection handling (Tom)
2321 Allow multiple SERIAL columns in a table (Tom)
2322 Prevent occasional index corruption (Vadim)
2326 Add OUTER JOINs (Tom)
2327 Function manager overhaul (Tom)
2328 Allow ALTER TABLE RENAME on indexes (Tom)
2329 Improve CLUSTER (Tom)
2330 Improve ps status display for more platforms (Peter E, Marc)
2331 Improve CREATE FUNCTION failure message (Ross)
2332 JDBC improvements (Peter, Travis Bauer, Christopher Cain, William Webber,
2334 Grand Unified Configuration scheme/GUC. Many options can now be set in
2335 data/postgresql.conf, postmaster/postgres flags, or SET commands (Peter E)
2336 Improved handling of file descriptor cache (Tom)
2337 New warning code about auto-created table alias entries (Bruce)
2338 Overhaul initdb process (Tom, Peter E)
2339 Overhaul of inherited tables; inherited tables now accessed by default;
2340 new ONLY key word prevents it (Chris Bitmead, Tom)
2341 ODBC cleanups/improvements (Nick Gorham, Stephan Szabo, Zoltan Kovacs,
2343 Allow renaming of temp tables (Tom)
2344 Overhaul memory manager contexts (Tom)
2345 pg_dumpall uses CREATE USER or CREATE GROUP rather using COPY (Peter E)
2346 Overhaul pg_dump (Philip Warner)
2347 Allow pg_hba.conf secondary password file to specify only username (Peter E)
2348 Allow TEMPORARY or TEMP key word when creating temporary tables (Bruce)
2349 New memory leak checker (Karel)
2350 New SET SESSION CHARACTERISTICS (Thomas)
2351 Allow nested block comments (Thomas)
2352 Add WITHOUT TIME ZONE type qualifier (Thomas)
2353 New ALTER TABLE ADD CONSTRAINT (Stephan)
2354 Use NUMERIC accumulators for INTEGER aggregates (Tom)
2355 Overhaul aggregate code (Tom)
2356 New VARIANCE and STDDEV() aggregates
2357 Improve dependency ordering of pg_dump (Philip)
2358 New pg_restore command (Philip)
2359 New pg_dump tar output option (Philip)
2360 New pg_dump of large objects (Philip)
2361 New ESCAPE option to LIKE (Thomas)
2362 New case-insensitive LIKE - ILIKE (Thomas)
2363 Allow functional indexes to use binary-compatible type (Tom)
2364 Allow SQL functions to be used in more contexts (Tom)
2365 New pg_config utility (Peter E)
2366 New PL/pgSQL EXECUTE command which allows dynamic SQL and utility statements
2368 New PL/pgSQL GET DIAGNOSTICS statement for SPI value access (Jan)
2369 New quote_identifiers() and quote_literal() functions (Jan)
2370 New ALTER TABLE table OWNER TO user command (Mark Hollomon)
2371 Allow subselects in FROM, i.e. FROM (SELECT ...) [AS] alias (Tom)
2372 Update PyGreSQL to version 3.1 (D'Arcy)
2373 Store tables as files named by OID (Vadim)
2374 New SQL function setval(seq,val,bool) for use in pg_dump (Philip)
2375 Require DROP VIEW to remove views, no DROP TABLE (Mark)
2376 Allow DROP VIEW view1, view2 (Mark)
2377 Allow multiple objects in DROP INDEX, DROP RULE, and DROP TYPE (Tom)
2378 Allow automatic conversion to/from Unicode (Tatsuo, Eiji)
2379 New /contrib/pgcrypto hashing functions (Marko Kreen)
2380 New pg_dumpall --globals-only option (Peter E)
2381 New CHECKPOINT command for WAL which creates new WAL log file (Vadim)
2382 New AT TIME ZONE syntax (Thomas)
2383 Allow location of Unix domain socket to be configurable (David J. MacKenzie)
2384 Allow postmaster to listen on a specific IP address (David J. MacKenzie)
2385 Allow socket path name to be specified in hostname by using leading slash
2386 (David J. MacKenzie)
2387 Allow CREATE DATABASE to specify template database (Tom)
2388 New utility to convert MySQL schema dumps to SQL92 and PostgreSQL (Thomas)
2389 New /contrib/rserv replication toolkit (Vadim)
2390 New file format for COPY BINARY (Tom)
2391 New /contrib/oid2name to map numeric files to table names (B Palmer)
2392 New "idle in transaction" ps status message (Marc)
2393 Update to pgaccess 0.98.7 (Constantin Teodorescu)
2394 pg_ctl now defaults to -w (wait) on shutdown, new -l (log) option
2395 Add rudimentary dependency checking to pg_dump (Philip)
2399 Fix INET/CIDR type ordering and add new functions (Tom)
2400 Make OID behave as an unsigned type (Tom)
2401 Allow BIGINT as synonym for INT8 (Peter E)
2402 New int2 and int8 comparison operators (Tom)
2403 New BIT and BIT VARYING types (Adriaan Joubert, Tom, Peter E)
2404 CHAR() no longer faster than VARCHAR() because of TOAST (Tom)
2405 New GIST seg/cube examples (Gene Selkov)
2406 Improved round(numeric) handling (Tom)
2407 Fix CIDR output formatting (Tom)
2408 New CIDR abbrev() function (Tom)
2412 Write-Ahead Log (WAL) to provide crash recovery with less performance
2414 ANALYZE stage of VACUUM no longer exclusively locks table (Bruce)
2415 Reduced file seeks (Denis Perchine)
2416 Improve BTREE code for duplicate keys (Tom)
2417 Store all large objects in a single table (Denis Perchine, Tom)
2418 Improve memory allocation performance (Karel, Tom)
2422 New function manager call conventions (Tom)
2423 SGI portability fixes (David Kaelbling)
2424 New configure --enable-syslog option (Peter E)
2425 New BSDI README (Bruce)
2426 configure script moved to top level, not /src (Peter E)
2427 Makefile/configuration/compilation overhaul (Peter E)
2428 New configure --with-python option (Peter E)
2429 Solaris cleanups (Peter E)
2430 Overhaul /contrib Makefiles (Karel)
2431 New OpenSSL configuration option (Magnus, Peter E)
2433 QNX fixes (Maurizio)
2434 New heap_open(), heap_openr() API (Tom)
2435 Remove colon and semi-colon operators (Thomas)
2436 New pg_class.relkind value for views (Mark Hollomon)
2437 Rename ichar() to chr() (Karel)
2438 New documentation for btrim(), ascii(), chr(), repeat() (Karel)
2439 Fixes for NT/Cygwin (Pete Forman)
2440 AIX port fixes (Andreas)
2441 New BeOS port (David Reid, Cyril Velter)
2442 Add proofreader's changes to docs (Addison-Wesley, Bruce)
2443 New Alpha spinlock code (Adriaan Joubert, Compaq)
2444 UnixWare port overhaul (Peter E)
2445 New Darwin/MacOS X port (Peter Bierman, Bruce Hartzler)
2446 New FreeBSD Alpha port (Alfred)
2447 Overhaul shared memory segments (Tom)
2448 Add IBM S/390 support (Neale Ferguson)
2449 Moved macmanuf to /contrib (Larry Rosenman)
2450 Syslog improvements (Larry Rosenman)
2451 New template0 database that contains no user additions (Tom)
2452 New /contrib/cube and /contrib/seg GIST sample code (Gene Selkov)
2453 Allow NetBSD's libedit instead of readline (Peter)
2454 Improved assembly language source code format (Bruce)
2455 New contrib/pg_logger
2456 New --template option to createdb
2457 New contrib/pg_control utility (Oliver)
2458 New FreeBSD tools ipc_check, start-scripts/freebsd
2460 _________________________________________________________________
2464 Release date: 2000-11-11
2466 This has a variety of fixes from 7.0.2.
2467 _________________________________________________________________
2469 Migration to version 7.0.3
2471 A dump/restore is *not* required for those running 7.0.*.
2472 _________________________________________________________________
2477 Large object fix (Tom)
2478 Fix lean in COPY WITH OIDS leak (Tom)
2479 Fix backwards-index-scan (Tom)
2480 Fix SELECT ... FOR UPDATE so it checks for duplicate keys (Hiroshi)
2481 Add --enable-syslog to configure (Marc)
2482 Fix abort transaction at backend exit in rare cases (Tom)
2483 Fix for psql \l+ when multibyte enabled (Tatsuo)
2484 Allow PL/pgSQL to accept non ascii identifiers (Tatsuo)
2485 Make vacuum always flush buffers (Tom)
2486 Fix to allow cancel while waiting for a lock (Hiroshi)
2487 Fix for memory aloocation problem in user authentication code (Tom)
2488 Remove bogus use of int4out() (Tom)
2489 Fixes for multiple subqueries in COALESCE or BETWEEN (Tom)
2490 Fix for failure of triggers on heap open in certain cases (Jeroen van
2492 Fix for erroneous selectivity of not-equals (Tom)
2493 Fix for erroneous use of strcmp() (Tom)
2494 Fix for bug where storage manager accesses items beyond end of file
2496 Fix to include kernel errno message in all smgr elog messages (Tom)
2497 Fix for '.' not in PATH at build time (SL Baur)
2498 Fix for out-of-file-descriptors error (Tom)
2499 Fix to make pg_dump dump 'iscachable' flag for functions (Tom)
2500 Fix for subselect in targetlist of Append node (Tom)
2501 Fix for mergejoin plans (Tom)
2502 Fix TRUNCATE failure on relations with indexes (Tom)
2503 Avoid database-wide restart on write error (Hiroshi)
2504 Fix nodeMaterial to honor chgParam by recomputing its output (Tom)
2505 Fix VACUUM problem with moving chain of update row versions when source
2506 and destination of a row version lie on the same page (Tom)
2507 Fix user.c CommandCounterIncrement (Tom)
2508 Fix for AM/PM boundary problem in to_char() (Karel Zak)
2509 Fix TIME aggregate handling (Tom)
2510 Fix to_char() to avoid coredump on NULL input (Tom)
2512 Fix for inserting/copying longer multibyte strings into char() data
2514 Fix for crash of backend, on abort (Tom)
2516 _________________________________________________________________
2520 Release date: 2000-06-05
2522 This is a repackaging of 7.0.1 with added documentation.
2523 _________________________________________________________________
2525 Migration to version 7.0.2
2527 A dump/restore is *not* required for those running 7.*.
2528 _________________________________________________________________
2532 Added documentation to tarball.
2534 _________________________________________________________________
2538 Release date: 2000-06-01
2540 This is a cleanup release for 7.0.
2541 _________________________________________________________________
2543 Migration to version 7.0.1
2545 A dump/restore is *not* required for those running 7.0.
2546 _________________________________________________________________
2550 Fix many CLUSTER failures (Tom)
2551 Allow ALTER TABLE RENAME works on indexes (Tom)
2552 Fix plpgsql to handle datetime->timestamp and timespan->interval (Bruce)
2553 New configure --with-setproctitle switch to use setproctitle() (Marc, Bruce)
2554 Fix the off by one errors in ResultSet from 6.5.3, and more.
2555 jdbc ResultSet fixes (Joseph Shraibman)
2556 optimizer tunings (Tom)
2557 Fix create user for pgaccess
2558 Fix for UNLISTEN failure
2559 IRIX fixes (David Kaelbling)
2560 QNX fixes (Andreas Kardos)
2561 Reduce COPY IN lock level (Tom)
2562 Change libpqeasy to use PQconnectdb() style parameters (Bruce)
2563 Fix pg_dump to handle OID indexes (Tom)
2564 Fix small memory leak (Tom)
2565 Solaris fix for createdb/dropdb (Tatsuo)
2566 Fix for non-blocking connections (Alfred Perlstein)
2567 Fix improper recovery after RENAME TABLE failures (Tom)
2568 Copy pg_ident.conf.sample into /lib directory in install (Bruce)
2569 Add SJIS UDC (NEC selection IBM kanji) support (Eiji Tokuya)
2570 Fix too long syslog message (Tatsuo)
2571 Fix problem with quoted indexes that are too long (Tom)
2572 JDBC ResultSet.getTimestamp() fix (Gregory Krasnow & Floyd Marinescu)
2573 ecpg changes (Michael)
2575 _________________________________________________________________
2579 Release date: 2000-05-08
2581 This release contains improvements in many areas, demonstrating the
2582 continued growth of PostgreSQL. There are more improvements and fixes
2583 in 7.0 than in any previous release. The developers have confidence
2584 that this is the best release yet; we do our best to put out only
2585 solid releases, and this one is no exception.
2587 Major changes in this release:
2590 Foreign keys are now implemented, with the exception of PARTIAL
2591 MATCH foreign keys. Many users have been asking for this
2592 feature, and we are pleased to offer it.
2595 Continuing on work started a year ago, the optimizer has been
2596 improved, allowing better query plan selection and faster
2597 performance with less memory usage.
2600 psql, our interactive terminal monitor, has been updated with a
2601 variety of new features. See the psql manual page for details.
2604 SQL92 join syntax is now supported, though only as INNER JOIN
2605 for this release. JOIN, NATURAL JOIN, JOIN/USING, and JOIN/ON
2606 are available, as are column correlation names.
2607 _________________________________________________________________
2609 Migration to version 7.0
2611 A dump/restore using pg_dump is required for those wishing to migrate
2612 data from any previous release of PostgreSQL. For those upgrading from
2613 6.5.*, you may instead use pg_upgrade to upgrade to this release;
2614 however, a full dump/reload installation is always the most robust
2615 method for upgrades.
2617 Interface and compatibility issues to consider for the new release
2620 * The date/time types datetime and timespan have been superseded by
2621 the SQL92-defined types timestamp and interval. Although there has
2622 been some effort to ease the transition by allowing PostgreSQL to
2623 recognize the deprecated type names and translate them to the new
2624 type names, this mechanism may not be completely transparent to
2625 your existing application.
2626 * The optimizer has been substantially improved in the area of query
2627 cost estimation. In some cases, this will result in decreased
2628 query times as the optimizer makes a better choice for the
2629 preferred plan. However, in a small number of cases, usually
2630 involving pathological distributions of data, your query times may
2631 go up. If you are dealing with large amounts of data, you may want
2632 to check your queries to verify performance.
2633 * The JDBC and ODBC interfaces have been upgraded and extended.
