4 Current maintainer: Bruce Momjian (bruce@momjian.us)
5 Last updated: Wed Aug 9 12:35:39 EDT 2006
7 The most recent version of this document can be viewed at
8 http://www.postgresql.org/docs/faqs.TODO.html.
10 #A hyphen, "-", marks changes that will appear in the upcoming 8.2 release.#
11 #A percent sign, "%", marks items that are easier to implement.#
13 Bracketed items, "[]", have more detail.
15 This list contains all known PostgreSQL bugs and feature requests. If
16 you would like to work on an item, please read the Developer's FAQ
23 * -Make postmater and postgres options distinct so the postmaster -o
24 option is no longer needed
25 * -Allow pooled connections to list all prepared statements
27 This would allow an application inheriting a pooled connection to know
28 the statements prepared in the current session.
30 * Allow major upgrades without dump/reload, perhaps using pg_upgrade
32 * Check for unreferenced table files created by transactions that were
33 in-progress when the server terminated abruptly
35 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
37 * Allow administrators to safely terminate individual sessions either
38 via an SQL function or SIGTERM
40 Lock table corruption following SIGTERM of an individual backend
41 has been reported in 8.0. A possible cause was fixed in 8.1, but
42 it is unknown whether other problems exist. This item mostly
43 requires additional testing rather than of writing any new code.
44 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00174.php
46 * %Set proper permissions on non-system schemas during db creation
48 Currently all schemas are owned by the super-user because they are
49 copied from the template1 database.
51 * Support table partitioning that allows a single table to be stored
52 in subtables that are partitioned based on the primary key or a WHERE
54 * Add function to report the time of the most recent server reload
55 * Allow statistics collector information to be pulled from the collector
56 process directly, rather than requiring the collector to write a
57 filesystem file twice a second?
58 * Allow log_min_messages to be specified on a per-module basis
60 This would allow administrators to see more detailed information from
61 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
62 Another idea is to allow separate configuration files for each module,
63 or allow arbitrary SET commands to be passed to them.
65 * -Re-enable the GUC full_page_writes in 8.2 when reliability issues have
69 * Improve replication solutions
73 You can use any of the master/slave replication servers to use a
74 standby server for data warehousing. To allow read/write queries to
75 multiple servers, you need multi-master replication like pgcluster.
77 o Allow replication over unreliable or non-persistent links
82 o -Add "include file" functionality in postgresql.conf
83 o %Allow commenting of variables in postgresql.conf to restore them
86 Currently, if a variable is commented out, it keeps the
87 previous uncommented value until a server restarted.
89 o -Allow per-database permissions to be set via GRANT
91 Allow database connection checks based on GRANT rules in
92 addition to the existing access checks in pg_hba.conf.
94 o Allow pg_hba.conf to specify host names along with IP addresses
96 Host name lookup could occur when the postmaster reads the
97 pg_hba.conf file, or when the backend starts. Another
98 solution would be to reverse lookup the connection IP and
99 check that hostname against the host names in pg_hba.conf.
100 We could also then check that the host name maps to the IP
103 o %Allow postgresql.conf file values to be changed via an SQL
104 API, perhaps using SET GLOBAL
105 o Allow the server to be stopped/restarted via an SQL API
106 o -Issue a warning if a change-on-restart-only postgresql.conf value
107 is modified and the server config files are reloaded
108 o Mark change-on-restart-only values in postgresql.conf
113 o Allow a database in tablespace t1 with tables created in
114 tablespace t2 to be used as a template for a new database created
115 with default tablespace t2
117 All objects in the default database tablespace must have default
118 tablespace specifications. This is because new databases are
119 created by copying directories. If you mix default tablespace
120 tables and tablespace-specified tables in the same directory,
121 creating a new database from such a mixed directory would create a
122 new database with tables that had incorrect explicit tablespaces.
123 To fix this would require modifying pg_class in the newly copied
124 database, which we don't currently do.
126 o Allow reporting of which objects are in which tablespaces
128 This item is difficult because a tablespace can contain objects
129 from multiple databases. There is a server-side function that
130 returns the databases which use a specific tablespace, so this
131 requires a tool that will call that function and connect to each
132 database to find the objects in each database for that tablespace.
134 o %Add a GUC variable to control the tablespace for temporary objects
137 It could start with a random tablespace from a supplied list and
138 cycle through the list.
140 o Allow WAL replay of CREATE TABLESPACE to work when the directory
141 structure on the recovery computer is different from the original
143 o Allow per-tablespace quotas
146 * Point-In-Time Recovery (PITR)
148 o Add command to archive partially filled write-ahead logs? [pitr]
150 Currently only full WAL files are archived. This means that the
151 most recent transactions aren't available for recovery in case
154 o -Automatically force archiving of partially-filled WAL files when
155 pg_stop_backup() is called or the server is stopped
156 o -Add reporting of the current WAL file and offset, perhaps as
157 part of partial log file archiving
158 o %Create dump tool for write-ahead logs for use in determining
159 transaction id for point-in-time recovery
160 o Allow a warm standby system to also allow read-only statements
163 This is useful for checking PITR recovery.
165 o Allow the PITR process to be debugged and data examined
171 * Allow server log information to be output as INSERT statements
173 This would allow server log information to be easily loaded into
174 a database for analysis.
176 * %Add ability to monitor the use of temporary sort files
177 * -Allow server logs to be remotely read and removed using SQL commands
178 * -Allow protocol-level BIND parameter values to be logged, if text mode
184 * Improve the MONEY data type
186 Change the MONEY data type to use DECIMAL internally, with special
187 locale-aware output formatting.
188 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
190 * Change NUMERIC to enforce the maximum precision
191 * Add NUMERIC division operator that doesn't round?
