3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Fri Aug 15 22:36:53 EDT 2008
6 The most recent version of this document can be viewed at
7 http://www.postgresql.org/docs/faqs.TODO.html.
9 #A hyphen, "-", marks changes that will appear in the upcoming 8.4 release.#
10 #A percent sign, "%", marks items that are easier to implement.#
12 This list contains all known PostgreSQL bugs and feature requests. If
13 you would like to work on an item, please read the Developer's FAQ
14 first. There is also a developer's wiki at
15 http://developer.postgresql.org.
21 * -Allow administrators to safely terminate individual sessions either
22 via an SQL function or SIGTERM
23 * Check for unreferenced table files created by transactions that were
24 in-progress when the server terminated abruptly
26 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
28 * Set proper permissions on non-system schemas during db creation
30 Currently all schemas are owned by the super-user because they are copied
31 from the template1 database. However, since all objects are inherited
32 from the template database, it is not clear that setting schemas to the db
35 * -Add function to report the time of the most recent server reload
36 * Allow log_min_messages to be specified on a per-module basis
38 This would allow administrators to see more detailed information from
39 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
40 Another idea is to allow separate configuration files for each module,
41 or allow arbitrary SET commands to be passed to them.
43 * Simplify ability to create partitioned tables
45 This would allow creation of partitioned tables without requiring
46 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
47 for rapid partition selection. Options could include range and hash
50 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
51 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
52 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00028.php
53 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00248.php
54 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00387.php
55 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00413.php
57 * Allow auto-selection of partitioned tables for min/max() operations
58 * Allow more complex user/database default GUC settings
60 Currently ALTER USER and ALTER DATABASE support per-user and
61 per-database defaults. Consider adding per-user-and-database
62 defaults so things like search_path can be defaulted for a
63 specific user connecting to a specific database.
65 * Allow custom variables to appear in pg_settings()
66 * Allow custom variable classes that can restrict who can set the values
68 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
70 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
71 the privilege granted by the invoking role, and not those granted
74 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
76 * Allow SSL authentication/encryption over unix domain sockets
78 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
80 * Allow SSL key file permission checks to be optionally disabled when
81 sharing SSL keys with other applications
83 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00069.php
85 * Allow SSL client certificate names to be checked against the client
88 This is already implemented in
89 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
92 * Add 'hostgss' pg_hba.conf option to allow GSS link-level encryption
94 http://archives.postgresql.org/pgsql-hackers/2008-07/msg01454.php
96 * Improve server security options
98 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01875.php
99 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00000.php
101 * Prevent query cancel packets from being replayed by an attacker,
102 especially when using SSL
104 http://archives.postgresql.org/pgsql-hackers/2008-08/msg00345.php
107 * Configuration files
109 o Allow pg_hba.conf to specify host names along with IP addresses
111 Host name lookup could occur when the postmaster reads the
112 pg_hba.conf file, or when the backend starts. Another
113 solution would be to reverse lookup the connection IP and
114 check that hostname against the host names in pg_hba.conf.
115 We could also then check that the host name maps to the IP
117 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00569.php
119 o %Allow postgresql.conf file values to be changed via an SQL
120 API, perhaps using SET GLOBAL
121 o Allow the server to be stopped/restarted via an SQL API
122 o Issue a warning if a change-on-restart-only postgresql.conf value
123 is modified and the server config files are reloaded
124 o Consider normalizing fractions in postgresql.conf, perhaps
127 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00550.php
129 o Allow Kerberos to disable stripping of realms so we can
130 check the username@realm against multiple realms
132 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00009.php
134 o Add functions to syntax check configuration files
136 o Improve LDAP authentication configuration options
138 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01745.php
140 o Add external tool to auto-tune some postgresql.conf parameters
142 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00000.php
148 o Allow a database in tablespace t1 with tables created in
149 tablespace t2 to be used as a template for a new database created
150 with default tablespace t2
152 Currently all objects in the default database tablespace must
153 have default tablespace specifications. This is because new
154 databases are created by copying directories. If you mix default
155 tablespace tables and tablespace-specified tables in the same
156 directory, creating a new database from such a mixed directory
157 would create a new database with tables that had incorrect
158 explicit tablespaces. To fix this would require modifying
159 pg_class in the newly copied database, which we don't currently
162 o Allow reporting of which objects are in which tablespaces
164 This item is difficult because a tablespace can contain objects
165 from multiple databases. There is a server-side function that
166 returns the databases which use a specific tablespace, so this
167 requires a tool that will call that function and connect to each
168 database to find the objects in each database for that tablespace.
170 o Allow WAL replay of CREATE TABLESPACE to work when the directory
171 structure on the recovery computer is different from the original
173 o Allow per-tablespace quotas
176 * Statistics Collector
178 o Allow statistics collector information to be pulled from the collector
179 process directly, rather than requiring the collector to write a
180 filesystem file twice a second?
181 o Reduce file system activity overhead of statistics file pgstat.stat
183 http://archives.postgresql.org/pgsql-general/2007-12/msg00106.php
185 o Allow statistics last vacuum/analyze execution times to be displayed
186 without requiring stats_row_level to be enabled
188 http://archives.postgresql.org/pgsql-docs/2007-04/msg00028.php
190 o Clear table counters on TRUNCATE
192 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00169.php
195 * Point-In-Time Recovery (PITR)
197 o Allow a warm standby system to also allow read-only statements
199 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
201 o %Create dump tool for write-ahead logs for use in determining
202 transaction id for point-in-time recovery
204 This is useful for checking PITR recovery.
206 o Allow recovery.conf to support the same syntax as
207 postgresql.conf, including quoting
209 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
211 o -Fix server restart problem when the server was shutdown during
214 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00800.php
216 o Recreate pg_xlog/archive_status/ if it doesn't exist after
217 restoring from a PITR backup
219 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00487.php
221 o Reduce PITR WAL file size by removing full page writes and
222 by removing trailing bytes to improve compression
228 * Change NUMERIC to enforce the maximum precision
229 * Reduce storage space for small NUMERICs
231 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
232 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
233 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
235 * Fix data types where equality comparison isn't intuitive, e.g. box
236 * Add support for public SYNONYMs
238 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
240 * Fix CREATE CAST on DOMAINs
242 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
243 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
245 * Allow domains to be cast
247 http://archives.postgresql.org/pgsql-hackers/2003-06/msg01206.php
248 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00289.php
250 * Add support for SQL-standard GENERATED/IDENTITY columns
252 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
253 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
254 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
255 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
256 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
258 * Improve XML support
260 http://developer.postgresql.org/index.php/XML_Support
262 * Consider placing all sequences in a single table, or create a system
265 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
267 * Allow the UUID type to accept non-standard formats
269 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
271 * Consider a special data type for regular expressions
273 http://archives.postgresql.org/pgsql-hackers/2007-08/msg01067.php
275 * Reduce BIT data type overhead using short varlena headers
277 http://archives.postgresql.org/pgsql-general/2007-12/msg00273.php
279 * Allow xml arrays to be cast to other data types
281 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00981.php
282 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00231.php
283 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00471.php
285 * Simplify integer cross-data-type operators
287 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00189.php
289 * Allow adding/renaming/removing enumerated values to an existing
292 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01718.php
296 o Allow infinite dates and intervals just like infinite timestamps
297 o Merge hardwired timezone names with the TZ database; allow either
298 kind everywhere a TZ name is currently taken
299 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
300 information, either zone name or offset from UTC
302 If the TIMESTAMP value is stored with a time zone name, interval
303 computations should adjust based on the time zone rules.