2634 * The string function CHAR_LENGTH is now a native function. Previous
2635 versions translated this into a call to LENGTH, which could result
2636 in ambiguity with other types implementing LENGTH such as the
2638 _________________________________________________________________
2644 Prevent function calls exceeding maximum number of arguments (Tom)
2645 Improve CASE construct (Tom)
2646 Fix SELECT coalesce(f1,0) FROM int4_tbl GROUP BY f1 (Tom)
2647 Fix SELECT sentence.words[0] FROM sentence GROUP BY sentence.words[0] (Tom)
2648 Fix GROUP BY scan bug (Tom)
2649 Improvements in SQL grammar processing (Tom)
2650 Fix for views involved in INSERT ... SELECT ... (Tom)
2651 Fix for SELECT a/2, a/2 FROM test_missing_target GROUP BY a/2 (Tom)
2652 Fix for subselects in INSERT ... SELECT (Tom)
2653 Prevent INSERT ... SELECT ... ORDER BY (Tom)
2654 Fixes for relations greater than 2GB, including vacuum
2655 Improve propagating system table changes to other backends (Tom)
2656 Improve propagating user table changes to other backends (Tom)
2657 Fix handling of temp tables in complex situations (Bruce, Tom)
2658 Allow table locking at table open, improving concurrent reliability (Tom)
2659 Properly quote sequence names in pg_dump (Ross J. Reedstrom)
2660 Prevent DROP DATABASE while others accessing
2661 Prevent any rows from being returned by GROUP BY if no rows processed (Tom)
2662 Fix SELECT COUNT(1) FROM table WHERE ...' if no rows matching WHERE (Tom)
2663 Fix pg_upgrade so it works for MVCC (Tom)
2664 Fix for SELECT ... WHERE x IN (SELECT ... HAVING SUM(x) > 1) (Tom)
2665 Fix for "f1 datetime DEFAULT 'now'" (Tom)
2666 Fix problems with CURRENT_DATE used in DEFAULT (Tom)
2667 Allow comment-only lines, and ;;; lines too. (Tom)
2668 Improve recovery after failed disk writes, disk full (Hiroshi)
2669 Fix cases where table is mentioned in FROM but not joined (Tom)
2670 Allow HAVING clause without aggregate functions (Tom)
2671 Fix for "--" comment and no trailing newline, as seen in perl interface
2672 Improve pg_dump failure error reports (Bruce)
2673 Allow sorts and hashes to exceed 2GB file sizes (Tom)
2674 Fix for pg_dump dumping of inherited rules (Tom)
2675 Fix for NULL handling comparisons (Tom)
2676 Fix inconsistent state caused by failed CREATE/DROP commands (Hiroshi)
2677 Fix for dbname with dash
2678 Prevent DROP INDEX from interfering with other backends (Tom)
2679 Fix file descriptor leak in verify_password()
2680 Fix for "Unable to identify an operator =$" problem
2681 Fix ODBC so no segfault if CommLog and Debug enabled (Dirk Niggemann)
2682 Fix for recursive exit call (Massimo)
2683 Fix for extra-long timezones (Jeroen van Vianen)
2684 Make pg_dump preserve primary key information (Peter E)
2685 Prevent databases with single quotes (Peter E)
2686 Prevent DROP DATABASE inside transaction (Peter E)
2687 ecpg memory leak fixes (Stephen Birch)
2688 Fix for SELECT null::text, SELECT int4fac(null) and SELECT 2 + (null) (Tom)
2689 Y2K timestamp fix (Massimo)
2690 Fix for VACUUM 'HEAP_MOVED_IN was not expected' errors (Tom)
2691 Fix for views with tables/columns containing spaces (Tom)
2692 Prevent privileges on indexes (Peter E)
2693 Fix for spinlock stuck problem when error is generated (Hiroshi)
2694 Fix ipcclean on Linux
2695 Fix handling of NULL constraint conditions (Tom)
2696 Fix memory leak in odbc driver (Nick Gorham)
2697 Fix for privilege check on UNION tables (Tom)
2698 Fix to allow SELECT 'a' LIKE 'a' (Tom)
2699 Fix for SELECT 1 + NULL (Tom)
2701 Fix log() on numeric type (Tom)
2702 Deprecate ':' and ';' operators
2703 Allow vacuum of temporary tables
2704 Disallow inherited columns with the same name as new columns
2705 Recover or force failure when disk space is exhausted (Hiroshi)
2706 Fix INSERT INTO ... SELECT with AS columns matching result columns
2707 Fix INSERT ... SELECT ... GROUP BY groups by target columns not source columns
2709 Fix CREATE TABLE test (a char(5) DEFAULT text '', b int4) with INSERT (Tom)
2710 Fix UNION with LIMIT
2711 Fix CREATE TABLE x AS SELECT 1 UNION SELECT 2
2712 Fix CREATE TABLE test(col char(2) DEFAULT user)
2713 Fix mismatched types in CREATE TABLE ... DEFAULT
2714 Fix SELECT * FROM pg_class where oid in (0,-1)
2715 Fix SELECT COUNT('asdf') FROM pg_class WHERE oid=12
2716 Prevent user who can create databases can modifying pg_database table (Peter E)
2717 Fix btree to give a useful elog when key > 1/2 (page - overhead) (Tom)
2718 Fix INSERT of 0.0 into DECIMAL(4,4) field (Tom)
2722 New CLI interface include file sqlcli.h, based on SQL3/SQL98
2723 Remove all limits on query length, row length limit still exists (Tom)
2724 Update jdbc protocol to 2.0 (Jens Glaser <jens@jens.de>)
2725 Add TRUNCATE command to quickly truncate relation (Mike Mascari)
2726 Fix to give super user and createdb user proper update catalog rights (Peter E)
2727 Allow ecpg bool variables to have NULL values (Christof)
2728 Issue ecpg error if NULL value for variable with no NULL indicator (Christof)
2729 Allow ^C to cancel COPY command (Massimo)
2730 Add SET FSYNC and SHOW PG_OPTIONS commands(Massimo)
2731 Function name overloading for dynamically-loaded C functions (Frankpitt)
2732 Add CmdTuples() to libpq++(Vince)
2733 New CREATE CONSTRAINT TRIGGER and SET CONSTRAINTS commands(Jan)
2734 Allow CREATE FUNCTION/WITH clause to be used for all language types
2735 configure --enable-debug adds -g (Peter E)
2736 configure --disable-debug removes -g (Peter E)
2737 Allow more complex default expressions (Tom)
2738 First real FOREIGN KEY constraint trigger functionality (Jan)
2739 Add FOREIGN KEY ... MATCH FULL ... ON DELETE CASCADE (Jan)
2740 Add FOREIGN KEY ... MATCH <unspecified> referential actions (Don Baccus)
2741 Allow WHERE restriction on ctid (physical heap location) (Hiroshi)
2742 Move pginterface from contrib to interface directory, rename to pgeasy (Bruce)
2743 Change pgeasy connectdb() parameter ordering (Bruce)
2744 Require SELECT DISTINCT target list to have all ORDER BY columns (Tom)
2745 Add Oracle's COMMENT ON command (Mike Mascari <mascarim@yahoo.com>)
2746 libpq's PQsetNoticeProcessor function now returns previous hook(Peter E)
2747 Prevent PQsetNoticeProcessor from being set to NULL (Peter E)
2748 Make USING in COPY optional (Bruce)
2749 Allow subselects in the target list (Tom)
2750 Allow subselects on the left side of comparison operators (Tom)
2751 New parallel regression test (Jan)
2752 Change backend-side COPY to write files with permissions 644 not 666 (Tom)
2753 Force permissions on PGDATA directory to be secure, even if it exists (Tom)
2754 Added psql LASTOID variable to return last inserted oid (Peter E)
2755 Allow concurrent vacuum and remove pg_vlock vacuum lock file (Tom)
2756 Add privilege check for vacuum (Peter E)
2757 New libpq functions to allow asynchronous connections: PQconnectStart(),
2758 PQconnectPoll(), PQresetStart(), PQresetPoll(), PQsetenvStart(),
2759 PQsetenvPoll(), PQsetenvAbort (Ewan Mellor)
2760 New libpq PQsetenv() function (Ewan Mellor)
2761 create/alter user extension (Peter E)
2762 New postmaster.pid and postmaster.opts under $PGDATA (Tatsuo)
2763 New scripts for create/drop user/db (Peter E)
2764 Major psql overhaul (Peter E)
2765 Add const to libpq interface (Peter E)
2766 New libpq function PQoidValue (Peter E)
2767 Show specific non-aggregate causing problem with GROUP BY (Tom)
2768 Make changes to pg_shadow recreate pg_pwd file (Peter E)
2769 Add aggregate(DISTINCT ...) (Tom)
2770 Allow flag to control COPY input/output of NULLs (Peter E)
2771 Make postgres user have a password by default (Peter E)
2772 Add CREATE/ALTER/DROP GROUP (Peter E)
2773 All administration scripts now support --long options (Peter E, Karel)
2774 Vacuumdb script now supports --all option (Peter E)
2775 ecpg new portable FETCH syntax
2776 Add ecpg EXEC SQL IFDEF, EXEC SQL IFNDEF, EXEC SQL ELSE, EXEC SQL ELIF
2777 and EXEC SQL ENDIF directives
2778 Add pg_ctl script to control backend start-up (Tatsuo)
2779 Add postmaster.opts.default file to store start-up flags (Tatsuo)
2780 Allow --with-mb=SQL_ASCII
2781 Increase maximum number of index keys to 16 (Bruce)
2782 Increase maximum number of function arguments to 16 (Bruce)
2783 Allow configuration of maximum number of index keys and arguments (Bruce)
2784 Allow unprivileged users to change their passwords (Peter E)
2785 Password authentication enabled; required for new users (Peter E)
2786 Disallow dropping a user who owns a database (Peter E)
2787 Change initdb option --with-mb to --enable-multibyte
2788 Add option for initdb to prompts for superuser password (Peter E)
2789 Allow complex type casts like col::numeric(9,2) and col::int2::float8 (Tom)
2790 Updated user interfaces on initdb, initlocation, pg_dump, ipcclean (Peter E)
2791 New pg_char_to_encoding() and pg_encoding_to_char() functions (Tatsuo)
2792 libpq non-blocking mode (Alfred Perlstein)
2793 Improve conversion of types in casts that don't specify a length
2794 New plperl internal programming language (Mark Hollomon)
2795 Allow COPY IN to read file that do not end with a newline (Tom)
2796 Indicate when long identifiers are truncated (Tom)
2797 Allow aggregates to use type equivalency (Peter E)
2798 Add Oracle's to_char(), to_date(), to_datetime(), to_timestamp(), to_number()
2799 conversion functions (Karel Zak <zakkr@zf.jcu.cz>)
2800 Add SELECT DISTINCT ON (expr [, expr ...]) targetlist ... (Tom)
2801 Check to be sure ORDER BY is compatible with the DISTINCT operation (Tom)
2802 Add NUMERIC and int8 types to ODBC
2803 Improve EXPLAIN results for Append, Group, Agg, Unique (Tom)
2804 Add ALTER TABLE ... ADD FOREIGN KEY (Stephan Szabo)
2805 Allow SELECT .. FOR UPDATE in PL/pgSQL (Hiroshi)
2806 Enable backward sequential scan even after reaching EOF (Hiroshi)
2807 Add btree indexing of boolean values, >= and <= (Don Baccus)
2808 Print current line number when COPY FROM fails (Massimo)
2809 Recognize POSIX time zone e.g. "PST+8" and "GMT-8" (Thomas)
2810 Add DEC as synonym for DECIMAL (Thomas)
2811 Add SESSION_USER as SQL92 key word, same as CURRENT_USER (Thomas)
2812 Implement SQL92 column aliases (aka correlation names) (Thomas)
2813 Implement SQL92 join syntax (Thomas)
2814 Make INTERVAL reserved word allowed as a column identifier (Thomas)
2815 Implement REINDEX command (Hiroshi)
2816 Accept ALL in aggregate function SUM(ALL col) (Tom)
2817 Prevent GROUP BY from using column aliases (Tom)
2818 New psql \encoding option (Tatsuo)
2819 Allow PQrequestCancel() to terminate when in waiting-for-lock state (Hiroshi)
2820 Allow negation of a negative number in all cases
2821 Add ecpg descriptors (Christof, Michael)
2822 Allow CREATE VIEW v AS SELECT f1::char(8) FROM tbl
2823 Allow casts with length, like foo::char(8)
2824 New libpq functions PQsetClientEncoding(), PQclientEncoding() (Tatsuo)
2825 Add support for SJIS user defined characters (Tatsuo)
2826 Larger views/rules supported
2827 Make libpq's PQconndefaults() thread-safe (Tom)
2828 Disable // as comment to be ANSI conforming, should use -- (Tom)
2829 Allow column aliases on views CREATE VIEW name (collist)
2830 Fixes for views with subqueries (Tom)
2831 Allow UPDATE table SET fld = (SELECT ...) (Tom)
2832 SET command options no longer require quotes
2833 Update pgaccess to 0.98.6
2834 New SET SEED command
2835 New pg_options.sample file
2836 New SET FSYNC command (Massimo)
2837 Allow pg_descriptions when creating tables
2838 Allow pg_descriptions when creating types, columns, and functions
2839 Allow psql \copy to allow delimiters (Peter E)
2840 Allow psql to print nulls as distinct from "" [null] (Peter E)
2844 Many array fixes (Tom)
2845 Allow bare column names to be subscripted as arrays (Tom)
2846 Improve type casting of int and float constants (Tom)
2847 Cleanups for int8 inputs, range checking, and type conversion (Tom)
2848 Fix for SELECT timespan('21:11:26'::time) (Tom)
2849 netmask('x.x.x.x/0') is 255.255.255.255 instead of 0.0.0.0 (Oleg Sharoiko)