193 Currently NUMERIC _rounds_ the result to the specified precision.
194 This means division can return a result that multiplied by the
195 divisor is greater than the dividend, e.g. this returns a value > 10:
197 SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
199 The positive modulus result returned by NUMERICs might be considered
200 inaccurate, in one sense.
202 * %Disallow changing DEFAULT expression of a SERIAL column?
204 This should be done only if the existing SERIAL problems cannot be
207 * %Disallow ALTER SEQUENCE changes for SERIAL sequences because pg_dump
208 does not dump the changes
209 * Fix data types where equality comparison isn't intuitive, e.g. box
210 * -Zero umasked bits in conversion from INET cast to CIDR
211 * -Prevent INET cast to CIDR from dropping netmask, SELECT '1.1.1.1'::inet::cidr
212 * -Allow INET + INT8 to increment the host part of the address or
213 throw an error on overflow
214 * -Add 'tid != tid ' operator for use in corruption recovery
215 * Allow user-defined types to specify a type modifier at table creation
217 * Allow user-defined types to accept 'typmod' parameters
219 http://archives.postgresql.org/pgsql-hackers/2005-08/msg01142.php
220 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00012.php
222 * Add support for public SYNONYMs
224 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
226 * Fix CREATE CAST on DOMAINs
228 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
230 * Add Globally/Universally Unique Identifier (GUID/UUID)
231 * Add support for SQL-standard GENERATED/IDENTITY columns
233 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
238 o Allow infinite dates and intervals just like infinite timestamps
239 o Merge hardwired timezone names with the TZ database; allow either
240 kind everywhere a TZ name is currently taken
241 o -Allow customization of the known set of TZ names (generalize the
242 present australian_timezones hack)
243 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
244 information, either zone name or offset from UTC [timezone]
246 If the TIMESTAMP value is stored with a time zone name, interval
247 computations should adjust based on the time zone rules.
249 o -Allow timezone names in SQL strings, '2006-05-24 21:11
250 Americas/New_York'::timestamptz
251 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
252 o Add a GUC variable to allow output of interval values in ISO8601
254 o Improve timestamptz subtraction to be DST-aware
256 Currently, subtracting one date from another that crosses a
257 daylight savings time adjustment can return '1 day 1 hour', but
258 adding that back to the first date returns a time one hour in
259 the future. This is caused by the adjustment of '25 hours' to
260 '1 day 1 hour', and '1 day' is the same time the next day, even
261 if daylight savings adjustments are involved.
263 o Fix interval display to support values exceeding 2^31 hours
264 o Add overflow checking to timestamp and interval arithmetic
265 o Add ISO INTERVAL handling
267 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
268 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
270 o -Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
272 o Support ISO INTERVAL syntax if units cannot be determined from
273 the string, and are supplied after the string
275 The SQL standard states that the units after the string
276 specify the units of the string, e.g. INTERVAL '2' MINUTE
277 should return '00:02:00'. The current behavior has the units
278 restrict the interval value to the specified unit or unit
279 range, INTERVAL '70' SECOND returns '00:00:10'.
281 For syntax that isn't uniquely ISO or PG syntax, like '1' or
282 '1:30', treat as ISO if there is a range specification clause,
283 and as PG if there no clause is present, e.g. interpret '1:30'
284 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
285 '1:30' as '1 hour, 30 minutes'.
287 This makes common cases like SELECT INTERVAL '1' MONTH
288 SQL-standard results. The SQL standard supports a limited
289 number of unit combinations and doesn't support unit names in
290 the string. The PostgreSQL syntax is more flexible in the
291 range of units supported, e.g. PostgreSQL supports '1 year 1
292 hour', while the SQL standard does not.
294 o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
295 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
296 INTERVAL MONTH), and this should return '12 months'
297 o Round or truncate values to the requested precision, e.g.
298 INTERVAL '11 months' AS YEAR should return one or zero
299 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
304 o -Allow NULLs in arrays
305 o Delay resolution of array expression's data type so assignment
306 coercion can be performed on empty array expressions
311 o Improve vacuum of large objects, like /contrib/vacuumlo?
312 o Add security checking for large objects
313 o Auto-delete large objects when referencing row is deleted
315 /contrib/lo offers this functionality.
317 o Allow read/write into TOAST values like large objects
319 This requires the TOAST column to be stored EXTERNAL.
321 o Add API for 64-bit large object access
323 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
329 * Allow INET subnet tests using non-constants to be indexed
330 * -Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
333 Current CURRENT_TIMESTAMP returns the start time of the current
334 transaction, and gettimeofday() returns the wallclock time. This will
335 make time reporting more consistent and will allow reporting of
336 the statement start time.
338 * %Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
339 pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
340 * -Allow to_char() to print localized month names
341 * Allow to_date() and to_timestamp() accept localized month names
342 * Add missing parameter handling in to_char()
344 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
346 * Allow functions to have a schema search path specified at creation time
347 * Allow substring/replace() to get/set bit values
348 * Allow to_char() on interval values to accumulate the highest unit
351 Some special format flag would be required to request such
352 accumulation. Such functionality could also be added to EXTRACT.
353 Prevent accumulation that crosses the month/day boundary because of
354 the uneven number of days in a month.
356 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
357 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
358 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
359 o to_char(INTERVAL '3 years 5 months','MM') => 41
361 * -Add sleep() function, remove from regress.c
362 * -Allow user-defined functions retuning a domain value to enforce domain
364 * Add SPI_gettypmod() to return the typemod for a TupleDesc
367 Multi-Language Support
368 ======================
370 * Add NCHAR (as distinguished from ordinary varchar),
371 * Allow locale to be set at database creation
373 Currently locale can only be set during initdb. No global tables have
374 locale-aware columns. However, the database template used during
375 database creation might have locale-aware indexes. The indexes would
376 need to be reindexed to match the new locale.