304 http://archives.postgresql.org/pgsql-hackers/2004-10/msg00705.php
306 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
307 o Add a GUC variable to allow output of interval values in ISO8601
309 o Have timestamp subtraction not call justify_hours()?
311 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
313 o Improve timestamptz subtraction to be DST-aware
315 Currently subtracting one date from another that crosses a
316 daylight savings time adjustment can return '1 day 1 hour', but
317 adding that back to the first date returns a time one hour in
318 the future. This is caused by the adjustment of '25 hours' to
319 '1 day 1 hour', and '1 day' is the same time the next day, even
320 if daylight savings adjustments are involved.
322 o Fix interval display to support values exceeding 2^31 hours
323 o Add overflow checking to timestamp and interval arithmetic
324 o Extend timezone code to allow 64-bit values so we can
325 represent years beyond 2038
327 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
329 o -Use LC_TIME for localized weekday/month names, rather than
332 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
334 o Add ISO INTERVAL handling
336 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
337 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
339 o Support ISO INTERVAL syntax if units cannot be determined from
340 the string, and are supplied after the string
342 The SQL standard states that the units after the string
343 specify the units of the string, e.g. INTERVAL '2' MINUTE
344 should return '00:02:00'. The current behavior has the units
345 restrict the interval value to the specified unit or unit
346 range, INTERVAL '70' SECOND returns '00:00:10'.
348 For syntax that isn't uniquely ISO or PG syntax, like '1' or
349 '1:30', treat as ISO if there is a range specification clause,
350 and as PG if there no clause is present, e.g. interpret '1:30'
351 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
352 '1:30' as '1 hour, 30 minutes'.
354 This makes common cases like SELECT INTERVAL '1' MONTH
355 SQL-standard results. The SQL standard supports a limited
356 number of unit combinations and doesn't support unit names in
357 the string. The PostgreSQL syntax is more flexible in the
358 range of units supported, e.g. PostgreSQL supports '1 year 1
359 hour', while the SQL standard does not.
361 o Add support for year-month syntax, INTERVAL '50-6' YEAR
363 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
364 year' AS INTERVAL MONTH), and this should return '12 months'
365 o Round or truncate values to the requested precision, e.g.
366 INTERVAL '11 months' AS YEAR should return one or zero
367 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
372 o Delay resolution of array expression's data type so assignment
373 coercion can be performed on empty array expressions
374 o Add support for arrays of domains
376 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
378 o Allow single-byte header storage for array elements
383 o Improve vacuum of large objects, like contrib/vacuumlo?
384 o Add security checking for large objects
385 o Auto-delete large objects when referencing row is deleted
387 contrib/lo offers this functionality.
389 o Allow read/write into TOAST values like large objects
391 This requires the TOAST column to be stored EXTERNAL.
393 o Add API for 64-bit large object access
395 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
399 * Add locale-aware MONEY type, and support multiple currencies
401 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
402 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
404 * MONEY dumps in a locale-specific format making it difficult to
405 restore to a system with a different locale
406 * Allow MONEY to be easily cast to/from other numeric data types
410 o Allow dictionaries to change the token that is passed on to
413 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
415 o Consider a function-based API for '@@' searches
417 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
419 o Improve text search error messages
421 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00966.php
422 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01146.php
424 o Consider changing error to warning for strings larger than one
427 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00190.php
428 http://archives.postgresql.org/pgsql-patches/2008-03/msg00062.php
435 * Allow INET subnet tests using non-constants to be indexed
436 * Allow to_date() and to_timestamp() accept localized month names
437 * Fix to_date()-related functions to consistently issue errors
439 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
441 * Add missing parameter handling in to_char()
443 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
445 * Allow substring/replace() to get/set bit values
446 * Allow to_char() on interval values to accumulate the highest unit
449 Some special format flag would be required to request such
450 accumulation. Such functionality could also be added to EXTRACT.
451 Prevent accumulation that crosses the month/day boundary because of
452 the uneven number of days in a month.
454 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
455 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
456 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
457 o to_char(INTERVAL '3 years 5 months','MM') => 41
459 * Implement inlining of set-returning functions defined in SQL
460 * Allow SQL-language functions to return results from RETURNING queries
462 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
464 * Allow SQL-language functions to reference parameters by parameter name
466 Currently SQL-language functions can only refer to dollar parameters,
469 * Add SPI_gettypmod() to return the typemod for a TupleDesc
470 * Enforce typmod for function inputs, function results and parameters for
471 spi_prepare'd statements called from PLs
473 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
475 * Allow holdable cursors in SPI
476 * Tighten function permission checks
478 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
480 * Fix IS OF so it matches the ISO specification, and add documentation
482 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
483 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
485 * Add missing operators for geometric data types
487 Some geometric types do not have the full suite of geometric operators,
490 * Implement Boyer-Moore searching in strpos()
492 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
494 * Prevent malicious functions from being executed with the permissions
495 of unsuspecting users
497 Index functions are safe, so VACUUM and ANALYZE are safe too.
498 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
499 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
501 * Reduce memory usage of aggregates in set returning functions
503 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
505 * -Add temporal versions of generate_series()
507 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
509 * Add array_accum() and array_to_set() functions for arrays
511 The standards specify array_agg() and UNNEST.
512 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
514 * Fix /contrib/ltree operator
516 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
518 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
520 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
522 * Fix regular expression bug when using complex back-references
524 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00000.php
526 * Have /contrib/dblink reuse unnamed connections
528 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00895.php
530 * Add SQL-standard array_agg() and unnest() array functions
532 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01017.php
534 * Allow calling of a procedure outside a SELECT that can control the
537 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01375.php
541 Multi-Language Support
542 ======================
544 * Add NCHAR (as distinguished from ordinary varchar),
545 * Allow locale to be set at database creation
547 Currently locale can only be set during initdb. No global tables have
548 locale-aware columns. However, the database template used during
549 database creation might have locale-aware indexes. The indexes would
550 need to be reindexed to match the new locale.