2850 Add btree index on NUMERIC (Jan)
2851 Perl fix for large objects containing NUL characters (Douglas Thomson)
2852 ODBC fix for for large objects (free)
2853 Fix indexing of cidr data type
2854 Fix for Ethernet MAC addresses (macaddr type) comparisons
2855 Fix for date/time types when overflows happened in computations (Tom)
2856 Allow array on int8 (Peter E)
2857 Fix for rounding/overflow of NUMERIC type, like NUMERIC(4,4) (Tom)
2858 Allow NUMERIC arrays
2859 Fix bugs in NUMERIC ceil() and floor() functions (Tom)
2860 Make char_length()/octet_length including trailing blanks (Tom)
2861 Made abstime/reltime use int4 instead of time_t (Peter E)
2862 New lztext data type for compressed text fields
2863 Revise code to handle coercion of int and float constants (Tom)
2864 Start at new code to implement a BIT and BIT VARYING type (Adriaan Joubert)
2865 NUMERIC now accepts scientific notation (Tom)
2866 NUMERIC to int4 rounds (Tom)
2867 Convert float4/8 to NUMERIC properly (Tom)
2868 Allow type conversion with NUMERIC (Thomas)
2869 Make ISO date style (2000-02-16 09:33) the default (Thomas)
2870 Add NATIONAL CHAR [ VARYING ] (Thomas)
2871 Allow NUMERIC round and trunc to accept negative scales (Tom)
2872 New TIME WITH TIME ZONE type (Thomas)
2873 Add MAX()/MIN() on time type (Thomas)
2874 Add abs(), mod(), fac() for int8 (Thomas)
2875 Rename functions to round(), sqrt(), cbrt(), pow() for float8 (Thomas)
2876 Add transcendental math functions (e.g. sin(), acos()) for float8 (Thomas)
2877 Add exp() and ln() for NUMERIC type
2878 Rename NUMERIC power() to pow() (Thomas)
2879 Improved TRANSLATE() function (Edwin Ramirez, Tom)
2880 Allow X=-Y operators (Tom)
2881 Allow SELECT float8(COUNT(*))/(SELECT COUNT(*) FROM t) FROM t GROUP BY f1; (Tom
2883 Allow LOCALE to use indexes in regular expression searches (Tom)
2884 Allow creation of functional indexes to use default types
2888 Prevent exponential space consumption with many AND's and OR's (Tom)
2889 Collect attribute selectivity values for system columns (Tom)
2890 Reduce memory usage of aggregates (Tom)
2891 Fix for LIKE optimization to use indexes with multibyte encodings (Tom)
2892 Fix r-tree index optimizer selectivity (Thomas)
2893 Improve optimizer selectivity computations and functions (Tom)
2894 Optimize btree searching for cases where many equal keys exist (Tom)
2895 Enable fast LIKE index processing only if index present (Tom)
2896 Re-use free space on index pages with duplicates (Tom)
2897 Improve hash join processing (Tom)
2898 Prevent descending sort if result is already sorted(Hiroshi)
2899 Allow commuting of index scan query qualifications (Tom)
2900 Prefer index scans in cases where ORDER BY/GROUP BY is required (Tom)
2901 Allocate large memory requests in fix-sized chunks for performance (Tom)
2902 Fix vacuum's performance by reducing memory allocation requests (Tom)
2903 Implement constant-expression simplification (Bernard Frankpitt, Tom)
2904 Use secondary columns to be used to determine start of index scan (Hiroshi)
2905 Prevent quadruple use of disk space when doing internal sorting (Tom)
2906 Faster sorting by calling fewer functions (Tom)
2907 Create system indexes to match all system caches (Bruce, Hiroshi)
2908 Make system caches use system indexes (Bruce)
2909 Make all system indexes unique (Bruce)
2910 Improve pg_statistics management for VACUUM speed improvement (Tom)
2911 Flush backend cache less frequently (Tom, Hiroshi)
2912 COPY now reuses previous memory allocation, improving performance (Tom)
2913 Improve optimization cost estimation (Tom)
2914 Improve optimizer estimate of range queries x > lowbound AND x < highbound (Tom
2916 Use DNF instead of CNF where appropriate (Tom, Taral)
2917 Further cleanup for OR-of-AND WHERE-clauses (Tom)
2918 Make use of index in OR clauses (x = 1 AND y = 2) OR (x = 2 AND y = 4) (Tom)
2919 Smarter optimizer computations for random index page access (Tom)
2920 New SET variable to control optimizer costs (Tom)
2921 Optimizer queries based on LIMIT, OFFSET, and EXISTS qualifications (Tom)
2922 Reduce optimizer internal housekeeping of join paths for speedup (Tom)
2923 Major subquery speedup (Tom)
2924 Fewer fsync writes when fsync is not disabled (Tom)
2925 Improved LIKE optimizer estimates (Tom)
2926 Prevent fsync in SELECT-only queries (Vadim)
2927 Make index creation use psort code, because it is now faster (Tom)
2928 Allow creation of sort temp tables > 1 Gig
2932 Fix for linux PPC compile
2933 New generic expression-tree-walker subroutine (Tom)
2934 Change form() to varargform() to prevent portability problems
2935 Improved range checking for large integers on Alphas
2936 Clean up #include in /include directory (Bruce)
2937 Add scripts for checking includes (Bruce)
2938 Remove un-needed #include's from *.c files (Bruce)
2939 Change #include's to use <> and "" as appropriate (Bruce)
2940 Enable Windows compilation of libpq
2941 Alpha spinlock fix from Uncle George <gatgul@voicenet.com>
2942 Overhaul of optimizer data structures (Tom)
2943 Fix to cygipc library (Yutaka Tanida)
2944 Allow pgsql to work on newer Cygwin snapshots (Dan)
2945 New catalog version number (Tom)
2947 Rename heap_replace to heap_update
2948 Update for QNX (Dr. Andreas Kardos)
2949 New platform-specific regression handling (Tom)
2950 Rename oid8 -> oidvector and int28 -> int2vector (Bruce)
2951 Included all yacc and lex files into the distribution (Peter E.)
2952 Remove lextest, no longer needed (Peter E)
2953 Fix for libpq and psql on Windows (Magnus)
2954 Internally change datetime and timespan into timestamp and interval (Thomas)
2955 Fix for plpgsql on BSD/OS
2956 Add SQL_ASCII test case to the regression test (Tatsuo)
2957 configure --with-mb now deprecated (Tatsuo)
2959 NetBSD fixes (Johnny C. Lam <lamj@stat.cmu.edu>)
2960 Fixes for Alpha compiles
2961 New multibyte encodings
2963 _________________________________________________________________
2967 Release date: 1999-10-13
2969 This is basically a cleanup release for 6.5.2. We have added a new
2970 PgAccess that was missing in 6.5.2, and installed an NT-specific fix.
2971 _________________________________________________________________
2973 Migration to version 6.5.3
2975 A dump/restore is *not* required for those running 6.5.*.
2976 _________________________________________________________________
2980 Updated version of pgaccess 0.98
2982 Fix dumping rules on inherited tables
2984 _________________________________________________________________
2988 Release date: 1999-09-15
2990 This is basically a cleanup release for 6.5.1. We have fixed a variety
2991 of problems reported by 6.5.1 users.
2992 _________________________________________________________________
2994 Migration to version 6.5.2
2996 A dump/restore is *not* required for those running 6.5.*.
2997 _________________________________________________________________
3001 subselect+CASE fixes(Tom)
3002 Add SHLIB_LINK setting for solaris_i386 and solaris_sparc ports(Daren Sefcik)
3003 Fixes for CASE in WHERE join clauses(Tom)
3004 Fix BTScan abort(Tom)
3005 Repair the check for redundant UNIQUE and PRIMARY KEY indexes(Thomas)
3006 Improve it so that it checks for multicolumn constraints(Thomas)
3007 Fix for Windows making problem with MB enabled(Hiroki Kataoka)
3008 Allow BSD yacc and bison to compile pl code(Bruce)
3009 Fix SET NAMES working
3011 Fix vacuum's memory consumption(Hiroshi,Tatsuo)
3012 Reduce the total memory consumption of vacuum(Tom)
3013 Fix for timestamp(datetime)
3014 Rule deparsing bugfixes(Tom)
3015 Fix quoting problems in mkMakefile.tcldefs.sh.in and mkMakefile.tkdefs.sh.in(To
3017 This is to re-use space on index pages freed by vacuum(Vadim)
3018 document -x for pg_dump(Bruce)
3019 Fix for unary operators in rule deparser(Tom)
3020 Comment out FileUnlink of excess segments during mdtruncate()(Tom)
3021 IRIX linking fix from Yu Cao >yucao@falcon.kla-tencor.com<
3022 Repair logic error in LIKE: should not return LIKE_ABORT
3023 when reach end of pattern before end of text(Tom)
3024 Repair incorrect cleanup of heap memory allocation during transaction abort(Tom
3026 Updated version of pgaccess 0.98
3028 _________________________________________________________________
3032 Release date: 1999-07-15
3034 This is basically a cleanup release for 6.5. We have fixed a variety
3035 of problems reported by 6.5 users.
3036 _________________________________________________________________
3038 Migration to version 6.5.1
3040 A dump/restore is *not* required for those running 6.5.
3041 _________________________________________________________________
3046 Portability fixes for linux_ppc, IRIX, linux_alpha, OpenBSD, alpha
3047 Remove QUERY_LIMIT, use SELECT...LIMIT
3048 Fix for EXPLAIN on inheritance(Tom)
3049 Patch to allow vacuum on multisegment tables(Hiroshi)
3050 R-Tree optimizer selectivity fix(Tom)
3051 ACL file descriptor leak fix(Atsushi Ogawa)
3052 New expresssion subtree code(Tom)
3053 Avoid disk writes for read-only transactions(Vadim)
3054 Fix for removal of temp tables if last transaction was aborted(Bruce)
3055 Fix to prevent too large row from being created(Bruce)
3057 Allow port numbers 32k - 64k(Bruce)
3058 Add ^ precidence(Bruce)
3059 Rename sort files called pg_temp to pg_sorttemp(Bruce)
3060 Fix for microseconds in time values(Tom)
3061 Tutorial source cleanup
3063 Fix for sorting of NULL's in some cases(Tom)
3064 Shared library dependencies fixed (Tom)
3065 Fixed glitches affecting GROUP BY in subselects(Tom)
3066 Fix some compiler warnings (Tomoaki Nishiyama)
3067 Add Win1250 (Czech) support (Pavel Behal)
3069 _________________________________________________________________
3073 Release date: 1999-06-09
3075 This release marks a major step in the development team's mastery of
3076 the source code we inherited from Berkeley. You will see we are now
3077 easily adding major features, thanks to the increasing size and
3078 experience of our world-wide development team.
3080 Here is a brief summary of the more notable changes:
3082 Multiversion concurrency control(MVCC)
3083 This removes our old table-level locking, and replaces it with
3084 a locking system that is superior to most commercial database
3085 systems. In a traditional system, each row that is modified is
3086 locked until committed, preventing reads by other users. MVCC
3087 uses the natural multiversion nature of PostgreSQL to allow
3088 readers to continue reading consistent data during writer
3089 activity. Writers continue to use the compact pg_log
3090 transaction system. This is all performed without having to
3091 allocate a lock for every row like traditional database
3092 systems. So, basically, we no longer are restricted by simple
3093 table-level locking; we have something better than row-level
3096 Hot backups from pg_dump
3097 pg_dump takes advantage of the new MVCC features to give a
3098 consistent database dump/backup while the database stays online
3099 and available for queries.
3102 We now have a true numeric data type, with user-specified
3106 Temporary tables are guaranteed to have unique names within a
3107 database session, and are destroyed on session exit.
3110 We now have CASE, INTERSECT, and EXCEPT statement support. We
3111 have new LIMIT/OFFSET, SET TRANSACTION ISOLATION LEVEL, SELECT
3112 ... FOR UPDATE, and an improved LOCK TABLE command.
3115 We continue to speed up PostgreSQL, thanks to the variety of
3116 talents within our team. We have sped up memory allocation,
3117 optimization, table joins, and row transfer routines.
3120 We continue to expand our port list, this time including
3121 Windows NT/ix86 and NetBSD/arm32.
3124 Most interfaces have new versions, and existing functionality
3128 New and updated material is present throughout the
3129 documentation. New FAQs have been contributed for SGI and AIX
3130 platforms. The Tutorial has introductory information on SQL
3131 from Stefan Simkovics. For the User's Guide, there are
3132 reference pages covering the postmaster and more utility
3133 programs, and a new appendix contains details on date/time
3134 behavior. The Administrator's Guide has a new chapter on
3135 troubleshooting from Tom Lane. And the Programmer's Guide has a
3136 description of query processing, also from Stefan, and details
3137 on obtaining the PostgreSQL source tree via anonymous CVS and
3139 _________________________________________________________________
3141 Migration to version 6.5
3143 A dump/restore using pg_dump is required for those wishing to migrate
3144 data from any previous release of PostgreSQL. pg_upgrade can *not* be
3145 used to upgrade to this release because the on-disk structure of the
3146 tables has changed compared to previous releases.
3148 The new Multiversion Concurrency Control (MVCC) features can give
3149 somewhat different behaviors in multiuser environments. *Read and
3150 understand the following section to ensure that your existing
3151 applications will give you the behavior you need.*
3152 _________________________________________________________________
3154 Multiversion Concurrency Control
3156 Because readers in 6.5 don't lock data, regardless of transaction
3157 isolation level, data read by one transaction can be overwritten by
3158 another. In other words, if a row is returned by "SELECT" it doesn't
3159 mean that this row really exists at the time it is returned (i.e.
3160 sometime after the statement or transaction began) nor that the row is
3161 protected from being deleted or updated by concurrent transactions
3162 before the current transaction does a commit or rollback.
3164 To ensure the actual existence of a row and protect it against
3165 concurrent updates one must use "SELECT FOR UPDATE" or an appropriate
3166 "LOCK TABLE" statement. This should be taken into account when porting
3167 applications from previous releases of PostgreSQL and other
3170 Keep the above in mind if you are using "contrib/refint.*" triggers
3171 for referential integrity. Additional techniques are required now. One
3172 way is to use "LOCK parent_table IN SHARE ROW EXCLUSIVE MODE" command
3173 if a transaction is going to update/delete a primary key and use "LOCK
3174 parent_table IN SHARE MODE" command if a transaction is going to
3175 update/insert a foreign key.
3177 Note: Note that if you run a transaction in SERIALIZABLE mode then
3178 you must execute the "LOCK" commands above before execution of any
3179 DML statement ("SELECT/INSERT/DELETE/UPDATE/FETCH/COPY_TO") in the
3182 These inconveniences will disappear in the future when the ability to
3183 read dirty (uncommitted) data (regardless of isolation level) and true
3184 referential integrity will be implemented.