378 * Allow encoding on a per-column basis
380 Right now only one encoding is allowed per database. [locale]
381 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
383 * Add CREATE COLLATE? [locale]
384 * Support multiple simultaneous character sets, per SQL92
385 * Improve UTF8 combined character handling?
386 * Add octet_length_server() and octet_length_client()
387 * Make octet_length_client() the same as octet_length()?
388 * Fix problems with wrong runtime encoding conversion for NLS message files
389 * Add URL to more complete multi-byte regression tests
391 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
393 * Fix ILIKE and regular expressions to handle case insensitivity
394 properly in multibyte encodings
396 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
397 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
403 * Automatically create rules on views so they are updateable, per SQL99
405 We can only auto-create rules for simple views. For more complex
406 cases users will still have to write rules manually.
407 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
409 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
410 * Allow NOTIFY in rules involving conditionals
411 * Allow VIEW/RULE recompilation when the underlying tables change
413 Another issue is whether underlying table changes should be reflected
414 in the view, e.g. should SELECT * show additional columns if they
415 are added after the view is created.
421 * -Change LIMIT/OFFSET and FETCH/MOVE to use int8
422 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
423 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
424 * %Allow SET CONSTRAINTS to be qualified by schema/table name
425 * -Allow TRUNCATE ... CASCADE/RESTRICT
427 This is like DELETE CASCADE, but truncates.
429 * %Add a separate TRUNCATE permission
431 Currently only the owner can TRUNCATE a table because triggers are not
432 called, and the table is locked in exclusive mode.
434 * Allow PREPARE of cursors
435 * -Allow PREPARE to automatically determine parameter types based on the SQL
437 * Allow finer control over the caching of prepared query plans
439 Currently, queries prepared via the libpq API are planned on first
440 execute using the supplied parameters --- allow SQL PREPARE to do the
441 same. Also, allow control over replanning prepared queries either
442 manually or automatically when statistics for execute parameters
443 differ dramatically from those used during planning.
445 * Invalidate prepared queries, like INSERT, when the table definition
447 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
449 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
450 such information in memory would improve performance.
452 * Add optional textual message to NOTIFY
454 This would allow an informational message to be added to the notify
455 message, perhaps indicating the row modified or other custom
458 * Add a GUC variable to warn about non-standard SQL usage in queries
459 * Add SQL-standard MERGE command, typically used to merge two tables
462 This is similar to UPDATE, then for unmatched rows, INSERT.
463 Whether concurrent access allows modifications which could cause
464 row loss is implementation independent.
466 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
469 To implement this cleanly requires that the table have a unique index
470 so duplicate checking can be easily performed. It is possible to
471 do it without a unique index if we require the user to LOCK the table
474 * Add NOVICE output level for helpful messages like automatic sequence/index
476 * -Add COMMENT ON for all cluster global objects (roles, databases
478 * -Make row-wise comparisons work per SQL spec
480 Right now, '(a, b) < (1, 2)' is processed as 'a < 1 and b < 2', but
481 the SQL standard requires it to be processed as a column-by-column
482 comparison, so the proper comparison is '(a < 1) OR (a = 1 AND b < 2)'.
484 * Add RESET CONNECTION command to reset all session state
486 This would include resetting of all variables (RESET ALL), dropping of
487 temporary tables, removing any NOTIFYs, cursors, open transactions,
488 prepared queries, currval()s, etc. This could be used for connection
489 pooling. We could also change RESET ALL to have this functionality.
490 The difficult of this features is allowing RESET ALL to not affect
491 changes made by the interface driver for its internal use. One idea
492 is for this to be a protocol-only feature. Another approach is to
493 notify the protocol when a RESET CONNECTION command is used.
494 http://archives.postgresql.org/pgsql-patches/2006-04/msg00192.php
496 * Add GUC to issue notice about statements that use unjoined tables
497 * Allow EXPLAIN to identify tables that were skipped because of
499 * Allow EXPLAIN output to be more easily processed by scripts
500 * -Enable escape_string_warning and standard_conforming_strings
501 * Make standard_conforming_strings the default in 8.3?
503 When this is done, backslash-quote should be prohibited in non-E''
504 strings because of possible confusion over how such strings treat
505 backslashes. Basically, '' is always safe for a literal single
506 quote, while \' might or might not be based on the backslash
509 * Simplify dropping roles that have objects in several databases
510 * Allow COMMENT ON to accept an expression rather than just a string
511 * Allow the count returned by SELECT, etc to be to represent as an int64
512 to allow a higher range of values
513 * Make CLUSTER preserve recently-dead tuples per MVCC requirements
514 * Add SQL99 WITH clause to SELECT
515 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
516 * Add DEFAULT .. AS OWNER so permission checks are done as the table
519 This would be useful for SERIAL nextval() calls and CHECK constraints.
521 * Add a GUC to control whether BEGIN inside a transcation should abort
523 * Allow DISTINCT to work in multiple-argument aggregate calls
524 * Add column to pg_stat_activity that shows the progress of long-running
525 commands like CREATE INDEX and VACUUM
530 o Allow CREATE TABLE AS to determine column lengths for complex
531 expressions like SELECT col1 || col2
532 o Use more reliable method for CREATE DATABASE to get a consistent
534 o -Add ON COMMIT capability to CREATE TABLE AS ... SELECT
538 o -Allow an alias to be provided for the target table in
540 o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
543 A subselect can also be used as the value source.
544 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
549 o %Have ALTER TABLE RENAME rename SERIAL sequence names
550 o Add ALTER DOMAIN to modify the underlying data type
551 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
553 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
555 o %Allow ALTER TABLE to change constraint deferrability and actions
556 o Add missing object types for ALTER ... SET SCHEMA
557 o Allow ALTER TABLESPACE to move to different directories
558 o Allow databases to be moved to different tablespaces
559 o Allow moving system tables to other tablespaces, where possible
561 Currently non-global system tables must be in the default database
562 tablespace. Global system tables can never be moved.