552 * Allow encoding on a per-column basis optionally using the ICU library;
555 Right now only one encoding is allowed per database.
557 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
558 http://archives.postgresql.org/pgsql-patches/2005-08/msg00039.php
559 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
560 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00110.php
561 http://archives.postgresql.org/pgsql-patches/2005-09/msg00020.php
562 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01121.php
563 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00767.php
564 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
565 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
566 http://wiki.postgresql.org/wiki/Todo:Collate
567 http://wiki.postgresql.org/wiki/Todo:ICU
569 * Support multiple simultaneous character sets, per SQL92
570 * Improve UTF8 combined character handling?
571 * Add octet_length_server() and octet_length_client()
572 * Make octet_length_client() the same as octet_length()?
573 * Fix problems with wrong runtime encoding conversion for NLS message files
574 * Add URL to more complete multi-byte regression tests
576 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
578 * Fix ILIKE and regular expressions to handle case insensitivity
579 properly in multibyte encodings
581 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
582 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
584 * Set client encoding based on the client operating system encoding
586 Currently client_encoding is set in postgresql.conf, which
587 defaults to the server encoding.
588 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
590 * Change memory allocation for multi-byte functions so memory is
591 allocated inside conversion functions
593 Currently we preallocate memory based on worst-case usage.
600 * Automatically create rules on views so they are updateable, per SQL99
602 We can only auto-create rules for simple views. For more complex
603 cases users will still have to write rules manually.
605 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
606 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
608 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
609 * Allow VIEW/RULE recompilation when the underlying tables change
611 Another issue is whether underlying table changes should be reflected
612 in the view, e.g. should SELECT * show additional columns if they
613 are added after the view is created.
615 * Make it possible to use RETURNING together with conditional DO INSTEAD
616 rules, such as for partitioning setups
618 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
620 * Add the ability to automatically create materialized views
622 Right now materialized views require the user to create triggers on the
623 main table to keep the summary table current. SQL syntax should be able
624 to manager the triggers and summary table automatically. A more
625 sophisticated implementation would automatically retrieve from the
626 summary table when the main table is referenced, if possible.
628 * Improve ability to modify views via ALTER TABLE
630 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00691.php
631 http://archives.postgresql.org/pgsql-hackers/2008-07/msg01410.php
632 http://archives.postgresql.org/pgsql-hackers/2008-08/msg00300.php
638 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
639 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
640 * %Allow SET CONSTRAINTS to be qualified by schema/table name
641 * %Add a separate TRUNCATE permission
643 Currently only the owner can TRUNCATE a table because triggers are not
644 called, and the table is locked in exclusive mode.
646 * Fix TRUNCATE ... RESTART IDENTITY so its affect on sequences is rolled
647 back on transaction abort
648 * Allow PREPARE of cursors
649 * Allow finer control over the caching of prepared query plans
651 Currently queries prepared via the libpq API are planned on first
652 execute using the supplied parameters --- allow SQL PREPARE to do the
653 same. Also, allow control over replanning prepared queries either
654 manually or automatically when statistics for execute parameters
655 differ dramatically from those used during planning.
657 * Improve logging of prepared transactions recovered during startup
659 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
661 * Improve failure message when DROP DATABASE is used on a database that
662 has prepared transactions
663 * Allow prepared transactions with temporary tables created and dropped
664 in the same transaction, and when an ON COMMIT DELETE ROWS temporary
667 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00047.php
669 * Add a GUC variable to warn about non-standard SQL usage in queries
670 * Add SQL-standard MERGE/REPLACE/UPSERT command
672 MERGE is typically used to merge two tables. REPLACE or UPSERT
673 command does UPDATE, or on failure, INSERT. This is similar to UPDATE,
674 then for unmatched rows, INSERT. Whether concurrent access allows
675 modifications which could cause row loss is implementation independent.
676 To implement this cleanly requires that the table have a unique index
677 so duplicate checking can be easily performed. It is possible to do it
678 without a unique index if we require the user to LOCK the table before
681 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
682 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
683 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01157.php
684 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01475.php
685 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01890.php
687 * Add NOVICE output level for helpful messages like automatic sequence/index
689 * Add GUC to issue notice about statements that use unjoined tables
690 * Allow EXPLAIN to identify tables that were skipped because of
692 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
693 * Enable standard_conforming_strings
694 * Make standard_conforming_strings the default in 8.5?
696 When this is done, backslash-quote should be prohibited in non-E''
697 strings because of possible confusion over how such strings treat
698 backslashes. Basically, '' is always safe for a literal single
699 quote, while \' might or might not be based on the backslash
702 * Simplify dropping roles that have objects in several databases
703 * Allow COMMENT ON to accept an expression rather than just a string
704 * Allow the count returned by SELECT, etc to be represented as an int64
705 to allow a higher range of values
706 * Add SQL99 WITH clause to SELECT
707 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
709 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
710 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
711 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
712 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
713 http://archives.postgresql.org/pgsql-patches/2008-01/msg00105.php
714 http://archives.postgresql.org/pgsql-patches/2008-03/msg00327.php
716 * Add DEFAULT .. AS OWNER so permission checks are done as the table
719 This would be useful for SERIAL nextval() calls and CHECK constraints.
721 * Allow DISTINCT to work in multiple-argument aggregate calls
722 * Add column to pg_stat_activity that shows the progress of long-running
723 commands like CREATE INDEX and VACUUM
725 http://archives.postgresql.org/pgsql-patches/2008-04/msg00203.php
727 * Implement SQL:2003 window functions
729 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00380.php
730 http://archives.postgresql.org/pgsql-hackers/2008-07/msg00232.php
732 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause or
735 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
736 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
737 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00124.php
739 * Increase locking when DROPing objects so dependent objects cannot
740 get dropped while the DROP operation is happening
742 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
744 * -Allow AS in "SELECT col AS label" to be optional in certain cases
746 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
747 RETURNING to supply values to the INSERT
748 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
750 * Add comments on system tables/columns using the information in
753 Ideally the information would be pulled from the SGML file
756 * Improve reporting of UNION type mismatches
758 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
759 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
764 o Allow CREATE TABLE AS to determine column lengths for complex
765 expressions like SELECT col1 || col2
767 o Have WITH CONSTRAINTS also create constraint indexes
769 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
771 o Have CONSTRAINT cname NOT NULL record the contraint name
773 Right now pg_attribute.attnotnull records the NOT NULL status
774 of the column, but does not record the contraint name
776 o Prevent concurrent CREATE TABLE table1 from sometimes returning
777 a cryptic error message
779 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00169.php
781 o Add CREATE SCHEMA ... LIKE that copies a schema
783 o Add CREATE TABLE LIKE ... INCLUDING COMMENTS
787 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
789 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
790 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
791 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
792 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
794 o Research self-referential UPDATEs that see inconsistent row versions
795 in read-committed mode
797 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
798 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
800 o Allow GLOBAL temporary tables to exist as empty by default in
803 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
808 o Have ALTER TABLE RENAME rename SERIAL sequence names
810 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
812 o Have ALTER SEQUENCE RENAME rename the sequence name stored
813 in the sequence table
815 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
816 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
817 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
819 o Add ALTER DOMAIN to modify the underlying data type
820 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
822 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
824 o %Allow ALTER TABLE to change constraint deferrability and actions
825 o Add missing object types for ALTER ... SET SCHEMA
826 o Allow ALTER TABLESPACE to move to different directories
827 o Allow databases to be moved to different tablespaces
828 o Allow moving system tables to other tablespaces, where possible
830 Currently non-global system tables must be in the default database
831 tablespace. Global system tables can never be moved.