3185 _________________________________________________________________
3191 Fix text<->float8 and text<->float4 conversion functions(Thomas)
3192 Fix for creating tables with mixed-case constraints(Billy)
3193 Change exp()/pow() behavior to generate error on underflow/overflow(Jan)
3194 Fix bug in pg_dump -z
3195 Memory overrun cleanups(Tatsuo)
3196 Fix for lo_import crash(Tatsuo)
3197 Adjust handling of data type names to suppress double quotes(Thomas)
3198 Use type coercion for matching columns and DEFAULT(Thomas)
3199 Fix deadlock so it only checks once after one second of sleep(Bruce)
3200 Fixes for aggregates and PL/pgsql(Hiroshi)
3201 Fix for subquery crash(Vadim)
3202 Fix for libpq function PQfnumber and case-insensitive names(Bahman Rafatjoo)
3203 Fix for large object write-in-middle, no extra block, memory consumption(Tatsuo
3205 Fix for pg_dump -d or -D and quote special characters in INSERT
3206 Repair serious problems with dynahash(Tom)
3207 Fix INET/CIDR portability problems
3208 Fix problem with selectivity error in ALTER TABLE ADD COLUMN(Bruce)
3209 Fix executor so mergejoin of different column types works(Tom)
3210 Fix for Alpha OR selectivity bug
3211 Fix OR index selectivity problem(Bruce)
3212 Fix so \d shows proper length for char()/varchar()(Ryan)
3213 Fix tutorial code(Clark)
3214 Improve destroyuser checking(Oliver)
3215 Fix for Kerberos(Rodney McDuff)
3216 Fix for dropping database while dirty buffers(Bruce)
3217 Fix so sequence nextval() can be case-sensitive(Bruce)
3219 Drop buffers before destroying database files(Bruce)
3220 Fix case where executor evaluates functions twice(Tatsuo)
3221 Allow sequence nextval actions to be case-sensitive(Bruce)
3222 Fix optimizer indexing not working for negative numbers(Bruce)
3223 Fix for memory leak in executor with fjIsNull
3224 Fix for aggregate memory leaks(Erik Riedel)
3225 Allow user name containing a dash to grant privileges
3226 Cleanup of NULL in inet types
3227 Clean up system table bugs(Tom)
3228 Fix problems of PAGER and \? command(Masaaki Sakaida)
3229 Reduce default multisegment file size limit to 1GB(Peter)
3230 Fix for dumping of CREATE OPERATOR(Tom)
3231 Fix for backward scanning of cursors(Hiroshi Inoue)
3232 Fix for COPY FROM STDIN when using \i(Tom)
3233 Fix for subselect is compared inside an expression(Jan)
3234 Fix handling of error reporting while returning rows(Tom)
3235 Fix problems with reference to array types(Tom,Jan)
3236 Prevent UPDATE SET oid(Jan)
3237 Fix pg_dump so -t option can handle case-sensitive tablenames
3238 Fixes for GROUP BY in special cases(Tom, Jan)
3239 Fix for memory leak in failed queries(Tom)
3240 DEFAULT now supports mixed-case identifiers(Tom)
3241 Fix for multisegment uses of DROP/RENAME table, indexes(Ole Gjerde)
3242 Disable use of pg_dump with both -o and -d options(Bruce)
3243 Allow pg_dump to properly dump group privileges(Bruce)
3244 Fix GROUP BY in INSERT INTO table SELECT * FROM table2(Jan)
3245 Fix for computations in views(Jan)
3246 Fix for aggregates on array indexes(Tom)
3247 Fix for DEFAULT handles single quotes in value requiring too many quotes
3248 Fix security problem with non-super users importing/exporting large objects(Tom
3250 Rollback of transaction that creates table cleaned up properly(Tom)
3251 Fix to allow long table and column names to generate proper serial names(Tom)
3255 Add "vacuumdb" utility
3256 Speed up libpq by allocating memory better(Tom)
3257 EXPLAIN all indexes used(Tom)
3258 Implement CASE, COALESCE, NULLIF expression(Thomas)
3259 New pg_dump table output format(Constantin)
3260 Add string min()/max() functions(Thomas)
3261 Extend new type coercion techniques to aggregates(Thomas)
3262 New moddatetime contrib(Terry)
3263 Update to pgaccess 0.96(Constantin)
3264 Add routines for single-byte "char" type(Thomas)
3265 Improved substr() function(Thomas)
3266 Improved multibyte handling(Tatsuo)
3267 Multiversion concurrency control/MVCC(Vadim)
3268 New Serialized mode(Vadim)
3269 Fix for tables over 2gigs(Peter)
3270 New SET TRANSACTION ISOLATION LEVEL(Vadim)
3271 New LOCK TABLE IN ... MODE(Vadim)
3272 Update ODBC driver(Byron)
3273 New NUMERIC data type(Jan)
3274 New SELECT FOR UPDATE(Vadim)
3275 Handle "NaN" and "Infinity" for input values(Jan)
3276 Improved date/year handling(Thomas)
3277 Improved handling of backend connections(Magnus)
3278 New options ELOG_TIMESTAMPS and USE_SYSLOG options for log files(Massimo)
3279 New TCL_ARRAYS option(Massimo)
3280 New INTERSECT and EXCEPT(Stefan)
3281 New pg_index.indisprimary for primary key tracking(D'Arcy)
3282 New pg_dump option to allow dropping of tables before creation(Brook)
3283 Speedup of row output routines(Tom)
3284 New READ COMMITTED isolation level(Vadim)
3285 New TEMP tables/indexes(Bruce)
3286 Prevent sorting if result is already sorted(Jan)
3287 New memory allocation optimization(Jan)
3288 Allow psql to do \p\g(Bruce)
3289 Allow multiple rule actions(Jan)
3290 Added LIMIT/OFFSET functionality(Jan)
3291 Improve optimizer when joining a large number of tables(Bruce)
3292 New intro to SQL from S. Simkovics' Master's Thesis (Stefan, Thomas)
3293 New intro to backend processing from S. Simkovics' Master's Thesis (Stefan)
3294 Improved int8 support(Ryan Bradetich, Thomas, Tom)
3295 New routines to convert between int8 and text/varchar types(Thomas)
3296 New bushy plans, where meta-tables are joined(Bruce)
3297 Enable right-hand queries by default(Bruce)
3298 Allow reliable maximum number of backends to be set at configure time
3299 (--with-maxbackends and postmaster switch (-N backends))(Tom)
3300 GEQO default now 10 tables because of optimizer speedups(Tom)
3301 Allow NULL=Var for MS-SQL portability(Michael, Bruce)
3302 Modify contrib check_primary_key() so either "automatic" or "dependent"(Anand)
3303 Allow psql \d on a view show query(Ryan)
3304 Speedup for LIKE(Bruce)
3305 Ecpg fixes/features, see src/interfaces/ecpg/ChangeLog file(Michael)
3306 JDBC fixes/features, see src/interfaces/jdbc/CHANGELOG(Peter)
3307 Make % operator have precedence like /(Bruce)
3308 Add new postgres -O option to allow system table structure changes(Bruce)
3309 Update contrib/pginterface/findoidjoins script(Tom)
3310 Major speedup in vacuum of deleted rows with indexes(Vadim)
3311 Allow non-SQL functions to run different versions based on arguments(Tom)
3312 Add -E option that shows actual queries sent by \dt and friends(Masaaki Sakaida
3314 Add version number in start-up banners for psql(Masaaki Sakaida)
3315 New contrib/vacuumlo removes large objects not referenced(Peter)
3316 New initialization for table sizes so non-vacuumed tables perform better(Tom)
3317 Improve error messages when a connection is rejected(Tom)
3318 Support for arrays of char() and varchar() fields(Massimo)
3319 Overhaul of hash code to increase reliability and performance(Tom)
3320 Update to PyGreSQL 2.4(D'Arcy)
3321 Changed debug options so -d4 and -d5 produce different node displays(Jan)
3322 New pg_options: pretty_plan, pretty_parse, pretty_rewritten(Jan)
3323 Better optimization statistics for system table access(Tom)
3324 Better handling of non-default block sizes(Massimo)
3325 Improve GEQO optimizer memory consumption(Tom)
3326 UNION now suppports ORDER BY of columns not in target list(Jan)
3327 Major libpq++ improvements(Vince Vielhaber)
3328 pg_dump now uses -z(ACL's) as default(Bruce)
3329 backend cache, memory speedups(Tom)
3330 have pg_dump do everything in one snapshot transaction(Vadim)
3331 fix for large object memory leakage, fix for pg_dumping(Tom)
3332 INET type now respects netmask for comparisons
3333 Make VACUUM ANALYZE only use a readlock(Vadim)
3334 Allow VIEWs on UNIONS(Jan)
3335 pg_dump now can generate consistent snapshots on active databases(Vadim)
3339 Improve port matching(Tom)
3340 Portability fixes for SunOS
3341 Add Windows NT backend port and enable dynamic loading(Magnus and Daniel Horak)
3342 New port to Cobalt Qube(Mips) running Linux(Tatsuo)
3343 Port to NetBSD/m68k(Mr. Mutsuki Nakajima)
3344 Port to NetBSD/sun3(Mr. Mutsuki Nakajima)
3345 Port to NetBSD/macppc(Toshimi Aoki)
3346 Fix for tcl/tk configuration(Vince)
3347 Removed CURRENT key word for rule queries(Jan)
3348 NT dynamic loading now works(Daniel Horak)
3349 Add ARM32 support(Andrew McMurry)
3350 Better support for HP-UX 11 and UnixWare
3351 Improve file handling to be more uniform, prevent file descriptor leak(Tom)
3352 New install commands for plpgsql(Jan)
3354 _________________________________________________________________
3358 Release date: 1998-12-20
3360 The 6.4.1 release was improperly packaged. This also has one
3362 _________________________________________________________________
3364 Migration to version 6.4.2
3366 A dump/restore is *not* required for those running 6.4.*.
3367 _________________________________________________________________
3371 Fix for datetime constant problem on some platforms(Thomas)
3372 _________________________________________________________________
3376 Release date: 1998-12-18
3378 This is basically a cleanup release for 6.4. We have fixed a variety
3379 of problems reported by 6.4 users.
3380 _________________________________________________________________
3382 Migration to version 6.4.1
3384 A dump/restore is *not* required for those running 6.4.
3385 _________________________________________________________________
3389 Add pg_dump -N flag to force double quotes around identifiers. This is
3391 Fix for NOT in where clause causing crash(Bruce)
3392 EXPLAIN VERBOSE coredump fix(Vadim)
3393 Fix shared-library problems on Linux
3394 Fix test for table existence to allow mixed-case and whitespace in
3395 the table name(Thomas)
3396 Fix a couple of pg_dump bugs
3397 Configure matches template/.similar entries better(Tom)
3398 Change builtin function names from SPI_* to spi_*
3399 OR WHERE clause fix(Vadim)
3400 Fixes for mixed-case table names(Billy)
3401 contrib/linux/postgres.init.csh/sh fix(Thomas)
3402 libpq memory overrun fix
3404 Change exp() behavior to generate error on underflow(Thomas)
3405 pg_dump fixes for memory leak, inheritance constraints, layout change
3406 update pgaccess to 0.93
3407 Fix prototype for 64-bit platforms
3408 Multibyte fixes(Tatsuo)
3410 Fix memory overruns(Tatsuo)
3411 Fix for lo_import() crash(Bruce)
3412 Better search for install program(Tom)
3415 Use implicit type coercion for matching DEFAULT values(Thomas)
3416 Add routines to help with single-byte (internal) character type(Thomas)
3417 Compilation of libpq for Windows fixes(Magnus)
3418 Upgrade to PyGreSQL 2.2(D'Arcy)
3419 _________________________________________________________________
3423 Release date: 1998-10-30
3425 There are *many* new features and improvements in this release. Thanks
3426 to our developers and maintainers, nearly every aspect of the system
3427 has received some attention since the previous release. Here is a
3428 brief, incomplete summary:
3430 * Views and rules are now functional thanks to extensive new code in
3431 the rewrite rules system from Jan Wieck. He also wrote a chapter
3432 on it for the Programmer's Guide.
3433 * Jan also contributed a second procedural language, PL/pgSQL, to go
3434 with the original PL/pgTCL procedural language he contributed last
3436 * We have optional multiple-byte character set support from Tatsuo
3437 Ishii to complement our existing locale support.
3438 * Client/server communications has been cleaned up, with better
3439 support for asynchronous messages and interrupts thanks to Tom
3441 * The parser will now perform automatic type coercion to match
3442 arguments to available operators and functions, and to match
3443 columns and expressions with target columns. This uses a generic
3444 mechanism which supports the type extensibility features of
3445 PostgreSQL. There is a new chapter in the User's Guide which
3447 * Three new data types have been added. Two types, inet and cidr,
3448 support various forms of IP network, subnet, and machine
3449 addressing. There is now an 8-byte integer type available on some
3450 platforms. See the chapter on data types in the User's Guide for
3451 details. A fourth type, serial, is now supported by the parser as
3452 an amalgam of the int4 type, a sequence, and a unique index.
3453 * Several more SQL92-compatible syntax features have been added,
3454 including "INSERT DEFAULT VALUES"
3455 * The automatic configuration and installation system has received
3456 some attention, and should be more robust for more platforms than
3458 _________________________________________________________________
3460 Migration to version 6.4
3462 A dump/restore using pg_dump or pg_dumpall is required for those
3463 wishing to migrate data from any previous release of PostgreSQL.
3464 _________________________________________________________________
3470 Fix for a tiny memory leak in PQsetdb/PQfinish(Bryan)
3471 Remove char2-16 data types, use char/varchar(Darren)
3472 Pqfn not handles a NOTICE message(Anders)
3473 Reduced busywaiting overhead for spinlocks with many backends (dg)
3474 Stuck spinlock detection (dg)
3475 Fix up "ISO-style" timespan decoding and encoding(Thomas)
3476 Fix problem with table drop after rollback of transaction(Vadim)
3477 Change error message and remove non-functional update message(Vadim)
3478 Fix for COPY array checking
3479 Fix for SELECT 1 UNION SELECT NULL
3480 Fix for buffer leaks in large object calls(Pascal)
3481 Change owner from oid to int4 type(Bruce)
3482 Fix a bug in the oracle compatibility functions btrim() ltrim() and rtrim()
3483 Fix for shared invalidation cache overflow(Massimo)
3484 Prevent file descriptor leaks in failed COPY's(Bruce)
3485 Fix memory leak in libpgtcl's pg_select(Constantin)
3486 Fix problems with username/passwords over 8 characters(Tom)
3487 Fix problems with handling of asynchronous NOTIFY in backend(Tom)
3488 Fix of many bad system table entries(Tom)
3492 Upgrade ecpg and ecpglib,see src/interfaces/ecpc/ChangeLog(Michael)
3493 Show the index used in an EXPLAIN(Zeugswetter)
3494 EXPLAIN invokes rule system and shows plan(s) for rewritten queries(Jan)
3495 Multibyte awareness of many data types and functions, via configure(Tatsuo)
3496 New configure --with-mb option(Tatsuo)
3497 New initdb --pgencoding option(Tatsuo)
3498 New createdb -E multibyte option(Tatsuo)
3499 Select version(); now returns PostgreSQL version(Jeroen)
3500 libpq now allows asynchronous clients(Tom)
3501 Allow cancel from client of backend query(Tom)
3502 psql now cancels query with Control-C(Tom)
3503 libpq users need not issue dummy queries to get NOTIFY messages(Tom)