564 o Prevent parent tables from altering or dropping constraints
565 like CHECK that are inherited by child tables unless CASCADE
567 o %Prevent child tables from altering or dropping constraints
568 like CHECK that were inherited from the parent table
569 o -Add ALTER TABLE tab INHERIT / NO INHERIT parent
570 o Have ALTER INDEX update the name of a constraint using that index
571 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
576 o Automatically maintain clustering on a table
578 This might require some background daemon to maintain clustering
579 during periods of low usage. It might also require tables to be only
580 partially filled for easier reorganization. Another idea would
581 be to create a merged heap/index data file so an index lookup would
582 automatically access the heap data too. A third idea would be to
583 store heap rows in hashed groups, perhaps using a user-supplied
586 o %Add default clustering to system tables
588 To do this, determine the ideal cluster index for each system
589 table and set the cluster setting during initdb.
594 o Allow COPY to report error lines and continue
596 This requires the use of a savepoint before each COPY line is
597 processed, with ROLLBACK on COPY failure.
599 o -Have COPY return the number of rows loaded/unloaded?
600 o Allow COPY on a newly-created table to skip WAL logging
602 On crash recovery, the table involved in the COPY would
603 be removed or have its heap and index files truncated. One
604 issue is that no other backend should be able to add to
605 the table at the same time, which is something that is
608 o Allow COPY to output from SELECT
610 COPY should also be able to output views.
611 http://archives.postgresql.org/pgsql-patches/2005-09/msg00148.php
616 o Allow column-level privileges
617 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
620 The proposed syntax is:
621 GRANT SELECT ON ALL TABLES IN public TO phpuser;
622 GRANT SELECT ON NEW TABLES IN public TO phpuser;
624 o Allow GRANT/REVOKE permissions to be inherited by objects based on
627 o Allow SERIAL sequences to inherit permissions from the base table?
632 o Allow UPDATE/DELETE WHERE CURRENT OF cursor
634 This requires using the row ctid to map cursor rows back to the
635 original heap row. This become more complicated if WITH HOLD cursors
636 are to be supported because WITH HOLD cursors have a copy of the row
637 and no FOR UPDATE lock.
639 o Prevent DROP TABLE from dropping a row referenced by its own open
642 o -Allow pooled connections to list all open WITH HOLD cursors
644 Because WITH HOLD cursors exist outside transactions, this allows
645 them to be listed so they can be closed.
650 o Allow INSERT/UPDATE of the system-generated oid value for a row
651 o -Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
652 o Allow INSERT/UPDATE ... RETURNING new.col or old.col
654 This is useful for returning the auto-generated key for an INSERT.
655 One complication is how to handle rules that run as part of
657 http://archives.postgresql.org/pgsql-patches/2005-07/msg00568.php
658 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
664 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
666 o Add SET PATH for schemas?
668 This is basically the same as SET search_path.
671 * Referential Integrity
673 o Add MATCH PARTIAL referential integrity
674 o Change foreign key constraint for array -> element to mean element
676 o Enforce referential integrity for system tables
677 o Fix problem when cascading referential triggers make changes on
678 cascaded tables, seeing the tables in an intermediate state
680 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
681 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
684 * Server-Side Languages
687 o Fix RENAME to work on variables other than OLD/NEW
688 o Allow function parameters to be passed by name,
689 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
690 o Add Oracle-style packages (Pavel)
692 A package would be a schema with session-local variables,
693 public/private functions, and initialization functions. It
694 is also possible to implement these capabilities
695 in all schemas and not use a separate "packages"
697 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
699 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
700 o Allow listing of record column names, and access to
701 record columns via variables, e.g. columns := r.(*),
704 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
705 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
706 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
709 o Add single-step debugging of functions
710 o Add support for WITH HOLD and SCROLL cursors
712 PL/pgSQL cursors should support the same syntax as
715 o Allow PL/RETURN to return row or record functions
717 http://archives.postgresql.org/pgsql-patches/2005-11/msg00045.php
719 o Fix memory leak from exceptions
721 http://archives.postgresql.org/pgsql-performance/2006-06/msg00305.php
723 o Fix problems with RETURN NEXT on tables with
724 dropped/added columns after function creation
726 http://archives.postgresql.org/pgsql-patches/2006-02/msg00165.php
729 o Add table function support to pltcl, plpython
730 o Add support for polymorphic arguments and return types to
731 languages other than PL/PgSQL
732 o Add capability to create and call PROCEDURES
733 o Add support for OUT and INOUT parameters to languages other
735 o Add PL/Python tracebacks
737 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
739 o Allow PL/python to composite types and result sets
740 once buggy assert-enabled versions of python can be detected
742 http://archives.postgresql.org/pgsql-patches/2006-04/msg00087.php
747 * -Have initdb set the input DateStyle (MDY or DMY) based on locale
748 * Have pg_ctl look at PGHOST in case it is a socket directory?
749 * Allow pg_ctl to work properly with configuration files located outside
752 pg_ctl can not read the pid file because it isn't located in the
753 config directory but in the PGDATA directory. The solution is to
754 allow pg_ctl to read and understand postgresql.conf to find the
755 data_directory value.
760 o Have psql show current values for a sequence
761 o Move psql backslash database information into the backend, use
762 mnemonic commands? [psql]
764 This would allow non-psql clients to pull the same information out
765 of the database as psql.
767 o Fix psql's \d commands more consistent
769 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
770 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
772 o Allow psql \pset boolean variables to set to fixed values, rather
774 o Consistently display privilege information for all objects in psql
775 o -Improve psql's handling of multi-line statements
777 Currently, while \e saves a single statement as one entry, interactive
778 statements are saved one line at a time. Ideally all statements
779 would be saved like \e does.