833 o -Prevent parent tables from altering or dropping constraints
834 like CHECK that are inherited by child tables unless CASCADE
836 o -Prevent child tables from altering or dropping constraints
837 like CHECK that were inherited from the parent table
838 o Have ALTER INDEX update the name of a constraint using that index
839 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
840 o Allow column display reordering by recording a display,
841 storage, and permanent id for every column?
843 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
845 o Allow an existing index to be marked as a table's primary key
847 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00500.php
852 o Automatically maintain clustering on a table
854 This might require some background daemon to maintain clustering
855 during periods of low usage. It might also require tables to be only
856 partially filled for easier reorganization. Another idea would
857 be to create a merged heap/index data file so an index lookup would
858 automatically access the heap data too. A third idea would be to
859 store heap rows in hashed groups, perhaps using a user-supplied
861 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
863 o %Add default clustering to system tables
865 To do this, determine the ideal cluster index for each system
866 table and set the cluster setting during initdb.
868 o %Add VERBOSE option to report tables as they are processed,
874 o Allow COPY to report error lines and continue
876 This requires the use of a savepoint before each COPY line is
877 processed, with ROLLBACK on COPY failure.
878 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
880 o Allow COPY on a newly-created table to skip WAL logging
882 On crash recovery, the table involved in the COPY would
883 be removed or have its heap and index files truncated. One
884 issue is that no other backend should be able to add to
885 the table at the same time, which is something that is
886 currently allowed. This currently is done if the table is
887 created inside the same transaction block as the COPY because
888 no other backends can see the table.
890 o Consider using a ring buffer for COPY FROM
892 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
893 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
895 o Allow COPY FROM to create index entries in bulk
897 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
899 o Allow COPY in CSV mode to control whether a quoted zero-length
900 string is treated as NULL
902 Currently this is always treated as a zero-length string,
903 which generates an error when loading into an integer column
904 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
906 o Impove COPY performance
908 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00954.php
910 o Allow COPY to report errors sooner
912 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01169.php
918 o Allow column-level privileges
919 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
922 The proposed syntax is:
923 GRANT SELECT ON ALL TABLES IN public TO phpuser;
924 GRANT SELECT ON NEW TABLES IN public TO phpuser;
926 o Allow GRANT/REVOKE permissions to be inherited by objects based on
929 o Allow SERIAL sequences to inherit permissions from the base table?
934 o Prevent DROP TABLE from dropping a row referenced by its own open
940 o Allow INSERT/UPDATE of the system-generated oid value for a row
941 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
947 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
953 o Allow LISTEN/NOTIFY to store info in memory rather than tables?
955 Currently LISTEN/NOTIFY information is stored in pg_listener.
956 Storing such information in memory would improve performance.
958 o Add optional textual message to NOTIFY
960 This would allow an informational message to be added to the notify
961 message, perhaps indicating the row modified or other custom
964 o Allow multiple identical NOTIFY events to always be communicated
965 to the client, rather than sent as a single notification to the
968 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
970 o Allow NOTIFY in rules involving conditionals
971 o Improve LISTEN concurrency
973 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01106.php
977 Referential Integrity
978 =====================
980 * Add MATCH PARTIAL referential integrity
981 * Change foreign key constraint for array -> element to mean element
983 * Fix problem when cascading referential triggers make changes on
984 cascaded tables, seeing the tables in an intermediate state
986 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
987 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
989 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
991 This would allow UPDATE tab SET col = col + 1 to work if col has
992 a unique index. Currently, uniqueness checks are done while the
993 command is being executed, rather than at the end of the statement
995 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
996 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
998 * Optimize referential integrity checks
1000 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
1001 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
1004 Server-Side Languages
1005 =====================
1008 o Fix RENAME to work on variables other than OLD/NEW
1010 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
1011 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
1012 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
1014 o Allow function parameters to be passed by name,
1015 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
1016 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
1017 o Allow listing of record column names, and access to
1018 record columns via variables, e.g. columns := r.(*),
1019 tval2 := r.(colname)
1021 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
1022 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
1023 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
1025 o Add support for SCROLL cursors
1026 o Add support for WITH HOLD cursors
1027 o Allow row and record variables to be set to NULL constants,
1028 and allow NULL tests on such variables
1030 Because a row is not scalar, do not allow assignment
1031 from NULL-valued scalars.
1033 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
1035 o Review handling of MOVE and FETCH
1037 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
1039 o Improve logic of determining if an identifier is a a
1040 variable or column name
1042 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
1044 o Consider keeping seperate cached copies when search_path changes
1046 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01009.php
1048 o -Add CASE capability to language (already in SQL)
1050 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00696.php
1056 o Add table function support to pltcl, plpythonu
1057 o Add support for polymorphic arguments and return types to
1058 languages other than PL/PgSQL
1059 o Add capability to create and call PROCEDURES
1060 o Add support for OUT and INOUT parameters to languages other
1062 o Add PL/PythonU tracebacks
1064 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
1066 o Allow data to be passed in native language formats, rather
1069 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
1071 o Add ability to obfuscate function bodies
1073 http://archives.postgresql.org/pgsql-patches/2008-01/msg00125.php
1080 * Have pg_ctl look at PGHOST in case it is a socket directory?
1081 * Allow pg_ctl to work properly with configuration files located outside
1082 the PGDATA directory
1084 pg_ctl can not read the pid file because it isn't located in the
1085 config directory but in the PGDATA directory. The solution is to
1086 allow pg_ctl to read and understand postgresql.conf to find the
1087 data_directory value.
1089 * Add a function like pg_get_indexdef() that report more detailed index
1092 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
1095 o -Have psql show current values for a sequence
1096 o Move psql backslash database information into the backend, use
1099 This would allow non-psql clients to pull the same information out
1100 of the database as psql.
1101 http://archives.postgresql.org/pgsql-hackers/2004-01/msg00191.php
1103 o Make psql's \d commands more consistent
1105 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1106 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1108 o Consistently display privilege information for all objects in psql
1109 o Add auto-expanded mode so expanded output is used if the row
1110 length is wider than the screen width.