3504 NOTIFY now sends sender's PID, so you can tell whether it was your own(Tom)
3505 PGresult struct now includes associated error message, if any(Tom)
3506 Define "tz_hour" and "tz_minute" arguments to date_part()(Thomas)
3507 Add routines to convert between varchar and bpchar(Thomas)
3508 Add routines to allow sizing of varchar and bpchar into target columns(Thomas)
3509 Add bit flags to support timezonehour and minute in data retrieval(Thomas)
3510 Allow more variations on valid floating point numbers (e.g. ".1", "1e6")(Thomas
3512 Fixes for unary minus parsing with leading spaces(Thomas)
3513 Implement TIMEZONE_HOUR, TIMEZONE_MINUTE per SQL92 specs(Thomas)
3514 Check for and properly ignore FOREIGN KEY column constraints(Thomas)
3515 Define USER as synonym for CURRENT_USER per SQL92 specs(Thomas)
3516 Enable HAVING clause but no fixes elsewhere yet.
3517 Make "char" type a synonym for "char(1)" (actually implemented as bpchar)(Thoma
3519 Save string type if specified for DEFAULT clause handling(Thomas)
3520 Coerce operations involving different data types(Thomas)
3521 Allow some index use for columns of different types(Thomas)
3522 Add capabilities for automatic type conversion(Thomas)
3523 Cleanups for large objects, so file is truncated on open(Peter)
3524 Readline cleanups(Tom)
3525 Allow psql \f \ to make spaces as delimiter(Bruce)
3526 Pass pg_attribute.atttypmod to the frontend for column field lengths(Tom,Bruce)
3527 Msql compatibility library in /contrib(Aldrin)
3528 Remove the requirement that ORDER/GROUP BY clause identifiers be
3529 included in the target list(David)
3530 Convert columns to match columns in UNION clauses(Thomas)
3531 Remove fork()/exec() and only do fork()(Bruce)
3532 Jdbc cleanups(Peter)
3533 Show backend status on ps command line(only works on some platforms)(Bruce)
3534 Pg_hba.conf now has a sameuser option in the database field
3535 Make lo_unlink take oid param, not int4
3536 New DISABLE_COMPLEX_MACRO for compilers that can't handle our macros(Bruce)
3537 Libpgtcl now handles NOTIFY as a Tcl event, need not send dummy queries(Tom)
3538 libpgtcl cleanups(Tom)
3539 Add -error option to libpgtcl's pg_result command(Tom)
3540 New locale patch, see docs/README/locale(Oleg)
3541 Fix for pg_dump so CONSTRAINT and CHECK syntax is correct(ccb)
3542 New contrib/lo code for large object orphan removal(Peter)
3543 New psql command "SET CLIENT_ENCODING TO 'encoding'" for multibytes
3544 feature, see /doc/README.mb(Tatsuo)
3545 contrib/noupdate code to revoke update permission on a column
3546 libpq can now be compiled on Windows(Magnus)
3547 Add PQsetdbLogin() in libpq
3548 New 8-byte integer type, checked by configure for OS support(Thomas)
3549 Better support for quoted table/column names(Thomas)
3550 Surround table and column names with double-quotes in pg_dump(Thomas)
3551 PQreset() now works with passwords(Tom)
3552 Handle case of GROUP BY target list column number out of range(David)
3553 Allow UNION in subselects
3554 Add auto-size to screen to \d? commands(Bruce)
3555 Use UNION to show all \d? results in one query(Bruce)
3556 Add \d? field search feature(Bruce)
3557 Pg_dump issues fewer \connect requests(Tom)
3558 Make pg_dump -z flag work better, document it in manual page(Tom)
3559 Add HAVING clause with full support for subselects and unions(Stephan)
3560 Full text indexing routines in contrib/fulltextindex(Maarten)
3561 Transaction ids now stored in shared memory(Vadim)
3562 New PGCLIENTENCODING when issuing COPY command(Tatsuo)
3563 Support for SQL92 syntax "SET NAMES"(Tatsuo)
3564 Support for LATIN2-5(Tatsuo)
3565 Add UNICODE regression test case(Tatsuo)
3566 Lock manager cleanup, new locking modes for LLL(Vadim)
3567 Allow index use with OR clauses(Bruce)
3568 Allows "SELECT NULL ORDER BY 1;"
3569 Explain VERBOSE prints the plan, and now pretty-prints the plan to
3570 the postmaster log file(Bruce)
3571 Add indexes display to \d command(Bruce)
3572 Allow GROUP BY on functions(David)
3573 New pg_class.relkind for large objects(Bruce)
3574 New way to send libpq NOTICE messages to a different location(Tom)
3575 New \w write command to psql(Bruce)
3576 New /contrib/findoidjoins scans oid columns to find join relationships(Bruce)
3577 Allow binary-compatible indexes to be considered when checking for valid
3578 Indexes for restriction clauses containing a constant(Thomas)
3579 New ISBN/ISSN code in /contrib/isbn_issn
3580 Allow NOT LIKE, IN, NOT IN, BETWEEN, and NOT BETWEEN constraint(Thomas)
3581 New rewrite system fixes many problems with rules and views(Jan)
3582 * Rules on relations work
3583 * Event qualifications on insert/update/delete work
3584 * New OLD variable to reference CURRENT, CURRENT will be remove in futu
3586 * Update rules can reference NEW and OLD in rule qualifications/actions
3587 * Insert/update/delete rules on views work
3588 * Multiple rule actions are now supported, surrounded by parentheses
3589 * Regular users can create views/rules on tables they have RULE permits
3590 * Rules and views inherit the privileges of the creator
3591 * No rules at the column level
3592 * No UPDATE NEW/OLD rules
3593 * New pg_tables, pg_indexes, pg_rules and pg_views system views
3594 * Only a single action on SELECT rules
3595 * Total rewrite overhaul, perhaps for 6.5
3597 * handle aggregates on views
3598 * handle insert into select from view works
3599 System indexes are now multikey(Bruce)
3600 Oidint2, oidint4, and oidname types are removed(Bruce)
3601 Use system cache for more system table lookups(Bruce)
3602 New backend programming language PL/pgSQL in backend/pl(Jan)
3603 New SERIAL data type, auto-creates sequence/index(Thomas)
3604 Enable assert checking without a recompile(Massimo)
3605 User lock enhancements(Massimo)
3606 New setval() command to set sequence value(Massimo)
3607 Auto-remove unix socket file on start-up if no postmaster running(Massimo)
3608 Conditional trace package(Massimo)
3609 New UNLISTEN command(Massimo)
3610 psql and libpq now compile under Windows using win32.mak(Magnus)
3611 Lo_read no longer stores trailing NULL(Bruce)
3612 Identifiers are now truncated to 31 characters internally(Bruce)
3613 Createuser options now availble on the command line
3614 Code for 64-bit integer supported added, configure tested, int8 type(Thomas)
3615 Prevent file descriptor leaf from failed COPY(Bruce)
3616 New pg_upgrade command(Bruce)
3617 Updated /contrib directories(Massimo)
3618 New CREATE TABLE DEFAULT VALUES statement available(Thomas)
3619 New INSERT INTO TABLE DEFAULT VALUES statement available(Thomas)
3620 New DECLARE and FETCH feature(Thomas)
3621 libpq's internal structures now not exported(Tom)
3622 Allow up to 8 key indexes(Bruce)
3623 Remove ARCHIVE key word, that is no longer used(Thomas)
3624 pg_dump -n flag to supress quotes around indentifiers
3625 disable system columns for views(Jan)
3626 new INET and CIDR types for network addresses(TomH, Paul)
3627 no more double quotes in psql output
3628 pg_dump now dumps views(Terry)
3629 new SET QUERY_LIMIT(Tatsuo,Jan)
3633 /contrib cleanup(Jun)
3634 Inline some small functions called for every row(Bruce)
3637 Multibyte regression tests(Soonmyung.)
3638 Remove --disabled options from configure
3639 Define PGDOC to use POSTGRESDIR by default
3640 Make regression optional
3641 Remove extra braces code to pgindent(Bruce)
3642 Add bsdi shared library support(Bruce)
3643 New --without-CXX support configure option(Brook)
3645 Update backend flowchart in tools/backend(Bruce)
3646 Change atttypmod from int16 to int32(Bruce, Tom)
3647 Getrusage() fix for platforms that do not have it(Tom)
3648 Add PQconnectdb, PGUSER, PGPASSWORD to libpq man page
3649 NS32K platform fixes(Phil Nelson, John Buller)
3650 SCO 7/UnixWare 2.x fixes(Billy,others)
3651 Sparc/Solaris 2.5 fixes(Ryan)
3652 Pgbuiltin.3 is obsolete, move to doc files(Thomas)
3653 Even more documention(Thomas)
3654 Nextstep support(Jacek)
3656 pginterface manual page(Bruce)
3657 shared libraries all have version numbers
3658 merged all OS-specific shared library defines into one file
3659 smarter TCL/TK configuration checking(Billy)
3660 smarter perl configuration(Brook)
3661 configure uses supplied install-sh if no install script found(Tom)
3662 new Makefile.shlib for shared library configuration(Tom)
3663 _________________________________________________________________
3667 Release date: 1998-04-07
3669 This is a bug-fix release for 6.3.x. Refer to the release notes for
3670 version 6.3 for a more complete summary of new features.
3674 * Repairs automatic configuration support for some platforms,
3675 including Linux, from breakage inadvertently introduced in version
3677 * Correctly handles function calls on the left side of BETWEEN and
3680 A dump/restore is NOT required for those running 6.3 or 6.3.1. A make
3681 distclean, make, and make install is all that is required. This last
3682 step should be performed while the postmaster is not running. You
3683 should re-link any custom applications that use PostgreSQL libraries.
3685 For upgrades from pre-6.3 installations, refer to the installation and
3686 migration instructions for version 6.3.
3687 _________________________________________________________________
3691 Configure detection improvements for tcl/tk(Brook Milligan, Alvin)
3692 Manual page improvements(Bruce)
3693 BETWEEN and LIKE fix(Thomas)
3694 fix for psql \connect used by pg_dump(Oliver Elphick)
3696 pgaccess, version 0.86
3697 qsort removed, now uses libc version, cleanups(Jeroen)
3698 fix for buffer over-runs detected(Maurice Gittens)
3699 fix for buffer overrun in libpgtcl(Randy Kunkee)
3700 fix for UNION with DISTINCT or ORDER BY(Bruce)
3701 gettimeofday configure check(Doug Winterburn)
3702 Fix "indexes not used" bug(Vadim)
3703 docs additions(Thomas)
3704 Fix for backend memory leak(Bruce)
3705 libreadline cleanup(Erwan MAS)
3706 Remove DISTDIR(Bruce)
3707 Makefile dependency cleanup(Jeroen van Vianen)
3710 _________________________________________________________________
3714 Release date: 1998-03-23
3718 * Additional support for multibyte character sets.
3719 * Repair byte ordering for mixed-endian clients and servers.
3720 * Minor updates to allowed SQL syntax.
3721 * Improvements to the configuration autodetection for installation.
3723 A dump/restore is NOT required for those running 6.3. A make
3724 distclean, make, and make install is all that is required. This last
3725 step should be performed while the postmaster is not running. You
3726 should re-link any custom applications that use PostgreSQL libraries.
3728 For upgrades from pre-6.3 installations, refer to the installation and
3729 migration instructions for version 6.3.
3730 _________________________________________________________________
3734 ecpg cleanup/fixes, now version 1.1(Michael Meskes)
3735 pg_user cleanup(Bruce)
3736 large object fix for pg_dump and tclsh (alvin)
3737 LIKE fix for multiple adjacent underscores
3738 fix for redefining builtin functions(Thomas)
3740 upgrade to pg_access 0.83
3741 updated CLUSTER manual page
3742 multibyte character set support, see doc/README.mb(Tatsuo)
3743 configure --with-pgport fix
3745 big-endian fix for backend communications(Kataoka)
3746 SUBSTR() and substring() fix(Jan)
3747 several jdbc fixes(Peter)
3748 libpgtcl improvements, see libptcl/README(Randy Kunkee)
3749 Fix for "Datasize = 0" error(Vadim)
3750 Prevent \do from wrapping(Bruce)
3751 Remove duplicate Russian character set entries
3753 Allow optional TABLE key word in LOCK and SELECT INTO(Thomas)
3754 CREATE SEQUENCE options to allow a negative integer(Thomas)
3755 Add "PASSWORD" as an allowed column identifier(Thomas)
3756 Add checks for UNION target fields(Bruce)
3757 Fix Alpha port(Dwayne Bailey)
3758 Fix for text arrays containing quotes(Doug Gibson)
3759 Solaris compile fix(Albert Chin-A-Young)
3760 Better identify tcl and tk libs and includes(Bruce)
3762 _________________________________________________________________
3766 Release date: 1998-03-01
3768 There are *many* new features and improvements in this release. Here
3769 is a brief, incomplete summary:
3771 * Many new SQL features, including full SQL92 subselect capability
3772 (everything is here but target-list subselects).
3773 * Support for client-side environment variables to specify time zone
3775 * Socket interface for client/server connection. This is the default
3776 now so you may need to start postmaster with the "-i" flag.
3777 * Better password authorization mechanisms. Default table privileges
3779 * Old-style time travel has been removed. Performance has been
3782 Note: Bruce Momjian wrote the following notes to introduce the new
3785 There are some general 6.3 issues that I want to mention. These are
3786 only the big items that can not be described in one sentence. A review
3787 of the detailed changes list is still needed.
3789 First, we now have subselects. Now that we have them, I would like to
3790 mention that without subselects, SQL is a very limited language.
3791 Subselects are a major feature, and you should review your code for
3792 places where subselects provide a better solution for your queries. I
3793 think you will find that there are more uses for subselects than you
3794 may think. Vadim has put us on the big SQL map with subselects, and
3795 fully functional ones too. The only thing you can't do with subselects
3796 is to use them in the target list.
3798 Second, 6.3 uses Unix domain sockets rather than TCP/IP by default. To
3799 enable connections from other machines, you have to use the new
3800 postmaster -i option, and of course edit "pg_hba.conf". Also, for this
3801 reason, the format of "pg_hba.conf" has changed.
3803 Third, char() fields will now allow faster access than varchar() or
3804 text. Specifically, the text and varchar() have a penalty for access
3805 to any columns after the first column of this type. char() used to
3806 also have this access penalty, but it no longer does. This may suggest
3807 that you redesign some of your tables, especially if you have short
3808 character columns that you have defined as varchar() or text. This and
3809 other changes make 6.3 even faster than earlier releases.
3811 We now have passwords definable independent of any Unix file. There
3812 are new SQL USER commands. See the Administrator's Guide for more
3813 information. There is a new table, pg_shadow, which is used to store
3814 user information and user passwords, and it by default only
3815 SELECT-able by the postgres super-user. pg_user is now a view of
3816 pg_shadow, and is SELECT-able by PUBLIC. You should keep using pg_user
3817 in your application without changes.
3819 User-created tables now no longer have SELECT privilege to PUBLIC by
3820 default. This was done because the ANSI standard requires it. You can
3821 of course GRANT any privileges you want after the table is created.
3822 System tables continue to be SELECT-able by PUBLIC.
3824 We also have real deadlock detection code. No more sixty-second
3825 timeouts. And the new locking code implements a FIFO better, so there
3826 should be less resource starvation during heavy use.
3828 Many complaints have been made about inadequate documentation in
3829 previous releases. Thomas has put much effort into many new manuals
3830 for this release. Check out the doc/ directory.
3832 For performance reasons, time travel is gone, but can be implemented
3833 using triggers (see "pgsql/contrib/spi/README"). Please check out the
3834 new \d command for types, operators, etc. Also, views have their own
3835 privileges now, not based on the underlying tables, so privileges on
3836 them have to be set separately. Check "/pgsql/interfaces" for some new
3837 ways to talk to PostgreSQL.
3839 This is the first release that really required an explanation for
3840 existing users. In many ways, this was necessary because the new
3841 release removes many limitations, and the work-arounds people were
3842 using are no longer needed.
3843 _________________________________________________________________
3845 Migration to version 6.3
3847 A dump/restore using pg_dump or pg_dumpall is required for those
3848 wishing to migrate data from any previous release of PostgreSQL.