781 o -Allow multi-line column values to align in the proper columns
783 If the second output column value is 'a\nb', the 'b' should appear
784 in the second display column, rather than the first column as it
787 o -Display IN, INOUT, and OUT parameters in \df
788 o Add auto-expanded mode so expanded output is used if the row
789 length is wider than the screen width.
791 Consider using auto-expanded mode for backslash commands like \df+.
793 o Prevent tab completion of SET TRANSACTION from querying the
794 database and therefore preventing the transaction isolation
795 level from being set.
797 Currently, SET <tab> causes a database lookup to check all
798 supported session variables. This query causes problems
799 because setting the transaction isolation level must be the
800 first statement of a transaction.
805 o -Allow pg_dump to use multiple -t and -n switches, exclusion
806 ability, and regular expression object matching
807 o %Add dumping of comments on index columns and composite type columns
808 o %Add full object name to the tag field. eg. for operators we need
809 '=(integer, integer)', instead of just '='.
810 o Add pg_dumpall custom format dumps?
811 o -Update pg_dump and psql to use the new COPY libpq API (Christopher)
812 o Remove unnecessary function pointer abstractions in pg_dump source
814 o Allow selection of individual object(s) of all types, not just
816 o In a selective dump, allow dumping of an object and all its
818 o Add options like pg_restore -l and -L to pg_dump
819 o Stop dumping CASCADE on DROP TYPE commands in clean mode
820 o Allow pg_dump --clean to drop roles that own objects or have
822 o Add -f to pg_dumpall
829 Document differences between ecpg and the SQL standard and
830 information about the Informix-compatibility module.
832 o Solve cardinality > 1 for input descriptors / variables?
833 o Add a semantic check level, e.g. check if a table really exists
834 o fix handling of DB attributes that are arrays
835 o Use backend PREPARE/EXECUTE facility for ecpg where possible
837 o Fix nested C comments
838 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
839 o Make SET CONNECTION thread-aware, non-standard?
840 o Allow multidimensional arrays
841 o Add internationalized message strings
842 o Implement COPY FROM STDIN
847 o Add a function to support Parse/DescribeStatement capability
848 o Add PQescapeIdentifierConn()
849 o Prevent PQfnumber() from lowercasing unquoted the column name
851 PQfnumber() should never have been doing lowercasing, but
852 historically it has so we need a way to prevent it
854 o Allow statement results to be automatically batched to the client
856 Currently, all statement results are transferred to the libpq
857 client before libpq makes the results available to the
858 application. This feature would allow the application to make
859 use of the first result rows while the rest are transferred, or
860 held on the server waiting for them to be requested by libpq.
861 One complexity is that a statement like SELECT 1/col could error
862 out mid-way through the result set.
867 * Add deferred trigger queue file
869 Right now all deferred trigger information is stored in backend
870 memory. This could exhaust memory for very large trigger queues.
871 This item involves dumping large queues into files.
873 * Allow DEFERRABLE UNIQUE constraints?
874 * Allow triggers to be disabled in only the current session.
876 This is currently possible by starting a multi-statement transaction,
877 modifying the system tables, performing the desired SQL, restoring the
878 system tables, and committing the transaction. ALTER TABLE ...
879 TRIGGER requires a table lock so it is not ideal for this usage.
881 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
883 If the dump is known to be valid, allow foreign keys to be added
884 without revalidating the data.
886 * Allow statement-level triggers to access modified rows
887 * Support triggers on columns
889 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
891 * Allow AFTER triggers on system tables
893 System tables are modified in many places in the backend without going
894 through the executor and therefore not causing triggers to fire. To
895 complete this item, the functions that modify system tables will have
902 * Flush cached query plans when the dependent objects change,
903 when the cardinality of parameters changes dramatically, or
904 when new ANALYZE statistics are available
906 A more complex solution would be to save multiple plans for different
907 cardinality and use the appropriate plan based on the EXECUTE values.
909 * Track dependencies in function bodies and recompile/invalidate
911 This is particularly important for references to temporary tables
912 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
913 in PL/PgSQL is to use EXECUTE. One complexity is that a function
914 might itself drop and recreate dependent tables, causing it to
915 invalidate its own query plan.
921 * Add pre-parsing phase that converts non-ISO syntax to supported
924 This could allow SQL written for other databases to run without
927 * Allow plug-in modules to emulate features from other databases
928 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
930 * Allow statements across databases or servers with transaction
933 This can be done using dblink and two-phase commit.
935 * Add the features of packages
937 o Make private objects accessible only to objects in the same schema
938 o Allow current_schema.objname to access current schema objects
939 o Add session variables
940 o Allow nested schemas
942 * Consider allowing control of upper/lower case folding of unquoted
945 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
950 * Allow inherited tables to inherit index, UNIQUE constraint, and primary
952 * UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
953 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
956 The main difficulty with this item is the problem of creating an index
957 that can span more than one table.
959 * Allow SELECT ... FOR UPDATE on inherited tables
960 * Add UNIQUE capability to non-btree indexes
961 * Prevent index uniqueness checks when UPDATE does not modify the column
963 Uniqueness (index) checks are done when updating a column even if the
964 column is not modified by the UPDATE.