1112 Consider using auto-expanded mode for backslash commands like \df+.
1114 o Prevent tab completion of SET TRANSACTION from querying the
1115 database and therefore preventing the transaction isolation
1116 level from being set.
1118 Currently SET <tab> causes a database lookup to check all
1119 supported session variables. This query causes problems
1120 because setting the transaction isolation level must be the
1121 first statement of a transaction.
1123 o Add a \set variable to control whether \s displays line numbers
1125 Another option is to add \# which lists line numbers, and
1126 allows command execution.
1128 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
1130 o Prevent escape string warnings when object names have
1133 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1135 o Have \d show foreign keys that reference a table's primary key
1137 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1139 o Have \d show child tables that inherit from the specified parent
1140 o -Have \l+ show database size, if permissions allow
1142 Ideally it will not generate an error for invalid permissions
1144 o Include the symbolic SQLSTATE name in verbose error reports
1146 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1148 o -Improve display of enums to show valid enum values
1149 o Add prompt escape to display the client and server versions
1150 o Add option to wrap column values at whitespace boundaries,
1151 rather than chopping them at a fixed width.
1153 Currently, "wrapped" format chops values into fixed
1154 widths. Perhaps the word wrapping could use the same
1155 algorithm documented in the W3C specification.
1156 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00404.php
1157 http://www.w3.org/TR/CSS21/tables.html#auto-table-layout
1158 o Add "auto" expanded mode that outputs in expanded format if
1159 "wrapped" mode can't wrap the output to the screen width
1161 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00417.php
1164 * pg_dump / pg_restore
1165 o %Add dumping of comments on index columns and composite type columns
1166 o %Add full object name to the tag field. eg. for operators we need
1167 '=(integer, integer)', instead of just '='.
1168 o Add pg_dumpall custom format dumps?
1169 o Allow selection of individual object(s) of all types, not just
1171 o In a selective dump, allow dumping of an object and all its
1173 o Add options like pg_restore -l and -L to pg_dump
1174 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1175 o Allow pg_dump --clean to drop roles that own objects or have
1177 o Change pg_dump so that a comment on the dumped database is
1178 applied to the loaded database, even if the database has a
1179 different name. This will require new backend syntax, perhaps
1180 COMMENT ON CURRENT DATABASE.
1181 o Remove unnecessary function pointer abstractions in pg_dump source
1183 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1184 multiple objects simultaneously
1186 The difficulty with this is getting multiple dump processes to
1187 produce a single dump output file. It also would require
1188 several sessions to share the same snapshot.
1189 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1191 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1192 restoring multiple objects simultaneously
1194 This might require a pg_restore flag to indicate how many
1195 simultaneous operations should be performed. Only pg_dump's
1196 -Fc format has the necessary dependency information.
1197 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1199 o To better utilize resources, allow pg_restore to check foreign
1200 keys simultaneously, where possible
1201 o Allow pg_restore to create all indexes of a table
1202 concurrently, via a single heap scan
1204 This requires a pg_dump -Fc file because that format contains
1205 the required dependency information.
1206 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1208 o Allow pg_restore to load different parts of the COPY data
1210 o -Prevent pg_dump/pg_restore from being affected by
1213 Using psql to restore a pg_dump dump is also affected.
1215 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1216 o Allow pre/data/post files when schema and data are dumped
1217 separately, for performance reasons
1219 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1226 Document differences between ecpg and the SQL standard and
1227 information about the Informix-compatibility module.
1229 o Solve cardinality > 1 for input descriptors / variables?
1230 o Add a semantic check level, e.g. check if a table really exists
1231 o fix handling of DB attributes that are arrays
1232 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1234 o Fix nested C comments
1235 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1236 o Make SET CONNECTION thread-aware, non-standard?
1237 o Allow multidimensional arrays
1238 o Add internationalized message strings
1239 o Implement COPY FROM STDIN
1243 o Add PQescapeIdentifierConn()
1244 o Prevent PQfnumber() from lowercasing unquoted the column name
1246 PQfnumber() should never have been doing lowercasing, but
1247 historically it has so we need a way to prevent it
1249 o Allow statement results to be automatically batched to the client
1251 Currently all statement results are transferred to the libpq
1252 client before libpq makes the results available to the
1253 application. This feature would allow the application to make
1254 use of the first result rows while the rest are transferred, or
1255 held on the server waiting for them to be requested by libpq.
1256 One complexity is that a statement like SELECT 1/col could error
1257 out mid-way through the result set.
1259 o Consider disallowing multiple queries in PQexec() as an
1260 additional barrier to SQL injection attacks
1262 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1264 o Add PQexecf() that allows complex parameter substitution
1266 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1268 o Add SQLSTATE severity to PGconn return status
1270 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1276 * Add deferred trigger queue file
1278 Right now all deferred trigger information is stored in backend
1279 memory. This could exhaust memory for very large trigger queues.
1280 This item involves dumping large queues into files, or doing some
1281 kind of join to process all the triggers, or some bulk operation.
1282 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00876.php
1284 * Allow triggers to be disabled in only the current session.
1286 This is currently possible by starting a multi-statement transaction,
1287 modifying the system tables, performing the desired SQL, restoring the
1288 system tables, and committing the transaction. ALTER TABLE ...
1289 TRIGGER requires a table lock so it is not ideal for this usage.
1291 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1293 If the dump is known to be valid, allow foreign keys to be added
1294 without revalidating the data.
1296 * Allow statement-level triggers to access modified rows
1297 * Support triggers on columns
1299 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1301 * Allow AFTER triggers on system tables
1303 System tables are modified in many places in the backend without going
1304 through the executor and therefore not causing triggers to fire. To
1305 complete this item, the functions that modify system tables will have
1308 * Tighten trigger permission checks
1310 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1312 * Allow BEFORE INSERT triggers on views
1314 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1316 * -Add ability to trigger on TRUNCATE
1317 * Add database and transaction-level triggers
1319 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00451.php
1320 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00620.php
1322 * Reduce locking requirements for creating a trigger
1324 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00635.php
1331 * Add UNIQUE capability to non-btree indexes
1332 * Prevent index uniqueness checks when UPDATE does not modify the column
1334 Uniqueness (index) checks are done when updating a column even if the
1335 column is not modified by the UPDATE.
1337 * Allow the creation of on-disk bitmap indexes which can be quickly
1338 combined with other bitmap indexes
1340 Such indexes could be more compact if there are only a few distinct values.
1341 Such indexes can also be compressed. Keeping such indexes updated can be
1344 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1345 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1346 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00265.php
1347 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01214.php
1348 http://archives.postgresql.org/pgsql-patches/2007-05/msg00013.php
1349 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00741.php
1351 * Allow accurate statistics to be collected on indexes with more than
1352 one column or expression indexes, perhaps using per-index statistics
1354 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1355 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1357 * Consider increasing the default and maximum number of statistics targets,
1358 and reduce statistics target overhead
1360 Also consider having a larger statistics target for indexed columns
1361 and expression indexes.