3849 _________________________________________________________________
3855 Fix binary cursors broken by MOVE implementation(Vadim)
3856 Fix for tcl library crash(Jan)
3857 Fix for array handling, from Gerhard Hintermayer
3858 Fix acl error, and remove duplicate pqtrace(Bruce)
3859 Fix psql \e for empty file(Bruce)
3860 Fix for textcat on varchar() fields(Bruce)
3861 Fix for DBT Sendproc (Zeugswetter Andres)
3862 Fix vacuum analyze syntax problem(Bruce)
3863 Fix for international identifiers(Tatsuo)
3864 Fix aggregates on inherited tables(Bruce)
3865 Fix substr() for out-of-bounds data
3866 Fix for select 1=1 or 2=2, select 1=1 and 2=2, and select sum(2+2)(Bruce)
3867 Fix notty output to show status result. -q option still turns it off(Bruce)
3868 Fix for count(*), aggs with views and multiple tables and sum(3)(Bruce)
3870 Fix for PQtrace start/stop several times(Bruce)
3871 Fix a variety of locking problems like newer lock waiters getting
3872 lock before older waiters, and having readlock people not share
3873 locks if a writer is waiting for a lock, and waiting writers not
3874 getting priority over waiting readers(Bruce)
3875 Fix crashes in psql when executing queries from external files(James)
3876 Fix problem with multiple order by columns, with the first one having
3878 Use correct hash table support functions for float8 and int4(Thomas)
3879 Re-enable JOIN= option in CREATE OPERATOR statement (Thomas)
3880 Change precedence for boolean operators to match expected behavior(Thomas)
3881 Generate elog(ERROR) on over-large integer(Bruce)
3882 Allow multiple-argument functions in constraint clauses(Thomas)
3883 Check boolean input literals for 'true','false','yes','no','1','0'
3884 and throw elog(ERROR) if unrecognized(Thomas)
3885 Major large objects fix
3886 Fix for GROUP BY showing duplicates(Vadim)
3887 Fix for index scans in MergeJion(Vadim)
3891 Subselects with EXISTS, IN, ALL, ANY key words (Vadim, Bruce, Thomas)
3892 New User Manual(Thomas, others)
3893 Speedup by inlining some frequently-called functions
3894 Real deadlock detection, no more timeouts(Bruce)
3895 Add SQL92 "constants" CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP,
3896 CURRENT_USER(Thomas)
3897 Modify constraint syntax to be SQL92-compliant(Thomas)
3898 Implement SQL92 PRIMARY KEY and UNIQUE clauses using indexes(Thomas)
3899 Recognize SQL92 syntax for FOREIGN KEY. Throw elog notice(Thomas)
3900 Allow NOT NULL UNIQUE constraint clause (each allowed separately before)(Thomas
3902 Allow PostgreSQL-style casting ("::") of non-constants(Thomas)
3903 Add support for SQL3 TRUE and FALSE boolean constants(Thomas)
3904 Support SQL92 syntax for IS TRUE/IS FALSE/IS NOT TRUE/IS NOT FALSE(Thomas)
3905 Allow shorter strings for boolean literals (e.g. "t", "tr", "tru")(Thomas)
3906 Allow SQL92 delimited identifiers(Thomas)
3907 Implement SQL92 binary and hexadecimal string decoding (b'10' and x'1F')(Thomas
3909 Support SQL92 syntax for type coercion of literal strings
3910 (e.g. "DATETIME 'now'")(Thomas)
3911 Add conversions for int2, int4, and OID types to and from text(Thomas)
3912 Use shared lock when building indexes(Vadim)
3913 Free memory allocated for an user query inside transaction block after
3914 this query is done, was turned off in <= 6.2.1(Vadim)
3915 New SQL statement CREATE PROCEDURAL LANGUAGE(Jan)
3916 New PostgreSQL Procedural Language (PL) backend interface(Jan)
3917 Rename pg_dump -H option to -h(Bruce)
3918 Add Java support for passwords, European dates(Peter)
3919 Use indexes for LIKE and ~, !~ operations(Bruce)
3920 Add hash functions for datetime and timespan(Thomas)
3921 Time Travel removed(Vadim, Bruce)
3922 Add paging for \d and \z, and fix \i(Bruce)
3923 Add Unix domain socket support to backend and to frontend library(Goran)
3924 Implement CREATE DATABASE/WITH LOCATION and initlocation utility(Thomas)
3925 Allow more SQL92 and/or PostgreSQL reserved words as column identifiers(Thomas)
3926 Augment support for SQL92 SET TIME ZONE...(Thomas)
3927 SET/SHOW/RESET TIME ZONE uses TZ backend environment variable(Thomas)
3928 Implement SET keyword = DEFAULT and SET TIME ZONE DEFAULT(Thomas)
3929 Enable SET TIME ZONE using TZ environment variable(Thomas)
3930 Add PGDATESTYLE environment variable to frontend and backend initialization(Tho
3932 Add PGTZ, PGCOSTHEAP, PGCOSTINDEX, PGRPLANS, PGGEQO
3933 frontend library initialization environment variables(Thomas)
3934 Regression tests time zone automatically set with "setenv PGTZ PST8PDT"(Thomas)
3935 Add pg_description table for info on tables, columns, operators, types, and
3937 Increase 16 char limit on system table/index names to 32 characters(Bruce)
3938 Rename system indexes(Bruce)
3939 Add 'GERMAN' option to SET DATESTYLE(Thomas)
3940 Define an "ISO-style" timespan output format with "hh:mm:ss" fields(Thomas)
3941 Allow fractional values for delta times (e.g. '2.5 days')(Thomas)
3942 Validate numeric input more carefully for delta times(Thomas)
3943 Implement day of year as possible input to date_part()(Thomas)
3944 Define timespan_finite() and text_timespan() functions(Thomas)
3945 Remove archive stuff(Bruce)
3946 Allow for a pg_password authentication database that is separate from
3947 the system password file(Todd)
3948 Dump ACLs, GRANT, REVOKE privileges(Matt)
3949 Define text, varchar, and bpchar string length functions(Thomas)
3950 Fix Query handling for inheritance, and cost computations(Bruce)
3951 Implement CREATE TABLE/AS SELECT (alternative to SELECT/INTO)(Thomas)
3952 Allow NOT, IS NULL, IS NOT NULL in constraints(Thomas)
3953 Implement UNIONs for SELECT(Bruce)
3954 Add UNION, GROUP, DISTINCT to INSERT(Bruce)
3955 varchar() stores only necessary bytes on disk(Bruce)
3956 Fix for BLOBs(Peter)
3957 Mega-Patch for JDBC...see README_6.3 for list of changes(Peter)
3958 Remove unused "option" from PQconnectdb()
3959 New LOCK command and lock manual page describing deadlocks(Bruce)
3960 Add new psql \da, \dd, \df, \do, \dS, and \dT commands(Bruce)
3961 Enhance psql \z to show sequences(Bruce)
3962 Show NOT NULL and DEFAULT in psql \d table(Bruce)
3963 New psql .psqlrc file start-up(Andrew)
3964 Modify sample start-up script in contrib/linux to show syslog(Thomas)
3965 New types for IP and MAC addresses in contrib/ip_and_mac(TomH)
3966 Unix system time conversions with date/time types in contrib/unixdate(Thomas)
3967 Update of contrib stuff(Massimo)
3968 Add Unix socket support to DBD::Pg(Goran)
3969 New python interface (PyGreSQL 2.0)(D'Arcy)
3970 New frontend/backend protocol has a version number, network byte order(Phil)
3971 Security features in pg_hba.conf enhanced and documented, many cleanups(Phil)
3972 CHAR() now faster access than VARCHAR() or TEXT
3973 ecpg embedded SQL preprocessor
3974 Reduce system column overhead(Vadmin)
3975 Remove pg_time table(Vadim)
3976 Add pg_type attribute to identify types that need length (bpchar, varchar)
3977 Add report of offending line when COPY command fails
3978 Allow VIEW privileges to be set separately from the underlying tables.
3979 For security, use GRANT/REVOKE on views as appropriate(Jan)
3980 Tables now have no default GRANT SELECT TO PUBLIC. You must
3981 explicitly grant such privileges.
3982 Clean up tutorial examples(Darren)
3986 Add new html development tools, and flow chart in /tools/backend
3987 Fix for SCO compiles
3988 Stratus computer port Robert Gillies
3989 Added support for shlib for BSD44_derived & i386_solaris
3990 Make configure more automated(Brook)
3991 Add script to check regression test results
3992 Break parser functions into smaller files, group together(Bruce)
3993 Rename heap_create to heap_create_and_catalog, rename heap_creatr
3994 to heap_create()(Bruce)
3995 Sparc/Linux patch for locking(TomS)
3996 Remove PORTNAME and reorganize port-specific stuff(Marc)
3997 Add optimizer README file(Bruce)
3998 Remove some recursion in optimizer and clean up some code there(Bruce)
3999 Fix for NetBSD locking(Henry)
4000 Fix for libptcl make(Tatsuo)
4002 Change IS TRUE, IS FALSE, ... to expressions using "=" rather than
4003 function calls to istrue() or isfalse() to allow optimization(Thomas)
4004 Various fixes NetBSD/Sparc related(TomH)
4005 Alpha linux locking(Travis,Ryan)
4006 Change elog(WARN) to elog(ERROR)(Bruce)
4007 FAQ for FreeBSD(Marc)
4008 Bring in the PostODBC source tree as part of our standard distribution(Marc)
4009 A minor patch for HP/UX 10 vs 9(Stan)
4010 New pg_attribute.atttypmod for type-specific info like varchar length(Bruce)
4011 UnixWare patches(Billy)
4012 New i386 'lock' for spinlock asm(Billy)
4013 Support for multiplexed backends is removed
4014 Start an OpenBSD port
4017 Add string functions to regression suite(Thomas)
4018 Expand a few function names formerly truncated to 16 characters(Thomas)
4019 Remove un-needed malloc() calls and replace with palloc()(Bruce)
4020 _________________________________________________________________
4024 Release date: 1997-10-17
4026 6.2.1 is a bug-fix and usability release on 6.2.
4030 * Allow strings to span lines, per SQL92.
4031 * Include example trigger function for inserting user names on table
4034 This is a minor bug-fix release on 6.2. For upgrades from pre-6.2
4035 systems, a full dump/reload is required. Refer to the 6.2 release
4036 notes for instructions.
4037 _________________________________________________________________
4039 Migration from version 6.2 to version 6.2.1
4041 This is a minor bug-fix release. A dump/reload is not required from
4042 version 6.2, but is required from any release prior to 6.2.
4044 In upgrading from version 6.2, if you choose to dump/reload you will
4045 find that avg(money) is now calculated correctly. All other bug fixes
4046 take effect upon updating the executables.
4048 Another way to avoid dump/reload is to use the following SQL command
4049 from "psql" to update the existing system table:
4050 update pg_aggregate set aggfinalfn = 'cash_div_flt8'
4051 where aggname = 'avg' and aggbasetype = 790;
4053 This will need to be done to every existing database, including
4055 _________________________________________________________________
4059 Allow TIME and TYPE column names(Thomas)
4060 Allow larger range of true/false as boolean values(Thomas)
4061 Support output of "now" and "current"(Thomas)
4062 Handle DEFAULT with INSERT of NULL properly(Vadim)
4063 Fix for relation reference counts problem in buffer manager(Vadim)
4064 Allow strings to span lines, like ANSI(Thomas)
4065 Fix for backward cursor with ORDER BY(Vadim)
4066 Fix avg(cash) computation(Thomas)
4067 Fix for specifying a column twice in ORDER/GROUP BY(Vadim)
4068 Documented new libpq function to return affected rows, PQcmdTuples(Bruce)
4069 Trigger function for inserting user names for INSERT/UPDATE(Brook Milligan)
4071 _________________________________________________________________
4075 Release date: 1997-10-02
4077 A dump/restore is required for those wishing to migrate data from
4078 previous releases of PostgreSQL.
4079 _________________________________________________________________
4081 Migration from version 6.1 to version 6.2
4083 This migration requires a complete dump of the 6.1 database and a
4084 restore of the database in 6.2.
4086 Note that the "pg_dump" and "pg_dumpall" utility from 6.2 should be
4087 used to dump the 6.1 database.
4088 _________________________________________________________________
4090 Migration from version 1.x to version 6.2
4092 Those migrating from earlier 1.* releases should first upgrade to 1.09
4093 because the COPY output format was improved from the 1.02 release.
4094 _________________________________________________________________
4100 Fix problems with pg_dump for inheritance, sequences, archive tables(Bruce)
4101 Fix compile errors on overflow due to shifts, unsigned, and bad prototypes
4102 from Solaris(Diab Jerius)
4103 Fix bugs in geometric line arithmetic (bad intersection calculations)(Thomas)
4104 Check for geometric intersections at endpoints to avoid rounding ugliness(Thoma
4106 Catch non-functional delete attempts(Vadim)
4107 Change time function names to be more consistent(Michael Reifenberg)
4108 Check for zero divides(Michael Reifenberg)
4109 Fix very old bug which made rows changed/inserted by a command
4110 visible to the command itself (so we had multiple update of
4111 updated rows, etc.)(Vadim)
4112 Fix for SELECT null, 'fail' FROM pg_am (Patrick)
4113 SELECT NULL as EMPTY_FIELD now allowed(Patrick)
4114 Remove un-needed signal stuff from contrib/pginterface
4115 Fix OR (where x != 1 or x isnull didn't return rows with x NULL) (Vadim)
4116 Fix time_cmp function (Vadim)
4117 Fix handling of functions with non-attribute first argument in
4118 WHERE clauses (Vadim)
4119 Fix GROUP BY when order of entries is different from order
4120 in target list (Vadim)
4121 Fix pg_dump for aggregates without sfunc1 (Vadim)
4125 Default genetic optimizer GEQO parameter is now 8(Bruce)
4126 Allow use parameters in target list having aggregates in functions(Vadim)
4127 Added JDBC driver as an interface(Adrian & Peter)
4129 Return number of rows inserted/affected by INSERT/UPDATE/DELETE etc.(Vadim)
4130 Triggers implemented with CREATE TRIGGER (SQL3)(Vadim)
4131 SPI (Server Programming Interface) allows execution of queries inside
4133 NOT NULL implemented (SQL92)(Robson Paniago de Miranda)
4134 Include reserved words for string handling, outer joins, and unions(Thomas)
4135 Implement extended comments ("/* ... */") using exclusive states(Thomas)
4136 Add "//" single-line comments(Bruce)
4137 Remove some restrictions on characters in operator names(Thomas)
4138 DEFAULT and CONSTRAINT for tables implemented (SQL92)(Vadim & Thomas)
4139 Add text concatenation operator and function (SQL92)(Thomas)
4140 Support WITH TIME ZONE syntax (SQL92)(Thomas)
4141 Support INTERVAL unit TO unit syntax (SQL92)(Thomas)
4142 Define types DOUBLE PRECISION, INTERVAL, CHARACTER,
4143 and CHARACTER VARYING (SQL92)(Thomas)
4144 Define type FLOAT(p) and rudimentary DECIMAL(p,s), NUMERIC(p,s) (SQL92)(Thomas)
4145 Define EXTRACT(), POSITION(), SUBSTRING(), and TRIM() (SQL92)(Thomas)
4146 Define CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP (SQL92)(Thomas)
4147 Add syntax and warnings for UNION, HAVING, INNER and OUTER JOIN (SQL92)(Thomas)
4148 Add more reserved words, mostly for SQL92 compliance(Thomas)
4149 Allow hh:mm:ss time entry for timespan/reltime types(Thomas)
4150 Add center() routines for lseg, path, polygon(Thomas)
4151 Add distance() routines for circle-polygon, polygon-polygon(Thomas)
4152 Check explicitly for points and polygons contained within polygons
4153 using an axis-crossing algorithm(Thomas)
4154 Add routine to convert circle-box(Thomas)
4155 Merge conflicting operators for different geometric data types(Thomas)
4156 Replace distance operator "<===>" with "<->"(Thomas)
4157 Replace "above" operator "!^" with ">^" and "below" operator "!|" with "<^"(Tho
4159 Add routines for text trimming on both ends, substring, and string position(Tho
4161 Added conversion routines circle(box) and poly(circle)(Thomas)
4162 Allow internal sorts to be stored in memory rather than in files(Bruce & Vadim)
4163 Allow functions and operators on internally-identical types to succeed(Bruce)
4164 Speed up backend start-up after profiling analysis(Bruce)
4165 Inline frequently called functions for performance(Bruce)
4166 Reduce open() calls(Bruce)
4167 psql: Add PAGER for \h and \?,\C fix
4168 Fix for psql pager when no tty(Bruce)
4169 New entab utility(Bruce)
4170 General trigger functions for referential integrity (Vadim)
4171 General trigger functions for time travel (Vadim)
4172 General trigger functions for AUTOINCREMENT/IDENTITY feature (Vadim)
4173 MOVE implementation (Vadim)
4177 HP-UX 10 patches (Vladimir Turin)
4178 Added SCO support, (Daniel Harris)
4179 MkLinux patches (Tatsuo Ishii)
4180 Change geometric box terminology from "length" to "width"(Thomas)
4181 Deprecate temporary unstored slope fields in geometric code(Thomas)
4182 Remove restart instructions from INSTALL(Bruce)
4183 Look in /usr/ucb first for install(Bruce)
4184 Fix c++ copy example code(Thomas)
4185 Add -o to psql manual page(Bruce)
4186 Prevent relname unallocated string length from being copied into database(Bruce
4188 Cleanup for NAMEDATALEN use(Bruce)
4189 Fix pg_proc names over 15 chars in output(Bruce)
4190 Add strNcpy() function(Bruce)
4191 remove some (void) casts that are unnecessary(Bruce)
4192 new interfaces directory(Marc)
4193 Replace fopen() calls with calls to fd.c functions(Bruce)
4194 Make functions static where possible(Bruce)
4195 enclose unused functions in #ifdef NOT_USED(Bruce)
4196 Remove call to difftime() in timestamp support to fix SunOS(Bruce & Thomas)
4197 Changes for Digital Unix
4198 Portability fix for pg_dumpall(Bruce)
4199 Rename pg_attribute.attnvals to attdispersion(Bruce)
4200 "intro/unix" manual page now "pgintro"(Bruce)
4201 "built-in" manual page now "pgbuiltin"(Bruce)
4202 "drop" manual page now "drop_table"(Bruce)
4203 Add "create_trigger", "drop_trigger" manual pages(Thomas)
4204 Add constraints regression test(Vadim & Thomas)
4205 Add comments syntax regression test(Thomas)
4206 Add PGINDENT and support program(Bruce)
4207 Massive commit to run PGINDENT on all *.c and *.h files(Bruce)
4208 Files moved to /src/tools directory(Bruce)
4209 SPI and Trigger programming guides (Vadim & D'Arcy)
4210 _________________________________________________________________
4214 Release date: 1997-07-22
4215 _________________________________________________________________
4217 Migration from version 6.1 to version 6.1.1
4219 This is a minor bug-fix release. A dump/reload is not required from
4220 version 6.1, but is required from any release prior to 6.1. Refer to
4221 the release notes for 6.1 for more details.