966 * Allow the creation of on-disk bitmap indexes which can be quickly
967 combined with other bitmap indexes
969 Such indexes could be more compact if there are only a few distinct values.
970 Such indexes can also be compressed. Keeping such indexes updated can be
972 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
974 * Allow use of indexes to search for NULLs
976 One solution is to create a partial index on an IS NULL expression.
978 * Allow accurate statistics to be collected on indexes with more than
979 one column or expression indexes, perhaps using per-index statistics
980 * -Add fillfactor to control reserved free space during index creation
981 * Allow the creation of indexes with mixed ascending/descending
984 This is possible now by creating an operator class with reversed sort
985 operators. One complexity is that NULLs would then appear at the start
986 of the result set, and this might affect certain sort types, like
989 * Allow constraint_exclusion to work for UNIONs like it does for
990 inheritance, allow it to work for UPDATE and DELETE statements, and allow
991 it to be used for all statements with little performance impact
992 * Allow CREATE INDEX to take an additional parameter for use with
994 * Consider compressing indexes by storing key values duplicated in
995 several rows as a single index entry
997 This is difficult because it requires datatype-specific knowledge.
1002 o Add more GIST index support for geometric data types
1003 o Allow GIST indexes to create certain complex index types, like
1004 digital trees (see Aoki)
1008 o Pack hash index buckets onto disk pages more efficiently
1010 Currently only one hash bucket can be stored on a page. Ideally
1011 several hash buckets could be stored on a single page and greater
1012 granularity used for the hash algorithm.
1014 o Consider sorting hash buckets so entries can be found using a
1015 binary search, rather than a linear scan
1017 o In hash indexes, consider storing the hash value with or instead
1020 o Add WAL logging for crash recovery
1021 o Allow multi-column hash indexes
1027 * Improve commit_delay handling to reduce fsync()
1028 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1030 Ideally this requires a separate test program that can be run
1031 at initdb time or optionally later. Consider O_SYNC when
1034 * %Add an option to sync() before fsync()'ing checkpoint files
1035 * Add program to test if fsync has a delay compared to non-fsync
1041 * Allow free-behind capability for large sequential scans, perhaps using
1044 Posix_fadvise() can control both sequential/random file caching and
1045 free-behind behavior, but it is unclear how the setting affects other
1046 backends that also have the file open, and the feature is not supported
1047 on all operating systems.
1051 We could use a fixed row count and a +/- count to follow MVCC
1052 visibility rules, or a single cached value could be used and
1053 invalidated if anyone modifies the table. Another idea is to
1054 get a count directly from a unique index, but for this to be
1055 faster than a sequential scan it must avoid access to the heap
1056 to obtain tuple visibility information.
1058 * Add estimated_count(*) to return an estimate of COUNT(*)
1060 This would use the planner ANALYZE statistics to return an estimated
1063 * Allow data to be pulled directly from indexes
1065 Currently indexes do not have enough tuple visibility information
1066 to allow data to be pulled from the index without also accessing
1067 the heap. One way to allow this is to set a bit on index tuples
1068 to indicate if a tuple is currently visible to all transactions
1069 when the first valid heap lookup happens. This bit would have to
1070 be cleared when a heap tuple is expired.
1072 Another idea is to maintain a bitmap of heap pages where all rows
1073 are visible to all backends, and allow index lookups to reference
1074 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1075 add someday to determine which heap pages need vacuuming. Frequently
1076 accessed bitmaps would have to be stored in shared memory. One 8k
1077 page of bitmaps could track 512MB of heap pages.
1079 * Consider automatic caching of statements at various levels:
1082 o Query execute plan
1085 * Allow sequential scans to take advantage of other concurrent
1086 sequential scans, also called "Synchronised Scanning"
1088 One possible implementation is to start sequential scans from the lowest
1089 numbered buffer in the shared cache, and when reaching the end wrap
1090 around to the beginning, rather than always starting sequential scans
1091 at the start of the table.
1093 * Consider increasing internal areas when shared buffers is increased
1095 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1101 * Improve speed with indexes
1103 For large table adjustments during VACUUM FULL, it is faster to
1104 reindex rather than update the index.
1106 * Reduce lock time during VACUUM FULL by moving tuples with read lock,
1107 then write lock and truncate table
1109 Moved tuples are invisible to other backends so they don't require a
1110 write lock. However, the read lock promotion to write lock could lead
1111 to deadlock situations.
1113 * Auto-fill the free space map by scanning the buffer cache or by
1114 checking pages written by the background writer
1116 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1117 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1119 * Create a bitmap of pages that need vacuuming
1121 Instead of sequentially scanning the entire table, have the background
1122 writer or some other process record pages that have expired rows, then
1123 VACUUM can look at just those pages rather than the entire table. In
1124 the event of a system crash, the bitmap would probably be invalidated.
1125 One complexity is that index entries still have to be vacuumed, and
1126 doing this without an index scan (by using the heap values to find the
1127 index entry) might be slow and unreliable, especially for user-defined
1130 * -Add system view to show free space map contents
1131 * Allow FSM to return free space toward the beginning of the heap file,
1132 in hopes that empty pages at the end can be truncated by VACUUM
1133 * Allow FSM page return free space based on table clustering, to assist
1134 in maintaining clustering?
1135 * Consider shrinking expired tuples to just their headers
1137 http://archives.postgresql.org/pgsql-patches/2006-03/msg00142.php
1139 * Allow heap reuse of UPDATEd rows if no indexed columns are changed,
1140 and old and new versions are on the same heap page?
1142 While vacuum handles DELETEs fine, updating of non-indexed columns, like
1143 counters, are difficult for VACUUM to handle efficiently. This method
1144 is possible for same-page updates because a single index row can be
1145 used to point to both old and new values.
1146 http://archives.postgresql.org/pgsql-hackers/2006-06/msg01305.php
1147 http://archives.postgresql.org/pgsql-hackers/2006-06/msg01534.php
1149 * Reuse index tuples that point to heap tuples that are not visible to
1153 o Use free-space map information to guide refilling
1154 o %Issue log message to suggest VACUUM FULL if a table is nearly
1156 o Improve xid wraparound detection by recording per-table rather
1158 o Consider logging activity either to the logs or a system view
1164 * Fix priority ordering of read and write light-weight locks (Neil)
1167 Startup Time Improvements
1168 =========================
1170 * Experiment with multi-threaded backend for backend creation [thread]
1172 This would prevent the overhead associated with process creation. Most
1173 operating systems have trivial process creation time compared to
1174 database startup overhead, but a few operating systems (Win32,
1175 Solaris) might benefit from threading. Also explore the idea of
1176 a single session using multiple threads to execute a statement faster.