1362 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1363 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1364 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1365 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00188.php
1367 * Consider smaller indexes that record a range of values per heap page,
1368 rather than having one index entry for every heap row
1370 This is useful if the heap is clustered by the indexed values.
1371 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1372 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1373 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1374 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1375 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00014.php
1376 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00487.php
1377 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01589.php
1379 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1381 This is difficult because you must upgrade to an exclusive table lock
1382 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1383 have this complication. This would allow index compaction without
1385 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1387 * Allow multiple indexes to be created concurrently, ideally via a
1388 single heap scan, and have pg_restore use it
1390 * Consider sorting entries before inserting into btree index
1392 http://archives.postgresql.org/pgsql-general/2008-01/msg01010.php
1394 * Allow index scans to return matching index keys, not just the matching
1397 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01657.php
1403 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1404 and primary/foreign keys
1405 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1406 on inherited table, e.g. INSERT INTO inherit_table
1407 (unique_index_col) VALUES (dup) should fail
1409 The main difficulty with this item is the problem of
1410 creating an index that can span multiple tables.
1412 o Allow SELECT ... FOR UPDATE on inherited tables
1413 o Require all CHECK constraints to be inherited
1415 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1417 o Add checks to prevent a CREATE RULE views on inherited tables
1419 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1420 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1425 o Add more GIST index support for geometric data types
1426 o Allow GIST indexes to create certain complex index types, like
1427 digital trees (see Aoki)
1432 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1434 o Pack hash index buckets onto disk pages more efficiently
1436 Currently only one hash bucket can be stored on a page. Ideally
1437 several hash buckets could be stored on a single page and greater
1438 granularity used for the hash algorithm.
1440 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1442 o Consider sorting hash buckets so entries can be found using a
1443 binary search, rather than a linear scan
1444 o In hash indexes, consider storing the hash value with or instead
1446 o Add WAL logging for crash recovery
1447 o Allow multi-column hash indexes
1448 o -During index creation, pre-sort the tuples to improve build speed
1452 * Consider using hash buckets to do DISTINCT, rather than sorting
1454 This would be beneficial when there are few distinct values. This is
1455 already used by GROUP BY.
1457 * Consider whether duplicate keys should be sorted by block/offset
1459 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00558.php
1461 * -Avoid tuple some tuple copying in sort routines
1462 * Consider being smarter about memory and external files used during
1465 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01101.php
1466 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00045.php
1468 * Consider detoasting keys before sorting
1474 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1476 Ideally this requires a separate test program that can be run
1477 at initdb time or optionally later. Consider O_SYNC when
1480 * Add program to test if fsync has a delay compared to non-fsync
1481 * Consider sorting writes during checkpoint
1483 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1492 We could use a fixed row count and a +/- count to follow MVCC
1493 visibility rules, or a single cached value could be used and
1494 invalidated if anyone modifies the table. Another idea is to
1495 get a count directly from a unique index, but for this to be
1496 faster than a sequential scan it must avoid access to the heap
1497 to obtain tuple visibility information.
1499 * Provide a way to calculate an "estimated COUNT(*)"
1501 Perhaps by using the optimizer's cardinality estimates or random
1504 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1506 * Allow data to be pulled directly from indexes
1508 Currently indexes do not have enough tuple visibility information
1509 to allow data to be pulled from the index without also accessing
1510 the heap. One way to allow this is to set a bit on index tuples
1511 to indicate if a tuple is currently visible to all transactions
1512 when the first valid heap lookup happens. This bit would have to
1513 be cleared when a heap tuple is expired.
1515 Another idea is to maintain a bitmap of heap pages where all rows
1516 are visible to all backends, and allow index lookups to reference
1517 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1518 add someday to determine which heap pages need vacuuming. Frequently
1519 accessed bitmaps would have to be stored in shared memory. One 8k
1520 page of bitmaps could track 512MB of heap pages.
1522 A third idea would be for a heap scan to check if all rows are visible
1523 and if so set a per-table flag which can be checked by index scans.
1524 Any change to the table would have to clear the flag. To detect
1525 changes during the heap scan a counter could be set at the start and
1526 checked at the end --- if it is the same, the table has not been
1527 modified --- any table change would increment the counter.
1529 http://archives.postgresql.org/pgsql-patches/2007-10/msg00166.php
1530 http://archives.postgresql.org/pgsql-patches/2008-01/msg00049.php
1532 * Consider automatic caching of statements at various levels:
1535 o Query execute plan
1537 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00823.php
1539 * Consider increasing internal areas when shared buffers is increased
1541 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1543 * Consider decreasing the amount of memory used by PrivateRefCount
1545 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1546 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1548 * Consider allowing higher priority queries to have referenced buffer
1549 cache pages stay in memory longer
1551 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1558 * Improve speed with indexes
1560 For large table adjustments during VACUUM FULL, it is faster to cluster
1561 or reindex rather than update the index. Also, index updates can bloat
1564 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1565 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1566 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1568 * Auto-fill the free space map by scanning the buffer cache or by
1569 checking pages written by the background writer
1571 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1572 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1574 * Create a bitmap of pages that need vacuuming
1576 Instead of sequentially scanning the entire table, have the background
1577 writer or some other process record pages that have expired rows, then
1578 VACUUM can look at just those pages rather than the entire table. In
1579 the event of a system crash, the bitmap would probably be invalidated.
1580 One complexity is that index entries still have to be vacuumed, and
1581 doing this without an index scan (by using the heap values to find the
1582 index entry) might be slow and unreliable, especially for user-defined
1585 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1586 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1587 http://archives.postgresql.org/pgsql-patches/2007-03/msg00508.php
1588 http://archives.postgresql.org/pgsql-patches/2007-04/msg00347.php
1589 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00156.php
1590 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00546.php
1591 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00416.php
1593 * Allow FSM to return free space toward the beginning of the heap file,
1594 in hopes that empty pages at the end can be truncated by VACUUM
1595 * Allow FSM page return free space based on table clustering, to assist
1596 in maintaining clustering?
1597 * -Improve dead row detection during multi-statement transactions usage
1599 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1601 * Consider a more compact data representation for dead tuples
1603 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1608 o %Issue log message to suggest VACUUM FULL if a table is nearly
1610 o Improve control of auto-vacuum
1612 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1614 o Prevent long-lived temporary tables from causing frozen-xid
1615 advancement starvation
1617 The problem is that autovacuum cannot vacuum them to set frozen xids;
1618 only the session that created them can do that.