4222 _________________________________________________________________
4226 fix for SET with options (Thomas)
4227 allow pg_dump/pg_dumpall to preserve ownership of all tables/objects(Bruce)
4228 new psql \connect option allows changing usernames without changing databases
4229 fix for initdb --debug option(Yoshihiko Ichikawa))
4230 lextest cleanup(Bruce)
4232 fix date/time month boundary arithmetic(Thomas)
4233 fix timezone daylight handling for some ports(Thomas, Bruce, Tatsuo)
4234 timestamp overhauled to use standard functions(Thomas)
4235 other code cleanup in date/time routines(Thomas)
4236 psql's \d now case-insensitive(Bruce)
4237 psql's backslash commands can now have trailing semicolon(Bruce)
4238 fix memory leak in psql when using \g(Bruce)
4239 major fix for endian handling of communication to server(Thomas, Tatsuo)
4240 Fix for Solaris assembler and include files(Yoshihiko Ichikawa)
4241 allow underscores in usernames(Bruce)
4242 pg_dumpall now returns proper status, portability fix(Bruce)
4244 _________________________________________________________________
4248 Release date: 1997-06-08
4250 The regression tests have been adapted and extensively modified for
4251 the 6.1 release of PostgreSQL.
4253 Three new data types (datetime, timespan, and circle) have been added
4254 to the native set of PostgreSQL types. Points, boxes, paths, and
4255 polygons have had their output formats made consistent across the data
4256 types. The polygon output in misc.out has only been spot-checked for
4257 correctness relative to the original regression output.
4259 PostgreSQL 6.1 introduces a new, alternate optimizer which uses
4260 genetic algorithms. These algorithms introduce a random behavior in
4261 the ordering of query results when the query contains multiple
4262 qualifiers or multiple tables (giving the optimizer a choice on order
4263 of evaluation). Several regression tests have been modified to
4264 explicitly order the results, and hence are insensitive to optimizer
4265 choices. A few regression tests are for data types which are
4266 inherently unordered (e.g. points and time intervals) and tests
4267 involving those types are explicitly bracketed with "set geqo to
4268 'off'" and "reset geqo".
4270 The interpretation of array specifiers (the curly braces around atomic
4271 values) appears to have changed sometime after the original regression
4272 tests were generated. The current "./expected/*.out" files reflect
4273 this new interpretation, which may not be correct!
4275 The float8 regression test fails on at least some platforms. This is
4276 due to differences in implementations of pow() and exp() and the
4277 signaling mechanisms used for overflow and underflow conditions.
4279 The "random" results in the random test should cause the "random" test
4280 to be "failed", since the regression tests are evaluated using a
4281 simple diff. However, "random" does not seem to produce random results
4282 on my test machine (Linux/gcc/i686).
4283 _________________________________________________________________
4285 Migration to version 6.1
4287 This migration requires a complete dump of the 6.0 database and a
4288 restore of the database in 6.1.
4290 Those migrating from earlier 1.* releases should first upgrade to 1.09
4291 because the COPY output format was improved from the 1.02 release.
4292 _________________________________________________________________
4298 packet length checking in library routines
4299 lock manager priority patch
4300 check for under/over flow of float8(Bruce)
4301 multitable join fix(Vadim)
4302 SIGPIPE crash fix(Darren)
4303 large object fixes(Sven)
4304 allow btree indexes to handle NULLs(Vadim)
4305 timezone fixes(D'Arcy)
4306 select SUM(x) can return NULL on no rows(Thomas)
4307 internal optimizer, executor bug fixes(Vadim)
4308 fix problem where inner loop in < or <= has no rows(Vadim)
4309 prevent re-commuting join index clauses(Vadim)
4310 fix join clauses for multiple tables(Vadim)
4311 fix hash, hashjoin for arrays(Vadim)
4312 fix btree for abstime type(Vadim)
4313 large object fixes(Raymond)
4314 fix buffer leak in hash indexes (Vadim)
4315 fix rtree for use in inner scan (Vadim)
4316 fix gist for use in inner scan, cleanups (Vadim, Andrea)
4317 avoid unnecessary local buffers allocation (Vadim, Massimo)
4318 fix local buffers leak in transaction aborts (Vadim)
4319 fix file manager memmory leaks, cleanups (Vadim, Massimo)
4320 fix storage manager memmory leaks (Vadim)
4321 fix btree duplicates handling (Vadim)
4322 fix deleted rows reincarnation caused by vacuum (Vadim)
4323 fix SELECT varchar()/char() INTO TABLE made zero-length fields(Bruce)
4324 many psql, pg_dump, and libpq memory leaks fixed using Purify (Igor)
4328 attribute optimization statistics(Bruce)
4329 much faster new btree bulk load code(Paul)
4330 BTREE UNIQUE added to bulk load code(Vadim)
4331 new lock debug code(Massimo)
4332 massive changes to libpg++(Leo)
4333 new GEQO optimizer speeds table multitable optimization(Martin)
4334 new WARN message for non-unique insert into unique key(Marc)
4335 update x=-3, no spaces, now valid(Bruce)
4336 remove case-sensitive identifier handling(Bruce,Thomas,Dan)
4337 debug backend now pretty-prints tree(Darren)
4338 new Oracle character functions(Edmund)
4339 new plaintext password functions(Dan)
4340 no such class or insufficient privilege changed to distinct messages(Dan)
4341 new ANSI timestamp function(Dan)
4342 new ANSI Time and Date types (Thomas)
4343 move large chunks of data in backend(Martin)
4344 multicolumn btree indexes(Vadim)
4345 new SET var TO value command(Martin)
4346 update transaction status on reads(Dan)
4347 new locale settings for character types(Oleg)
4348 new SEQUENCE serial number generator(Vadim)
4349 GROUP BY function now possible(Vadim)
4350 re-organize regression test(Thomas,Marc)
4351 new optimizer operation weights(Vadim)
4352 new psql \z grant/permit option(Marc)
4353 new MONEY data type(D'Arcy,Thomas)
4354 tcp socket communication speed improved(Vadim)
4355 new VACUUM option for attribute statistics, and for certain columns (Vadim)
4356 many geometric type improvements(Thomas,Keith)
4357 additional regression tests(Thomas)
4358 new datestyle variable(Thomas,Vadim,Martin)
4359 more comparison operators for sorting types(Thomas)
4360 new conversion functions(Thomas)
4361 new more compact btree format(Vadim)
4362 allow pg_dumpall to preserve database ownership(Bruce)
4363 new SET GEQO=# and R_PLANS variable(Vadim)
4364 old (!GEQO) optimizer can use right-sided plans (Vadim)
4365 typechecking improvement in SQL parser(Bruce)
4366 new SET, SHOW, RESET commands(Thomas,Vadim)
4367 new \connect database USER option
4368 new destroydb -i option (Igor)
4369 new \dt and \di psql commands (Darren)
4370 SELECT "\n" now escapes newline (A. Duursma)
4371 new geometry conversion functions from old format (Thomas)
4375 new configuration script(Marc)
4376 readline configuration option added(Marc)
4377 OS-specific configuration options removed(Marc)
4378 new OS-specific template files(Marc)
4379 no more need to edit Makefile.global(Marc)
4380 re-arrange include files(Marc)
4381 nextstep patches (Gregor Hoffleit)
4382 removed Windows-specific code(Bruce)
4383 removed postmaster -e option, now only postgres -e option (Bruce)
4384 merge duplicate library code in front/backends(Martin)
4385 now works with eBones, international Kerberos(Jun)
4386 more shared library support
4387 c++ include file cleanup(Bruce)
4388 warn about buggy flex(Bruce)
4389 DG/UX, Ultrix, IRIX, AIX portability fixes
4390 _________________________________________________________________
4394 Release date: 1997-01-29
4396 A dump/restore is required for those wishing to migrate data from
4397 previous releases of PostgreSQL.
4398 _________________________________________________________________
4400 Migration from version 1.09 to version 6.0
4402 This migration requires a complete dump of the 1.09 database and a
4403 restore of the database in 6.0.
4404 _________________________________________________________________
4406 Migration from pre-1.09 to version 6.0
4408 Those migrating from earlier 1.* releases should first upgrade to 1.09
4409 because the COPY output format was improved from the 1.02 release.
4410 _________________________________________________________________
4416 ALTER TABLE bug - running postgress process needs to re-read table definition
4417 Allow vacuum to be run on one table or entire database(Bruce)
4419 Fix array over-runs of memory writes(Kurt)
4420 Fix elusive btree range/non-range bug(Dan)
4421 Fix for hash indexes on some types like time and date
4422 Fix for pg_log size explosion
4423 Fix permissions on lo_export()(Bruce)
4424 Fix unitialized reads of memory(Kurt)
4425 Fixed ALTER TABLE ... char(3) bug(Bruce)
4426 Fixed a few small memory leaks
4427 Fixed EXPLAIN handling of options and changed full_path option name
4428 Fixed output of group acl privileges
4429 Memory leaks (hunt and destroy with tools like Purify(Kurt)
4430 Minor improvements to rules system
4432 New asserts for run-checking
4433 Overhauled parser/analyze code to properly report errors and increase speed
4434 Pg_dump -d now handles NULL's properly(Bruce)
4435 Prevent SELECT NULL from crashing server (Bruce)
4436 Properly report errors when INSERT ... SELECT columns did not match
4437 Properly report errors when insert column names were not correct
4438 psql \g filename now works(Bruce)
4439 psql fixed problem with multiple statements on one line with multiple outputs
4440 Removed duplicate system OIDs
4441 SELECT * INTO TABLE . GROUP/ORDER BY gives unlink error if table exists(Bruce)
4442 Several fixes for queries that crashed the backend
4443 Starting quote in insert string errors(Bruce)
4444 Submitting an empty query now returns empty status, not just " " query(Bruce)
4448 Add EXPLAIN manual page(Bruce)
4449 Add UNIQUE index capability(Dan)
4450 Add hostname/user level access control rather than just hostname and user
4451 Add synonym of != for <>(Bruce)
4452 Allow "select oid,* from table"
4453 Allow BY,ORDER BY to specify columns by number, or by non-alias table.column(Br
4455 Allow COPY from the frontend(Bryan)
4456 Allow GROUP BY to use alias column name(Bruce)
4457 Allow actual compression, not just reuse on the same page(Vadim)
4458 Allow installation-configuration option to auto-add all local users(Bryan)
4459 Allow libpq to distinguish between text value '' and null(Bruce)
4460 Allow non-postgres users with createdb privs to destroydb's
4461 Allow restriction on who can create C functions(Bryan)
4462 Allow restriction on who can do backend COPY(Bryan)
4463 Can shrink tables, pg_time and pg_log(Vadim & Erich)
4464 Change debug level 2 to print queries only, changed debug heading layout(Bruce)
4465 Change default decimal constant representation from float4 to float8(Bruce)
4466 European date format now set when postmaster is started
4467 Execute lowercase function names if not found with exact case
4468 Fixes for aggregate/GROUP processing, allow 'select sum(func(x),sum(x+y) from z
4470 Gist now included in the distrubution(Marc)
4471 Idend authentication of local users(Bryan)
4472 Implement BETWEEN qualifier(Bruce)
4473 Implement IN qualifier(Bruce)
4474 libpq has PQgetisnull()(Bruce)
4475 libpq++ improvements
4476 New options to initdb(Bryan)
4477 Pg_dump allow dump of OIDs(Bruce)
4478 Pg_dump create indexes after tables are loaded for speed(Bruce)
4479 Pg_dumpall dumps all databases, and the user table
4480 Pginterface additions for NULL values(Bruce)
4481 Prevent postmaster from being run as root
4482 psql \h and \? is now readable(Bruce)
4483 psql allow backslashed, semicolons anywhere on the line(Bruce)
4484 psql changed command prompt for lines in query or in quotes(Bruce)
4485 psql char(3) now displays as (bp)char in \d output(Bruce)
4486 psql return code now more accurate(Bryan?)
4487 psql updated help syntax(Bruce)
4488 Re-visit and fix vacuum(Vadim)
4489 Reduce size of regression diffs, remove timezone name difference(Bruce)
4490 Remove compile-time parameters to enable binary distributions(Bryan)
4491 Reverse meaning of HBA masks(Bryan)
4492 Secure Authentication of local users(Bryan)
4493 Speed up vacuum(Vadim)
4494 Vacuum now had VERBOSE option(Bruce)
4498 All functions now have prototypes that are compared against the calls
4499 Allow asserts to be disabled easly from Makefile.global(Bruce)
4500 Change oid constants used in code to #define names
4501 Decoupled sparc and solaris defines(Kurt)
4502 Gcc -Wall compiles cleanly with warnings only from unfixable constructs
4503 Major include file reorganization/reduction(Marc)
4504 Make now stops on compile failure(Bryan)
4505 Makefile restructuring(Bryan, Marc)
4506 Merge bsdi_2_1 to bsdi(Bruce)
4507 Monitor program removed
4508 Name change from Postgres95 to PostgreSQL
4509 New config.h file(Marc, Bryan)
4510 PG_VERSION now set to 6.0 and used by postmaster
4511 Portability additions, including Ultrix, DG/UX, AIX, and Solaris
4512 Reduced the number of #define's, centeralized #define's
4513 Remove duplicate OIDS in system tables(Dan)
4514 Remove duplicate system catalog info or report mismatches(Dan)
4515 Removed many os-specific #define's
4516 Restructured object file generation/location(Bryan, Marc)
4517 Restructured port-specific file locations(Bryan, Marc)
4518 Unused/uninialized variables corrected
4519 _________________________________________________________________
4523 Release date: 1996-11-04
4525 Sorry, we didn't keep track of changes from 1.02 to 1.09. Some of the
4526 changes listed in 6.0 were actually included in the 1.02.1 to 1.09
4528 _________________________________________________________________
4532 Release date: 1996-08-01
4533 _________________________________________________________________
4535 Migration from version 1.02 to version 1.02.1
4537 Here is a new migration file for 1.02.1. It includes the 'copy' change
4538 and a script to convert old ASCII files.
4540 Note: The following notes are for the benefit of users who want to
4541 migrate databases from Postgres95 1.01 and 1.02 to Postgres95
4544 If you are starting afresh with Postgres95 1.02.1 and do not need
4545 to migrate old databases, you do not need to read any further.
4547 In order to upgrade older Postgres95 version 1.01 or 1.02 databases to
4548 version 1.02.1, the following steps are required:
4549 1. Start up a new 1.02.1 postmaster
4550 2. Add the new built-in functions and operators of 1.02.1 to 1.01 or
4551 1.02 databases. This is done by running the new 1.02.1 server
4552 against your own 1.01 or 1.02 database and applying the queries
4553 attached at the end of the file. This can be done easily through
4554 "psql". If your 1.01 or 1.02 database is named testdb and you have
4555 cut the commands from the end of this file and saved them in
4557 % psql testdb -f addfunc.sql
4558 Those upgrading 1.02 databases will get a warning when executing
4559 the last two statements in the file because they are already
4560 present in 1.02. This is not a cause for concern.
4561 _________________________________________________________________
4563 Dump/Reload Procedure
4565 If you are trying to reload a pg_dump or text-mode, copy tablename to
4566 stdout generated with a previous version, you will need to run the
4567 attached "sed" script on the ASCII file before loading it into the
4568 database. The old format used '.' as end-of-data, while '\.' is now
4569 the end-of-data marker. Also, empty strings are now loaded in as ''
4570 rather than NULL. See the copy manual page for full details.
4571 sed 's/^\.$/\\./g' <in_file >out_file
4573 If you are loading an older binary copy or non-stdout copy, there is
4574 no end-of-data character, and hence no conversion necessary.
4575 -- following lines added by agc to reflect the case-insensitive
4576 -- regexp searching for varchar (in 1.02), and bpchar (in 1.02.1)
4577 create operator ~* (leftarg = bpchar, rightarg = text, procedure = texticregexe
4579 create operator !~* (leftarg = bpchar, rightarg = text, procedure = texticregex
4581 create operator ~* (leftarg = varchar, rightarg = text, procedure = texticregex
4583 create operator !~* (leftarg = varchar, rightarg = text, procedure = texticrege
4585 _________________________________________________________________
4589 Source code maintenance and development
4590 * worldwide team of volunteers
4591 * the source tree now in CVS at ftp.ki.net
4594 * psql (and underlying libpq library) now has many more options for
4595 formatting output, including HTML
4596 * pg_dump now output the schema and/or the data, with many fixes to
4597 enhance completeness.
4598 * psql used in place of monitor in administration shell scripts.
4599 monitor to be deprecated in next release.
4600 * date/time functions enhanced
4601 * NULL insert/update/comparison fixed/enhanced
4602 * TCL/TK lib and shell fixed to work with both tck7.4/tk4.0 and tcl7.5/tk4.1
4604 Bug Fixes (almost too numerous to mention)
4606 * storage management
4607 * check for NULL pointer before dereferencing
4611 * added SolarisX86 port
4612 * added BSD/OS 2.1 port
4614 _________________________________________________________________
4618 Release date: 1996-02-23
4619 _________________________________________________________________
4621 Migration from version 1.0 to version 1.01
4623 The following notes are for the benefit of users who want to migrate
4624 databases from Postgres95 1.0 to Postgres95 1.01.
4626 If you are starting afresh with Postgres95 1.01 and do not need to
4627 migrate old databases, you do not need to read any further.
4629 In order to Postgres95 version 1.01 with databases created with
4630 Postgres95 version 1.0, the following steps are required:
4631 1. Set the definition of NAMEDATALEN in "src/Makefile.global" to 16
4632 and OIDNAMELEN to 20.
4633 2. Decide whether you want to use Host based authentication.
4634 a. If you do, you must create a file name pg_hba in your
4635 top-level data directory (typically the value of your
4636 $PGDATA). "src/libpq/pg_hba" shows an example syntax.
4637 b. If you do not want host-based authentication, you can comment
4640 in "src/Makefile.global"
4641 Note that host-based authentication is turned on by default,
4642 and if you do not take steps A or B above, the out-of-the-box
4643 1.01 will not allow you to connect to 1.0 databases.
4644 3. Compile and install 1.01, but DO NOT do the "initdb" step.
4645 4. Before doing anything else, terminate your 1.0 postmaster, and
4646 backup your existing $PGDATA directory.
4647 5. Set your PGDATA environment variable to your 1.0 databases, but
4648 set up path up so that 1.01 binaries are being used.
4649 6. Modify the file "$PGDATA/PG_VERSION" from 5.0 to 5.1
4650 7. Start up a new 1.01 postmaster
4651 8. Add the new built-in functions and operators of 1.01 to 1.0
4652 databases. This is done by running the new 1.01 server against
4653 your own 1.0 database and applying the queries attached and saving
4654 in the file 1.0_to_1.01.sql. This can be done easily through
4655 "psql". If your 1.0 database is name testdb:
4656 % psql testdb -f 1.0_to_1.01.sql
4657 and then execute the following commands (cut and paste from here):
4658 -- add builtin functions that are new to 1.01
4660 create function int4eqoid (int4, oid) returns bool as 'foo'
4661 language 'internal';
4662 create function oideqint4 (oid, int4) returns bool as 'foo'
4663 language 'internal';
4664 create function char2icregexeq (char2, text) returns bool as 'foo'
4665 language 'internal';
4666 create function char2icregexne (char2, text) returns bool as 'foo'
4667 language 'internal';
4668 create function char4icregexeq (char4, text) returns bool as 'foo'
4669 language 'internal';
4670 create function char4icregexne (char4, text) returns bool as 'foo'
4671 language 'internal';
4672 create function char8icregexeq (char8, text) returns bool as 'foo'
4673 language 'internal';
4674 create function char8icregexne (char8, text) returns bool as 'foo'
4675 language 'internal';
4676 create function char16icregexeq (char16, text) returns bool as 'foo'
4677 language 'internal';
4678 create function char16icregexne (char16, text) returns bool as 'foo'
4679 language 'internal';
4680 create function texticregexeq (text, text) returns bool as 'foo'
4681 language 'internal';
4682 create function texticregexne (text, text) returns bool as 'foo'
4683 language 'internal';
4685 -- add builtin functions that are new to 1.01
4687 create operator = (leftarg = int4, rightarg = oid, procedure = int4eqoid);
4688 create operator = (leftarg = oid, rightarg = int4, procedure = oideqint4);
4689 create operator ~* (leftarg = char2, rightarg = text, procedure = char2icregexe
4691 create operator !~* (leftarg = char2, rightarg = text, procedure = char2icregex
4693 create operator ~* (leftarg = char4, rightarg = text, procedure = char4icregexe
4695 create operator !~* (leftarg = char4, rightarg = text, procedure = char4icregex
4697 create operator ~* (leftarg = char8, rightarg = text, procedure = char8icregexe
4699 create operator !~* (leftarg = char8, rightarg = text, procedure = char8icregex
4701 create operator ~* (leftarg = char16, rightarg = text, procedure = char16icrege
4703 create operator !~* (leftarg = char16, rightarg = text, procedure = char16icreg
4705 create operator ~* (leftarg = text, rightarg = text, procedure = texticregexeq)
4707 create operator !~* (leftarg = text, rightarg = text, procedure = texticregexne
4709 _________________________________________________________________
4714 * 1.01 is backwards compatible with 1.0 database provided the user
4715 follow the steps outlined in the MIGRATION_from_1.0_to_1.01 file.
4716 If those steps are not taken, 1.01 is not compatible with 1.0 database.
4719 * added PQdisplayTuples() to libpq and changed monitor and psql to use it
4720 * added NeXT port (requires SysVIPC implementation)
4721 * added CAST .. AS ... syntax
4722 * added ASC and DESC key words
4723 * added 'internal' as a possible language for CREATE FUNCTION
4724 internal functions are C functions which have been statically linked
4725 into the postgres backend.
4726 * a new type "name" has been added for system identifiers (table names,
4727 attribute names, etc.) This replaces the old char16 type. The
4728 of name is set by the NAMEDATALEN #define in src/Makefile.global
4729 * a readable reference manual that describes the query language.
4730 * added host-based access control. A configuration file ($PGDATA/pg_hba)
4731 is used to hold the configuration data. If host-based access control
4732 is not desired, comment out HBA=1 in src/Makefile.global.
4733 * changed regex handling to be uniform use of Henry Spencer's regex code
4734 regardless of platform. The regex code is included in the distribution
4735 * added functions and operators for case-insensitive regular expressions.
4736 The operators are ~* and !~*.
4737 * pg_dump uses COPY instead of SELECT loop for better performance
4740 * fixed an optimizer bug that was causing core dumps when
4741 functions calls were used in comparisons in the WHERE clause
4742 * changed all uses of getuid to geteuid so that effective uids are used
4743 * psql now returns non-zero status on errors when using -c
4744 * applied public patches 1-14
4745 _________________________________________________________________
4749 Release date: 1995-09-05
4750 _________________________________________________________________
4755 * The copyright of Postgres 1.0 has been loosened to be freely modifiable
4756 and modifiable for any purpose. Please read the COPYRIGHT file.
4757 Thanks to Professor Michael Stonebraker for making this possible.
4760 * date formats have to be MM-DD-YYYY (or DD-MM-YYYY if you're using
4761 EUROPEAN STYLE). This follows SQL-92 specs.
4762 * "delimiters" is now a key word
4765 * sql LIKE syntax has been added
4766 * copy command now takes an optional USING DELIMITER specification.
4767 delimiters can be any single-character string.
4768 * IRIX 5.3 port has been added.
4769 Thanks to Paul Walmsley and others.
4770 * updated pg_dump to work with new libpq
4771 * \d has been added psql
4772 Thanks to Keith Parks
4773 * regexp performance for architectures that use POSIX regex has been
4774 improved due to caching of precompiled patterns.
4775 Thanks to Alistair Crooks
4776 * a new version of libpq++
4777 Thanks to William Wanders
4780 * arbitrary userids can be specified in the createuser script
4781 * \c to connect to other databases in psql now works.
4782 * bad pg_proc entry for float4inc() is fixed
4783 * users with usecreatedb field set can now create databases without
4784 having to be usesuper
4785 * remove access control entries when the entry no longer has any
4787 * fixed non-portable datetimes implementation
4788 * added kerberos flags to the src/backend/Makefile
4789 * libpq now works with kerberos
4790 * typographic errors in the user manual have been corrected.
4791 * btrees with multiple index never worked, now we tell you they don't
4792 work when you try to use them
4793 _________________________________________________________________
4795 Postgres95 Release 0.03
4797 Release date: 1995-07-21
4798 _________________________________________________________________
4802 Incompatible changes:
4803 * BETA-0.3 IS INCOMPATIBLE WITH DATABASES CREATED WITH PREVIOUS VERSIONS
4804 (due to system catalog changes and indexing structure changes).
4805 * double-quote (") is deprecated as a quoting character for string literals;
4806 you need to convert them to single quotes (').
4807 * name of aggregates (eg. int4sum) are renamed in accordance with the
4808 SQL standard (eg. sum).
4809 * CHANGE ACL syntax is replaced by GRANT/REVOKE syntax.
4810 * float literals (eg. 3.14) are now of type float4 (instead of float8 in
4811 previous releases); you might have to do typecasting if you depend on it
4812 being of type float8. If you neglect to do the typecasting and you assign
4813 a float literal to a field of type float8, you may get incorrect values
4815 * LIBPQ has been totally revamped so that frontend applications
4816 can connect to multiple backends
4817 * the usesysid field in pg_user has been changed from int2 to int4 to
4818 allow wider range of Unix user ids.
4819 * the netbsd/freebsd/bsd o/s ports have been consolidated into a
4820 single BSD44_derived port. (thanks to Alistair Crooks)
4822 SQL standard-compliance (the following details changes that makes postgres95
4823 more compliant to the SQL-92 standard):
4824 * the following SQL types are now built-in: smallint, int(eger), float, real,
4825 char(N), varchar(N), date and time.
4827 The following are aliases to existing postgres types:
4829 integer, int -> int4
4830 float, real -> float4
4831 char(N) and varchar(N) are implemented as truncated text types. In
4832 addition, char(N) does blank-padding.
4833 * single-quote (') is used for quoting string literals; '' (in addition to
4834 \') is supported as means of inserting a single quote in a string
4835 * SQL standard aggregate names (MAX, MIN, AVG, SUM, COUNT) are used
4836 (Also, aggregates can now be overloaded, i.e. you can define your
4837 own MAX aggregate to take in a user-defined type.)
4838 * CHANGE ACL removed. GRANT/REVOKE syntax added.
4839 - Privileges can be given to a group using the "GROUP" key word.
4841 GRANT SELECT ON foobar TO GROUP my_group;
4842 The key word 'PUBLIC' is also supported to mean all users.
4844 Privileges can only be granted or revoked to one user or group
4847 "WITH GRANT OPTION" is not supported. Only class owners can change
4849 - The default access control is to to grant users readonly access.
4850 You must explicitly grant insert/update access to users. To change
4851 this, modify the line in
4852 src/backend/utils/acl.h
4853 that defines ACL_WORLD_DEFAULT
4856 * the bug where aggregates of empty tables were not run has been fixed. Now,
4857 aggregates run on empty tables will return the initial conditions of the
4858 aggregates. Thus, COUNT of an empty table will now properly return 0.
4859 MAX/MIN of an empty table will return a row of value NULL.
4860 * allow the use of \; inside the monitor
4861 * the LISTEN/NOTIFY asynchronous notification mechanism now work
4862 * NOTIFY in rule action bodies now work
4863 * hash indexes work, and access methods in general should perform better.
4864 creation of large btree indexes should be much faster. (thanks to Paul
4867 Other changes and enhancements:
4868 * addition of an EXPLAIN statement used for explaining the query execution
4869 plan (eg. "EXPLAIN SELECT * FROM EMP" prints out the execution plan for
4871 * WARN and NOTICE messages no longer have timestamps on them. To turn on
4872 timestamps of error messages, uncomment the line in
4873 src/backend/utils/elog.h:
4874 /* define ELOG_TIMESTAMPS */
4875 * On an access control violation, the message
4876 "Either no such class or insufficient privilege"
4877 will be given. This is the same message that is returned when
4878 a class is not found. This dissuades non-privileged users from
4879 guessing the existence of privileged classes.
4880 * some additional system catalog changes have been made that are not
4881 visible to the user.
4884 * The -oid option has been added to the "pg_result" tcl command.
4885 pg_result -oid returns oid of the last row inserted. If the
4886 last command was not an INSERT, then pg_result -oid returns "".
4887 * the large object interface is available as pg_lo* tcl commands:
4888 pg_lo_open, pg_lo_close, pg_lo_creat, etc.
4890 Portability enhancements and New Ports:
4891 * flex/lex problems have been cleared up. Now, you should be able to use
4892 flex instead of lex on any platforms. We no longer make assumptions of
4893 what lexer you use based on the platform you use.
4894 * The Linux-ELF port is now supported. Various configuration have been
4895 tested: The following configuration is known to work:
4896 kernel 1.2.10, gcc 2.6.3, libc 4.7.2, flex 2.5.2, bison 1.24
4897 with everything in ELF format,
4900 * ipcclean added to the distribution
4901 ipcclean usually does not need to be run, but if your backend crashes
4902 and leaves shared memory segments hanging around, ipcclean will
4903 clean them up for you.
4906 * the user manual has been revised and libpq documentation added.
4907 _________________________________________________________________
4909 Postgres95 Release 0.02
4911 Release date: 1995-05-25
4912 _________________________________________________________________
4916 Incompatible changes:
4917 * The SQL statement for creating a database is 'CREATE DATABASE' instead
4918 of 'CREATEDB'. Similarly, dropping a database is 'DROP DATABASE' instead
4919 of 'DESTROYDB'. However, the names of the executables 'createdb' and
4920 'destroydb' remain the same.
4923 * pgperl - a Perl (4.036) interface to Postgres95
4924 * pg_dump - a utility for dumping out a postgres database into a
4925 script file containing query commands. The script files are in a ASCII
4926 format and can be used to reconstruct the database, even on other
4927 machines and other architectures. (Also good for converting
4928 a Postgres 4.2 database to Postgres95 database.)
4930 The following ports have been incorporated into postgres95-beta-0.02:
4931 * the NetBSD port by Alistair Crooks
4932 * the AIX port by Mike Tung
4933 * the Windows NT port by Jon Forrest (more stuff but not done yet)
4934 * the Linux ELF port by Brian Gallew
4936 The following bugs have been fixed in postgres95-beta-0.02:
4937 * new lines not escaped in COPY OUT and problem with COPY OUT when first
4939 * cannot type return to use the default user id in createuser
4940 * SELECT DISTINCT on big tables crashes
4941 * Linux installation problems
4942 * monitor doesn't allow use of 'localhost' as PGHOST
4943 * psql core dumps when doing \c or \l
4944 * the "pgtclsh" target missing from src/bin/pgtclsh/Makefile
4945 * libpgtcl has a hard-wired default port number
4946 * SELECT DISTINCT INTO TABLE hangs
4947 * CREATE TYPE doesn't accept 'variable' as the internallength
4948 * wrong result using more than 1 aggregate in a SELECT
4949 _________________________________________________________________
4951 Postgres95 Release 0.01
4953 Release date: 1995-05-01