1178 * Experiment with multi-threaded backend better resource utilization
1180 This would allow a single query to make use of multiple CPU's or
1181 multiple I/O channels simultaneously. One idea is to create a
1182 background reader that can pre-fetch sequential and index scan
1183 pages needed by other backends. This could be expanded to allow
1184 concurrent reads from multiple devices in a partitioned table.
1186 * Add connection pooling
1188 It is unclear if this should be done inside the backend code or done
1189 by something external like pgpool. The passing of file descriptors to
1190 existing backends is one of the difficulties with a backend approach.
1196 * Eliminate need to write full pages to WAL before page modification [wal]
1198 Currently, to protect against partial disk page writes, we write
1199 full page images to WAL before they are modified so we can correct any
1200 partial page writes during recovery. These pages can also be
1201 eliminated from point-in-time archive files.
1203 o When off, write CRC to WAL and check file system blocks
1206 If CRC check fails during recovery, remember the page in case
1207 a later CRC for that page properly matches.
1209 o Write full pages during file system write and not when
1210 the page is modified in the buffer cache
1212 This allows most full page writes to happen in the background
1213 writer. It might cause problems for applying WAL on recovery
1214 into a partially-written page, but later the full page will be
1217 * Allow WAL traffic to be streamed to another server for stand-by
1219 * Reduce WAL traffic so only modified values are written rather than
1221 * Allow the pg_xlog directory location to be specified during initdb
1222 with a symlink back to the /data location
1223 * Allow WAL information to recover corrupted pg_controldata
1225 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1227 * Find a way to reduce rotational delay when repeatedly writing
1230 Currently fsync of WAL requires the disk platter to perform a full
1231 rotation to fsync again. One idea is to write the WAL to different
1232 offsets that might reduce the rotational delay.
1234 * Allow buffered WAL writes and fsync
1236 Instead of guaranteeing recovery of all committed transactions, this
1237 would provide improved performance by delaying WAL writes and fsync
1238 so an abrupt operating system restart might lose a few seconds of
1239 committed transactions but still be consistent. We could perhaps
1240 remove the 'fsync' parameter (which results in an an inconsistent
1241 database) in favor of this capability.
1243 * Allow WAL logging to be turned off for a table, but the table
1244 might be dropped or truncated during crash recovery [walcontrol]
1246 Allow tables to bypass WAL writes and just fsync() dirty pages on
1247 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1248 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1249 non-default logging should not use referential integrity with
1250 default-logging tables. A table without dirty buffers during a
1251 crash could perhaps avoid the drop/truncate.
1253 * Allow WAL logging to be turned off for a table, but the table would
1254 avoid being truncated/dropped [walcontrol]
1256 To do this, only a single writer can modify the table, and writes
1257 must happen only on new pages so the new pages can be removed during
1258 crash recovery. Readers can continue accessing the table. Such
1259 tables probably cannot have indexes. One complexity is the handling
1260 of indexes on TOAST tables.
1263 Optimizer / Executor
1264 ====================
1266 * Improve selectivity functions for geometric operators
1267 * Allow ORDER BY ... LIMIT # to select high/low value without sort or
1268 index using a sequential scan for highest/lowest values
1270 Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
1271 all values to return the high/low value. Instead The idea is to do a
1272 sequential scan to find the high/low value, thus avoiding the sort.
1273 MIN/MAX already does this, but not for LIMIT > 1.
1275 * Precompile SQL functions to avoid overhead
1276 * Create utility to compute accurate random_page_cost value
1277 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1278 * Have EXPLAIN ANALYZE highlight poor optimizer estimates
1279 * Consider using hash buckets to do DISTINCT, rather than sorting
1281 This would be beneficial when there are few distinct values. This is
1282 already used by GROUP BY.
1284 * Log statements where the optimizer row estimates were dramatically
1285 different from the number of rows actually found?
1286 * Consider compressed annealing to search for query plans
1288 This might replace GEQO, http://sixdemonbag.org/Djinni.
1291 Miscellaneous Performance
1292 =========================
1294 * Do async I/O for faster random read-ahead of data
1296 Async I/O allows multiple I/O requests to be sent to the disk with
1297 results coming back asynchronously.
1299 * Use mmap() rather than SYSV shared memory or to write WAL files?
1301 This would remove the requirement for SYSV SHM but would introduce
1302 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1303 to prevent I/O overhead.
1305 * Consider mmap()'ing files into a backend?
1307 Doing I/O to large tables would consume a lot of address space or
1308 require frequent mapping/unmapping. Extending the file also causes
1309 mapping problems that might require mapping only individual pages,
1310 leading to thousands of mappings. Another problem is that there is no
1311 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1312 could hit disk before WAL is written.
1314 * Add a script to ask system configuration questions and tune postgresql.conf
1315 * Merge xmin/xmax/cmin/cmax back into three header fields
1317 Before subtransactions, there used to be only three fields needed to
1318 store these four values. This was possible because only the current
1319 transaction looks at the cmin/cmax values. If the current transaction
1320 created and expired the row the fields stored where xmin (same as
1321 xmax), cmin, cmax, and if the transaction was expiring a row from a
1322 another transaction, the fields stored were xmin (cmin was not
1323 needed), xmax, and cmax. Such a system worked because a transaction
1324 could only see rows from another completed transaction. However,
1325 subtransactions can see rows from outer transactions, and once the
1326 subtransaction completes, the outer transaction continues, requiring
1327 the storage of all four fields. With subtransactions, an outer
1328 transaction can create a row, a subtransaction expire it, and when the
1329 subtransaction completes, the outer transaction still has to have
1330 proper visibility of the row's cmin, for example, for cursors.