1619 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1621 o Store per-table autovacuum settings in pg_class.reloptions.
1623 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1624 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1626 o Prevent autovacuum from running if an old transaction is still
1627 running from the last vacuum
1629 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1636 * Fix priority ordering of read and write light-weight locks (Neil)
1638 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1639 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1641 * Fix problem when multiple subtransactions of the same outer transaction
1642 hold different types of locks, and one subtransaction aborts
1644 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1645 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1646 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1647 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1649 * Allow UPDATEs on only non-referential integrity columns not to conflict
1650 with referential integrity locks
1652 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1654 * Add idle_in_transaction_timeout GUC so locks are not held for long
1657 * Improve deadlock detection when a page cleaning lock conflicts
1658 with a shared buffer that is pinned
1660 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1661 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1662 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1664 * Detect deadlocks involving LockBufferForCleanup()
1666 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1670 Startup Time Improvements
1671 =========================
1673 * Experiment with multi-threaded backend for backend creation
1675 This would prevent the overhead associated with process creation. Most
1676 operating systems have trivial process creation time compared to
1677 database startup overhead, but a few operating systems (Win32,
1678 Solaris) might benefit from threading. Also explore the idea of
1679 a single session using multiple threads to execute a statement faster.
1686 * Eliminate need to write full pages to WAL before page modification
1688 Currently, to protect against partial disk page writes, we write
1689 full page images to WAL before they are modified so we can correct any
1690 partial page writes during recovery. These pages can also be
1691 eliminated from point-in-time archive files.
1692 http://archives.postgresql.org/pgsql-hackers/2002-06/msg00655.php
1694 o When off, write CRC to WAL and check file system blocks
1697 If CRC check fails during recovery, remember the page in case
1698 a later CRC for that page properly matches.
1700 o Write full pages during file system write and not when
1701 the page is modified in the buffer cache
1703 This allows most full page writes to happen in the background
1704 writer. It might cause problems for applying WAL on recovery
1705 into a partially-written page, but later the full page will be
1708 * Allow WAL traffic to be streamed to another server for stand-by
1710 * Reduce WAL traffic so only modified values are written rather than
1713 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1715 * Allow WAL information to recover corrupted pg_controldata
1717 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1719 * Find a way to reduce rotational delay when repeatedly writing
1722 Currently fsync of WAL requires the disk platter to perform a full
1723 rotation to fsync again. One idea is to write the WAL to different
1724 offsets that might reduce the rotational delay.
1725 http://archives.postgresql.org/pgsql-hackers/2002-11/msg00483.php
1727 * Allow WAL logging to be turned off for a table, but the table
1728 might be dropped or truncated during crash recovery
1730 Allow tables to bypass WAL writes and just fsync() dirty pages on
1731 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1732 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1733 non-default logging should not use referential integrity with
1734 default-logging tables. A table without dirty buffers during a
1735 crash could perhaps avoid the drop/truncate.
1736 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1738 * Allow WAL logging to be turned off for a table, but the table would
1739 avoid being truncated/dropped
1741 To do this, only a single writer can modify the table, and writes
1742 must happen only on new pages so the new pages can be removed during
1743 crash recovery. Readers can continue accessing the table. Such
1744 tables probably cannot have indexes. One complexity is the handling
1745 of indexes on TOAST tables.
1746 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1748 * Speed WAL recovery by allowing more than one page to be prefetched
1750 This should be done utilizing the same infrastructure used for
1751 prefetching in general to avoid introducing complex error-prone code
1753 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1754 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1755 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1757 * Improve WAL concurrency by increasing lock granularity
1759 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1761 * Be more aggressive about creating WAL files
1763 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01325.php
1764 http://archives.postgresql.org/pgsql-hackers/2004-07/msg01075.php
1765 http://archives.postgresql.org/pgsql-hackers/2005-04/msg00556.php
1767 * Have resource managers report the duration of their status changes
1769 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01468.php
1771 * Move pgfoundry's xlogdump to /contrib and have it rely more closely
1772 on the WAL backend code
1774 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00035.php
1779 Optimizer / Executor
1780 ====================
1782 * Improve selectivity functions for geometric operators
1783 * Precompile SQL functions to avoid overhead
1784 * Create utility to compute accurate random_page_cost value
1785 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1786 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1787 actual row counts differ by a specified percentage
1788 * Improve how ANALYZE computes in-doubt tuples
1790 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00771.php
1792 * Log statements where the optimizer row estimates were dramatically
1793 different from the number of rows actually found?
1794 * Consider compressed annealing to search for query plans
1796 This might replace GEQO, http://sixdemonbag.org/Djinni.
1798 * Improve merge join performance by allowing mark/restore of
1801 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1803 * Consider using a hash for joining to a large IN (VALUES ...) list
1805 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1812 * Consider having the background writer update the transaction status
1813 hint bits before writing out the page
1815 Implementing this requires the background writer to have access to system
1816 catalogs and the transaction status log.
1818 * Consider adding buffers the background writer finds reusable to the
1821 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1823 * Automatically tune bgwriter_delay based on activity rather then using a
1826 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1828 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1830 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1832 * Test to see if calling PreallocXlogFiles() from the background writer
1833 will help with WAL segment creation latency
1835 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1839 Miscellaneous Performance
1840 =========================
1842 * Do async I/O for faster random read-ahead of data
1844 Async I/O allows multiple I/O requests to be sent to the disk with
1845 results coming back asynchronously.
1847 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1848 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1849 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1850 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1852 * Use mmap() rather than SYSV shared memory or to write WAL files?
1854 This would remove the requirement for SYSV SHM but would introduce
1855 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1856 to prevent I/O overhead.
1858 * Consider mmap()'ing files into a backend?
1860 Doing I/O to large tables would consume a lot of address space or
1861 require frequent mapping/unmapping. Extending the file also causes
1862 mapping problems that might require mapping only individual pages,
1863 leading to thousands of mappings. Another problem is that there is no
1864 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1865 could hit disk before WAL is written.
1867 * Add a script to ask system configuration questions and tune postgresql.conf
1868 * Consider ways of storing rows more compactly on disk
1870 o Reduce the row header size?
1871 o Consider reducing on-disk varlena length from four bytes to
1872 two because a heap row cannot be more than 64k in length
1874 * Consider increasing NUM_CLOG_BUFFERS
1876 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1877 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1879 * Consider transaction start/end performance improvements
1881 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00948.php
1882 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1884 * Allow user configuration of TOAST thresholds
1886 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1887 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1889 * Allow configuration of backend priorities via the operating system
1891 Though backend priorities make priority inversion during lock
1892 waits possible, research shows that this is not a huge problem.
1894 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1896 * Experiment with multi-threaded backend better I/O utilization
1898 This would allow a single query to make use of multiple I/O channels
1899 simultaneously. One idea is to create a background reader that can
1900 pre-fetch sequential and index scan pages needed by other backends.
1901 This could be expanded to allow concurrent reads from multiple devices
1902 in a partitioned table.
1904 * Experiment with multi-threaded backend better CPU utilization
1906 This would allow several CPUs to be used for a single query, such as
1907 for sorting or query execution.
1909 * Consider increasing the minimum allowed number of shared buffers
1911 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1913 * Expire published xmin for read-only and idle transactions
1915 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1917 * Consider if CommandCounterIncrement() can avoid its
1918 AcceptInvalidationMessages() call
1920 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1922 * Improve performance of shared invalidation queue for multiple CPUs
1924 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1926 * Consider Cartesian joins when both relations are needed to form an
1927 indexscan qualification for a third relation
1929 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1931 * Consider not storing a NULL bitmap on disk if all the NULLs are
1934 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1935 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1937 * Sort large UPDATE/DELETEs so it is done in heap order
1939 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1941 * SMP scalability improvements
1943 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1944 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1945 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1947 * Research reducing deTOASTing in more places
1949 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1951 * Allow one transaction to see tuples using the snapshot of another
1954 This would assist multiple backends in working together.
1955 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00400.php
1957 * Consider decreasing the I/O caused by updating tuple hint bits
1959 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00847.php
1960 http://archives.postgresql.org/pgsql-patches/2008-07/msg00199.php
1967 * Add use of 'const' for variables in source tree
1968 * Move some things from contrib into main tree
1969 * %Remove warnings created by -Wcast-align
1970 * Move platform-specific ps status display info from ps_status.c to ports
1971 * Add optional CRC checksum to heap and index pages
1972 * Improve documentation to build only interfaces (Marc)
1973 * Remove or relicense modules that are not under the BSD license, if possible
1974 * Acquire lock on a relation before building a relcache entry for it
1975 * Allow cross-compiling by generating the zic database on the target system
1976 * Improve NLS maintenance of libpgport messages linked onto applications
1977 * Clean up casting in contrib/isn
1979 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1981 * Improve the /contrib installation experience
1983 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00132.php
1985 * Use UTF8 encoding for NLS messages so all server encodings can
1987 * Update Bonjour to work with newer cross-platform SDK
1989 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1990 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1992 * Consider GnuTLS if OpenSSL license becomes a problem
1994 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1995 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1997 * Consider changing documentation format from SGML to XML
1999 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
2001 * Consider making NAMEDATALEN more configurable in future releases
2002 * Update our code to handle 64-bit timezone files to match the zic
2003 source code, which now uses them
2004 * Have configure choose integer datetimes by default
2006 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
2008 * Support scoped IPv6 addresses
2010 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
2012 * Consider allowing 64-bit integers and floats to be passed by value on
2015 Also change 32-bit floats (float4) to be passed by value at the same
2018 * Research use of signals and sleep wake ups
2020 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
2022 * Add automated check for invalid C++ source code constructs
2024 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
2026 * Consider simplifying how memory context resets handle child contexts
2028 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
2030 * Remove use of MAKE_PTR and MAKE_OFFSET macros
2032 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
2034 * Convert single quotes to apostrophes in the PDF documentation
2036 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
2038 * Create three versions of libpgport to simplify client code
2040 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00154.php
2042 * Remove old-style routines for manipulating tuples
2044 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00851.php
2046 * Improve detection of shared memory segments being used by others
2047 by checking the SysV shared memory field 'nattch'
2049 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00656.php
2050 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00673.php
2052 * Implement the non-threaded Avahi service discovery protocol
2054 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00939.php
2055 http://archives.postgresql.org/pgsql-patches/2008-02/msg00097.php
2056 http://archives.postgresql.org/pgsql-hackers/2008-03/msg01211.php
2057 http://archives.postgresql.org/pgsql-patches/2008-04/msg00001.php
2059 * Implement a module capability for loading /contrib-style extensions
2061 http://archives.postgresql.org/pgsql-patches/2008-04/msg00164.php
2063 * Fix system views like pg_stat_all_tables to use set-returning
2064 functions, rather than views of per-column functions
2066 * Allow table and index WITH options to be specified via hooks, for use
2067 with plugins like GiST index methods
2071 o Remove configure.in check for link failure when cause is found
2072 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
2074 o -Remove psql newline patch when we find out why mingw outputs an
2076 o Allow psql to use readline once non-US code pages work with
2078 o Fix problem with shared memory on the Win32 Terminal Server
2079 o Diagnose problem where shared memory can sometimes not be
2080 attached by postmaster children
2082 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
2084 o Improve signal handling
2086 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
2088 o Convert MSVC build system to remove most batch files
2090 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
2092 o Prevent SSL from sending network packets to avoid interference
2093 with Win32 signal emulation
2095 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
2097 o Support pgxs when using MSVC
2099 o Fix MSVC NLS support, like for to_char()
2101 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
2102 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
2104 o Find a correct rint() substitute on Windows
2106 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00808.php
2110 * Wire Protocol Changes
2112 o Allow dynamic character set handling
2113 o Add decoded type, length, precision
2115 o Update clients to use data types, typmod, schema.table.column names
2116 of result sets using new statement protocol
2122 * Add pre-parsing phase that converts non-ISO syntax to supported
2125 This could allow SQL written for other databases to run without
2128 * Allow plug-in modules to emulate features from other databases
2129 * Add features of Oracle-style packages (Pavel)
2131 A package would be a schema with session-local variables,
2132 public/private functions, and initialization functions. It
2133 is also possible to implement these capabilities
2134 in any schema and not use a separate "packages"
2137 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
2139 * Consider allowing control of upper/lower case folding of unquoted
2142 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
2143 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
2144 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00849.php
2145 http://archives.postgresql.org/pgsql-hackers/2008-07/msg00415.php
2147 * Add autonomous transactions
2149 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
2153 Features We Do _Not_ Want
2154 =========================
2156 * All backends running as threads in a single process (not wanted)
2158 This eliminates the process protection we get from the current setup.
2159 Thread creation is usually the same overhead as process creation on
2160 modern systems, so it seems unwise to use a pure threaded model.
2162 * Optimizer hints (not wanted)
2164 Optimizer hints are used to work around problems in the optimizer. We
2165 would rather have the problems reported and fixed.
2167 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
2168 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
2169 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
2172 Because we support postfix operators, it isn't possible to make AS
2173 optional and continue to use bison.
2174 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
2176 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
2178 * Embedded server (not wanted)
2180 While PostgreSQL clients runs fine in limited-resource environments, the
2181 server requires multiple processes and a stable pool of resources to
2182 run reliabily and efficiently. Stripping down the PostgreSQL server
2183 to run in the same process address space as the client application
2184 would add too much complexity and failure cases.