1332 One possible solution is to create a phantom cid which represents a
1333 cmin/cmax pair and is stored in local memory. Another idea is to
1334 store both cmin and cmax only in local memory.
1336 * Research storing disk pages with no alignment/padding
1342 * Add use of 'const' for variables in source tree
1343 * Move some things from /contrib into main tree
1344 * Move some /contrib modules out to their own project sites
1346 Particularly, move GPL-licensed /contrib/userlock and
1347 /contrib/dbmirror/clean_pending.pl.
1349 * %Remove warnings created by -Wcast-align
1350 * Move platform-specific ps status display info from ps_status.c to ports
1351 * Add optional CRC checksum to heap and index pages
1352 * Improve documentation to build only interfaces (Marc)
1353 * Remove or relicense modules that are not under the BSD license, if possible
1354 * %Remove memory/file descriptor freeing before ereport(ERROR)
1355 * Acquire lock on a relation before building a relcache entry for it
1356 * %Promote debug_query_string into a server-side function current_query()
1357 * %Allow the identifier length to be increased via a configure option
1358 * Allow cross-compiling by generating the zic database on the target system
1359 * Improve NLS maintenance of libpgport messages linked onto applications
1360 * Allow ecpg to work with MSVC and BCC
1361 * Add xpath_array() to /contrib/xml2 to return results as an array
1362 * Allow building in directories containing spaces
1364 This is probably not possible because 'gmake' and other compiler tools
1365 do not fully support quoting of paths with spaces.
1367 * -Allow installing to directories containing spaces
1369 This is possible if proper quoting is added to the makefiles for the
1370 install targets. Because PostgreSQL supports relocatable installs, it
1371 is already possible to install into a directory that doesn't contain
1372 spaces and then copy the install to a directory with spaces.
1374 * Fix sgmltools so PDFs can be generated with bookmarks
1375 * -%Clean up compiler warnings (especially with gcc version 4)
1376 * Use UTF8 encoding for NLS messages so all server encodings can
1378 * Update Bonjour to work with newer cross-platform SDK
1379 * -Remove BeOS and QNX-specific code
1380 * Split out libpq pgpass and environment documentation sections to make
1381 it easier for non-developers to find
1382 * Improve port/qsort() to handle sorts with 50% unique and 50% duplicate
1385 This involves choosing better pivot points for the quicksort.
1387 * Consider detoasting keys before sorting
1388 * Consider GnuTLS if OpenSSL license becomes a problem
1390 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1395 o Remove configure.in check for link failure when cause is found
1396 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1398 o Remove psql newline patch when we find out why mingw outputs an
1400 o Allow psql to use readline once non-US code pages work with
1402 o Re-enable timezone output on log_line_prefix '%t' when a
1403 shorter timezone string is available
1404 o Fix problem with shared memory on the Win32 Terminal Server
1405 o Improve signal handling
1407 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1409 o Add long file support for binary pg_dump output
1411 While Win32 supports 64-bit files, the MinGW API does not,
1412 meaning we have to build an fseeko replacement on top of the
1413 Win32 API, and we have to make sure MinGW handles it. Another
1414 option is to wait for the MinGW project to fix it, or use the
1415 code from the LibGW32C project as a guide.
1417 o -Port contrib/xml2
1418 o Check WSACancelBlockingCall() for interrupts [win32intr]
1421 * Wire Protocol Changes
1423 o Allow dynamic character set handling
1424 o Add decoded type, length, precision
1426 o Update clients to use data types, typmod, schema.table.column names
1427 of result sets using new statement protocol
1430 Features We Do _Not_ Want
1431 =========================
1433 * All backends running as threads in a single process (not want)
1435 This eliminates the process protection we get from the current setup.
1436 Thread creation is usually the same overhead as process creation on
1437 modern systems, so it seems unwise to use a pure threaded model.
1439 * Optimizer hints (not want)
1441 Optimizer hints are used to work around problems in the optimizer. We
1442 would rather have the problems reported and fixed.
1443 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1446 ---------------------------------------------------------------------------
1449 Developers who have claimed items are:
1450 --------------------------------------
1451 * Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
1452 * Andrew is Andrew Dunstan <andrew@dunslane.net>
1453 * Bruce is Bruce Momjian <bruce@momjian.us> of EnterpriseDB
1454 * Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
1455 Family Health Network
1456 * D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
1457 * Fabien is Fabien Coelho <coelho@cri.ensmp.fr>
1458 * Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
1459 * Greg is Greg Sabino Mullane <greg@turnstep.com>
1460 * Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
1461 * Joe is Joe Conway <mail@joeconway.com>
1462 * Karel is Karel Zak <zakkr@zf.jcu.cz>
1463 * Magnus is Magnus Hagander <mha@sollentuna.net>
1464 * Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
1465 * Matthew T. O'Connor <matthew@zeut.net>
1466 * Michael is Michael Meskes <meskes@postgresql.org> of Credativ
1467 * Neil is Neil Conway <neilc@samurai.com>
1468 * Oleg is Oleg Bartunov <oleg@sai.msu.su>
1469 * Pavel is Pavel Stehule <pavel.stehule@hotmail.com>
1470 * Peter is Peter Eisentraut <peter_e@gmx.net>
1471 * Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
1472 * Rod is Rod Taylor <pg@rbt.ca>
1473 * Simon is Simon Riggs <simon@2ndquadrant.com>
1474 * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
1475 * Tatsuo is Tatsuo Ishii <ishii@sraoss.co.jp> of SRA OSS, Inc. Japan
1476 * Teodor is Teodor Sigaev <teodor@sigaev.ru>
1477 * Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat