3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Wed May 28 21:58:24 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 variable classes that can restrict who can set the values
67 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
69 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
70 the privilege granted by the invoking role, and not those granted
73 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
75 * Allow SSL authentication/encryption over unix domain sockets
77 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
79 * Allow SSL key file permission checks to be optionally disabled when
80 sharing SSL keys with other applications
82 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00069.php
84 * Allow client certificate names to be checked against the client
87 This is already implemented in
88 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
93 o Allow pg_hba.conf to specify host names along with IP addresses
95 Host name lookup could occur when the postmaster reads the
96 pg_hba.conf file, or when the backend starts. Another
97 solution would be to reverse lookup the connection IP and
98 check that hostname against the host names in pg_hba.conf.
99 We could also then check that the host name maps to the IP
102 o %Allow postgresql.conf file values to be changed via an SQL
103 API, perhaps using SET GLOBAL
104 o Allow the server to be stopped/restarted via an SQL API
105 o Issue a warning if a change-on-restart-only postgresql.conf value
106 is modified and the server config files are reloaded
107 o Consider normalizing fractions in postgresql.conf, perhaps
110 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00550.php
112 o Allow Kerberos to disable stripping of realms so we can
113 check the username@realm against multiple realms
115 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00009.php
120 o Allow a database in tablespace t1 with tables created in
121 tablespace t2 to be used as a template for a new database created
122 with default tablespace t2
124 Currently all objects in the default database tablespace must
125 have default tablespace specifications. This is because new
126 databases are created by copying directories. If you mix default
127 tablespace tables and tablespace-specified tables in the same
128 directory, creating a new database from such a mixed directory
129 would create a new database with tables that had incorrect
130 explicit tablespaces. To fix this would require modifying
131 pg_class in the newly copied database, which we don't currently
134 o Allow reporting of which objects are in which tablespaces
136 This item is difficult because a tablespace can contain objects
137 from multiple databases. There is a server-side function that
138 returns the databases which use a specific tablespace, so this
139 requires a tool that will call that function and connect to each
140 database to find the objects in each database for that tablespace.
142 o Allow WAL replay of CREATE TABLESPACE to work when the directory
143 structure on the recovery computer is different from the original
145 o Allow per-tablespace quotas
148 * Statistics Collector
150 o Allow statistics collector information to be pulled from the collector
151 process directly, rather than requiring the collector to write a
152 filesystem file twice a second?
153 o Reduce file system activity overhead of statistics file pgstat.stat
155 http://archives.postgresql.org/pgsql-general/2007-12/msg00106.php
157 o Allow statistics last vacuum/analyze execution times to be displayed
158 without requiring stats_row_level to be enabled
160 http://archives.postgresql.org/pgsql-docs/2007-04/msg00028.php
162 o Clear table counters on TRUNCATE
164 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00169.php
167 * Point-In-Time Recovery (PITR)
169 o Allow a warm standby system to also allow read-only statements
171 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
173 o %Create dump tool for write-ahead logs for use in determining
174 transaction id for point-in-time recovery
176 This is useful for checking PITR recovery.
178 o Allow recovery.conf to support the same syntax as
179 postgresql.conf, including quoting
181 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
183 o Fix server restart problem when the server was shutdown during
186 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00800.php
188 o Recreate pg_xlog/archive_status/ if it doesn't exist after
189 restoring from a PITR backup
191 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00487.php
197 * Change NUMERIC to enforce the maximum precision
198 * Reduce storage space for small NUMERICs
200 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
201 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
202 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
204 * Fix data types where equality comparison isn't intuitive, e.g. box
205 * Add support for public SYNONYMs
207 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
209 * Fix CREATE CAST on DOMAINs
211 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
212 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
214 * Allow domains to be cast
216 http://archives.postgresql.org/pgsql-hackers/2003-06/msg01206.php
217 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00289.php
219 * Add support for SQL-standard GENERATED/IDENTITY columns
221 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
222 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
223 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
224 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
225 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
227 * Improve XML support
229 http://developer.postgresql.org/index.php/XML_Support
231 * Consider placing all sequences in a single table, or create a system
234 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
236 * Allow the UUID type to accept non-standard formats
238 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
240 * Consider a special data type for regular expressions
242 http://archives.postgresql.org/pgsql-hackers/2007-08/msg01067.php
244 * Reduce BIT data type overhead using short varlena headers
246 http://archives.postgresql.org/pgsql-general/2007-12/msg00273.php
248 * Allow xml arrays to be cast to other data types
250 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00981.php
251 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00231.php
252 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00471.php
254 * Simplify integer cross-data-type operators
256 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00189.php
258 * Allow XML to accept more liberal DOCTYPE specifications
260 http://archives.postgresql.org/pgsql-general/2008-02/msg00347.php
262 * Allow adding/renaming/removing enumerated values to an existing
265 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01718.php
269 o Allow infinite dates and intervals just like infinite timestamps
270 o Merge hardwired timezone names with the TZ database; allow either
271 kind everywhere a TZ name is currently taken
272 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
273 information, either zone name or offset from UTC
275 If the TIMESTAMP value is stored with a time zone name, interval
276 computations should adjust based on the time zone rules.
277 http://archives.postgresql.org/pgsql-hackers/2004-10/msg00705.php
279 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
280 o Add a GUC variable to allow output of interval values in ISO8601
282 o Have timestamp subtraction not call justify_hours()?
284 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
286 o Improve timestamptz subtraction to be DST-aware
288 Currently subtracting one date from another that crosses a
289 daylight savings time adjustment can return '1 day 1 hour', but
290 adding that back to the first date returns a time one hour in
291 the future. This is caused by the adjustment of '25 hours' to
292 '1 day 1 hour', and '1 day' is the same time the next day, even
293 if daylight savings adjustments are involved.
295 o Fix interval display to support values exceeding 2^31 hours
296 o Add overflow checking to timestamp and interval arithmetic
297 o Extend timezone code to allow 64-bit values so we can
298 represent years beyond 2038
300 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
302 o -Use LC_TIME for localized weekday/month names, rather than
305 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
307 o Add ISO INTERVAL handling
309 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
310 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
312 o Support ISO INTERVAL syntax if units cannot be determined from
313 the string, and are supplied after the string
315 The SQL standard states that the units after the string
316 specify the units of the string, e.g. INTERVAL '2' MINUTE
317 should return '00:02:00'. The current behavior has the units
318 restrict the interval value to the specified unit or unit
319 range, INTERVAL '70' SECOND returns '00:00:10'.
321 For syntax that isn't uniquely ISO or PG syntax, like '1' or
322 '1:30', treat as ISO if there is a range specification clause,
323 and as PG if there no clause is present, e.g. interpret '1:30'
324 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
325 '1:30' as '1 hour, 30 minutes'.
327 This makes common cases like SELECT INTERVAL '1' MONTH
328 SQL-standard results. The SQL standard supports a limited
329 number of unit combinations and doesn't support unit names in
330 the string. The PostgreSQL syntax is more flexible in the
331 range of units supported, e.g. PostgreSQL supports '1 year 1
332 hour', while the SQL standard does not.
334 o Add support for year-month syntax, INTERVAL '50-6' YEAR
336 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
337 year' AS INTERVAL MONTH), and this should return '12 months'
338 o Round or truncate values to the requested precision, e.g.
339 INTERVAL '11 months' AS YEAR should return one or zero
340 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
345 o Delay resolution of array expression's data type so assignment
346 coercion can be performed on empty array expressions
347 o Add support for arrays of domains
349 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
351 o Allow single-byte header storage for array elements
356 o Improve vacuum of large objects, like contrib/vacuumlo?
357 o Add security checking for large objects
358 o Auto-delete large objects when referencing row is deleted
360 contrib/lo offers this functionality.
362 o Allow read/write into TOAST values like large objects
364 This requires the TOAST column to be stored EXTERNAL.
366 o Add API for 64-bit large object access
368 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
372 * Add locale-aware MONEY type, and support multiple currencies
374 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
375 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
377 * MONEY dumps in a locale-specific format making it difficult to
378 restore to a system with a different locale
379 * Allow MONEY to be easily cast to/from other numeric data types
383 o Allow dictionaries to change the token that is passed on to
386 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
388 o Consider a function-based API for '@@' searches
390 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
392 o Improve text search error messages
394 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00966.php
395 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01146.php
397 o Consider changing error to warning for strings larger than one
400 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00190.php
401 http://archives.postgresql.org/pgsql-patches/2008-03/msg00062.php
408 * Allow INET subnet tests using non-constants to be indexed
409 * Allow to_date() and to_timestamp() accept localized month names
410 * Fix to_date()-related functions to consistently issue errors
412 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
414 * Add missing parameter handling in to_char()
416 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
418 * Allow substring/replace() to get/set bit values
419 * Allow to_char() on interval values to accumulate the highest unit
422 Some special format flag would be required to request such
423 accumulation. Such functionality could also be added to EXTRACT.
424 Prevent accumulation that crosses the month/day boundary because of
425 the uneven number of days in a month.
427 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
428 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
429 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
430 o to_char(INTERVAL '3 years 5 months','MM') => 41
432 * Implement inlining of set-returning functions defined in SQL
433 * Allow SQL-language functions to return results from RETURNING queries
435 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
437 * Allow SQL-language functions to reference parameters by parameter name
439 Currently SQL-language functions can only refer to dollar parameters,
442 * Add SPI_gettypmod() to return the typemod for a TupleDesc
443 * Enforce typmod for function inputs, function results and parameters for
444 spi_prepare'd statements called from PLs
446 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
448 * Allow holdable cursors in SPI
449 * Tighten function permission checks
451 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
453 * Fix IS OF so it matches the ISO specification, and add documentation
455 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
456 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
458 * Add missing operators for geometric data types
460 Some geometric types do not have the full suite of geometric operators,
463 * Implement Boyer-Moore searching in strpos()
465 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
467 * Prevent malicious functions from being executed with the permissions
468 of unsuspecting users
470 Index functions are safe, so VACUUM and ANALYZE are safe too.
471 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
472 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
474 * Reduce memory usage of aggregates in set returning functions
476 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
478 * Add temporal versions of generate_series()
480 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
482 * Add array_accum() and array_to_set() functions for arrays
484 The standards specify array_agg() and UNNEST.
485 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
487 * Fix /contrib/ltree operator
489 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
491 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
493 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
495 * Fix regular expression bug when using complex back-references
497 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00000.php
499 * Have /contrib/dblink reuse unnamed connections
501 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00895.php
503 * Add SQL-standard array_agg() and unnest() array functions
505 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01017.php
507 * Allow calling of a procedure outside a SELECT that can control the
510 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01375.php
514 Multi-Language Support
515 ======================
517 * Add NCHAR (as distinguished from ordinary varchar),
518 * Allow locale to be set at database creation
520 Currently locale can only be set during initdb. No global tables have
521 locale-aware columns. However, the database template used during
522 database creation might have locale-aware indexes. The indexes would
523 need to be reindexed to match the new locale.
525 * Allow encoding on a per-column basis optionally using the ICU library;
528 Right now only one encoding is allowed per database.
530 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
531 http://archives.postgresql.org/pgsql-patches/2005-08/msg00039.php
532 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
533 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00110.php
534 http://archives.postgresql.org/pgsql-patches/2005-09/msg00020.php
535 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01121.php
536 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00767.php
537 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
538 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
539 http://wiki.postgresql.org/wiki/Todo:Collate
540 http://wiki.postgresql.org/wiki/Todo:ICU
542 * Support multiple simultaneous character sets, per SQL92
543 * Improve UTF8 combined character handling?
544 * Add octet_length_server() and octet_length_client()
545 * Make octet_length_client() the same as octet_length()?
546 * Fix problems with wrong runtime encoding conversion for NLS message files
547 * Add URL to more complete multi-byte regression tests
549 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
551 * Fix ILIKE and regular expressions to handle case insensitivity
552 properly in multibyte encodings
554 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
555 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
557 * Set client encoding based on the client operating system encoding
559 Currently client_encoding is set in postgresql.conf, which
560 defaults to the server encoding.
561 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
563 * Change memory allocation for multi-byte functions so memory is
564 allocated inside conversion functions
566 Currently we preallocate memory based on worst-case usage.
573 * Automatically create rules on views so they are updateable, per SQL99
575 We can only auto-create rules for simple views. For more complex
576 cases users will still have to write rules manually.
578 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
579 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
581 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
582 * Allow VIEW/RULE recompilation when the underlying tables change
584 Another issue is whether underlying table changes should be reflected
585 in the view, e.g. should SELECT * show additional columns if they
586 are added after the view is created.
588 * Make it possible to use RETURNING together with conditional DO INSTEAD
589 rules, such as for partitioning setups
591 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
593 * Add the ability to automatically create materialized views
595 Right now materialized views require the user to create triggers on the
596 main table to keep the summary table current. SQL syntax should be able
597 to manager the triggers and summary table automatically. A more
598 sophisticated implementation would automatically retrieve from the
599 summary table when the main table is referenced, if possible.
606 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
607 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
608 * %Allow SET CONSTRAINTS to be qualified by schema/table name
609 * %Add a separate TRUNCATE permission
611 Currently only the owner can TRUNCATE a table because triggers are not
612 called, and the table is locked in exclusive mode.
614 * Allow PREPARE of cursors
615 * Allow finer control over the caching of prepared query plans
617 Currently queries prepared via the libpq API are planned on first
618 execute using the supplied parameters --- allow SQL PREPARE to do the
619 same. Also, allow control over replanning prepared queries either
620 manually or automatically when statistics for execute parameters
621 differ dramatically from those used during planning.
623 * Improve logging of prepared transactions recovered during startup
625 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
627 * Improve failure message when DROP DATABASE is used on a database that
628 has prepared transactions
629 * Allow prepared transactions with temporary tables created and dropped
630 in the same transaction, and when an ON COMMIT DELETE ROWS temporary
633 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00047.php
635 * Add a GUC variable to warn about non-standard SQL usage in queries
636 * Add SQL-standard MERGE/REPLACE/UPSERT command
638 MERGE is typically used to merge two tables. REPLACE or UPSERT
639 command does UPDATE, or on failure, INSERT. This is similar to UPDATE,
640 then for unmatched rows, INSERT. Whether concurrent access allows
641 modifications which could cause row loss is implementation independent.
642 To implement this cleanly requires that the table have a unique index
643 so duplicate checking can be easily performed. It is possible to do it
644 without a unique index if we require the user to LOCK the table before
647 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
648 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
649 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01475.php
651 * Add NOVICE output level for helpful messages like automatic sequence/index
653 * Add GUC to issue notice about statements that use unjoined tables
654 * Allow EXPLAIN to identify tables that were skipped because of
656 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
657 * Enable standard_conforming_strings
658 * Make standard_conforming_strings the default in 8.5?
660 When this is done, backslash-quote should be prohibited in non-E''
661 strings because of possible confusion over how such strings treat
662 backslashes. Basically, '' is always safe for a literal single
663 quote, while \' might or might not be based on the backslash
666 * Simplify dropping roles that have objects in several databases
667 * Allow COMMENT ON to accept an expression rather than just a string
668 * Allow the count returned by SELECT, etc to be represented as an int64
669 to allow a higher range of values
670 * Add SQL99 WITH clause to SELECT
671 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
673 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
674 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
675 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
676 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
677 http://archives.postgresql.org/pgsql-patches/2008-01/msg00105.php
678 http://archives.postgresql.org/pgsql-patches/2008-03/msg00327.php
680 * Add DEFAULT .. AS OWNER so permission checks are done as the table
683 This would be useful for SERIAL nextval() calls and CHECK constraints.
685 * Allow DISTINCT to work in multiple-argument aggregate calls
686 * Add column to pg_stat_activity that shows the progress of long-running
687 commands like CREATE INDEX and VACUUM
689 http://archives.postgresql.org/pgsql-patches/2008-04/msg00203.php
691 * Implement SQL:2003 window functions
692 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
694 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
695 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
697 * Increase locking when DROPing objects so dependent objects cannot
698 get dropped while the DROP operation is happening
700 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
702 * -Allow AS in "SELECT col AS label" to be optional in certain cases
704 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
705 RETURNING to supply values to the INSERT
706 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
708 * Add comments on system tables/columns using the information in
711 Ideally the information would be pulled from the SGML file
714 * Improve reporting of UNION type mismatches
716 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
717 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
722 o Allow CREATE TABLE AS to determine column lengths for complex
723 expressions like SELECT col1 || col2
725 o Have WITH CONSTRAINTS also create constraint indexes
727 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
729 o Have CONSTRAINT cname NOT NULL record the contraint name
731 Right now pg_attribute.attnotnull records the NOT NULL status
732 of the column, but does not record the contraint name
734 o Prevent concurrent CREATE TABLE table1 from sometimes returning
735 a cryptic error message
737 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00169.php
739 o Add CREATE SCHEMA ... LIKE that copies a schema
744 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
746 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
747 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
748 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
749 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
751 o Research self-referential UPDATEs that see inconsistent row versions
752 in read-committed mode
754 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
755 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
757 o Allow GLOBAL temporary tables to exist as empty by default in
760 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
765 o Have ALTER TABLE RENAME rename SERIAL sequence names
767 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
769 o Have ALTER SEQUENCE RENAME rename the sequence name stored
770 in the sequence table
772 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
773 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
774 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
776 o Add ALTER DOMAIN to modify the underlying data type
777 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
779 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
781 o %Allow ALTER TABLE to change constraint deferrability and actions
782 o Add missing object types for ALTER ... SET SCHEMA
783 o Allow ALTER TABLESPACE to move to different directories
784 o Allow databases to be moved to different tablespaces
785 o Allow moving system tables to other tablespaces, where possible
787 Currently non-global system tables must be in the default database
788 tablespace. Global system tables can never be moved.
790 o -Prevent parent tables from altering or dropping constraints
791 like CHECK that are inherited by child tables unless CASCADE
793 o -Prevent child tables from altering or dropping constraints
794 like CHECK that were inherited from the parent table
795 o Have ALTER INDEX update the name of a constraint using that index
796 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
797 o Allow column display reordering by recording a display,
798 storage, and permanent id for every column?
800 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
802 o Allow an existing index to be marked as a table's primary key
804 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00500.php
809 o Automatically maintain clustering on a table
811 This might require some background daemon to maintain clustering
812 during periods of low usage. It might also require tables to be only
813 partially filled for easier reorganization. Another idea would
814 be to create a merged heap/index data file so an index lookup would
815 automatically access the heap data too. A third idea would be to
816 store heap rows in hashed groups, perhaps using a user-supplied
818 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
820 o %Add default clustering to system tables
822 To do this, determine the ideal cluster index for each system
823 table and set the cluster setting during initdb.
825 o %Add VERBOSE option to report tables as they are processed,
831 o Allow COPY to report error lines and continue
833 This requires the use of a savepoint before each COPY line is
834 processed, with ROLLBACK on COPY failure.
835 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
837 o Allow COPY on a newly-created table to skip WAL logging
839 On crash recovery, the table involved in the COPY would
840 be removed or have its heap and index files truncated. One
841 issue is that no other backend should be able to add to
842 the table at the same time, which is something that is
843 currently allowed. This currently is done if the table is
844 created inside the same transaction block as the COPY because
845 no other backends can see the table.
847 o Consider using a ring buffer for COPY FROM
849 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
850 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
852 o Allow COPY FROM to create index entries in bulk
854 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
856 o Allow COPY in CSV mode to control whether a quoted zero-length
857 string is treated as NULL
859 Currently this is always treated as a zero-length string,
860 which generates an error when loading into an integer column
861 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
863 o Impove COPY performance
865 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00954.php
871 o Allow column-level privileges
872 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
875 The proposed syntax is:
876 GRANT SELECT ON ALL TABLES IN public TO phpuser;
877 GRANT SELECT ON NEW TABLES IN public TO phpuser;
879 o Allow GRANT/REVOKE permissions to be inherited by objects based on
882 o Allow SERIAL sequences to inherit permissions from the base table?
887 o Prevent DROP TABLE from dropping a row referenced by its own open
893 o Allow INSERT/UPDATE of the system-generated oid value for a row
894 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
900 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
906 o Allow LISTEN/NOTIFY to store info in memory rather than tables?
908 Currently LISTEN/NOTIFY information is stored in pg_listener.
909 Storing such information in memory would improve performance.
911 o Add optional textual message to NOTIFY
913 This would allow an informational message to be added to the notify
914 message, perhaps indicating the row modified or other custom
917 o Allow multiple identical NOTIFY events to always be communicated
918 to the client, rather than sent as a single notification to the
921 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
923 o Allow NOTIFY in rules involving conditionals
924 o Improve LISTEN concurrency
926 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01106.php
930 Referential Integrity
931 =====================
933 * Add MATCH PARTIAL referential integrity
934 * Change foreign key constraint for array -> element to mean element
936 * Fix problem when cascading referential triggers make changes on
937 cascaded tables, seeing the tables in an intermediate state
939 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
940 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
942 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
944 This would allow UPDATE tab SET col = col + 1 to work if col has
945 a unique index. Currently, uniqueness checks are done while the
946 command is being executed, rather than at the end of the statement
948 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
949 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
951 * Optimize referential integrity checks
953 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
954 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
957 Server-Side Languages
958 =====================
961 o Fix RENAME to work on variables other than OLD/NEW
963 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
964 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
965 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
967 o Allow function parameters to be passed by name,
968 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
969 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
970 o Allow listing of record column names, and access to
971 record columns via variables, e.g. columns := r.(*),
974 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
975 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
976 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
978 o Add support for SCROLL cursors
979 o Add support for WITH HOLD cursors
980 o Allow row and record variables to be set to NULL constants,
981 and allow NULL tests on such variables
983 Because a row is not scalar, do not allow assignment
984 from NULL-valued scalars.
986 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
988 o Review handling of MOVE and FETCH
990 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
992 o Improve logic of determining if an identifier is a a
993 variable or column name
995 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
997 o Consider keeping seperate cached copies when search_path changes
999 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01009.php
1001 o -Add CASE capability to language (already in SQL)
1003 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00696.php
1009 o Add table function support to pltcl, plpythonu
1010 o Add support for polymorphic arguments and return types to
1011 languages other than PL/PgSQL
1012 o Add capability to create and call PROCEDURES
1013 o Add support for OUT and INOUT parameters to languages other
1015 o Add PL/PythonU tracebacks
1017 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
1019 o Allow data to be passed in native language formats, rather
1022 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
1024 o Add ability to obfuscate function bodies
1026 http://archives.postgresql.org/pgsql-patches/2008-01/msg00125.php
1033 * Have pg_ctl look at PGHOST in case it is a socket directory?
1034 * Allow pg_ctl to work properly with configuration files located outside
1035 the PGDATA directory
1037 pg_ctl can not read the pid file because it isn't located in the
1038 config directory but in the PGDATA directory. The solution is to
1039 allow pg_ctl to read and understand postgresql.conf to find the
1040 data_directory value.
1042 * Add a function like pg_get_indexdef() that report more detailed index
1045 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
1048 o Have psql show current values for a sequence
1049 o Move psql backslash database information into the backend, use
1052 This would allow non-psql clients to pull the same information out
1053 of the database as psql.
1054 http://archives.postgresql.org/pgsql-hackers/2004-01/msg00191.php
1056 o Make psql's \d commands more consistent
1058 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1059 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1061 o Consistently display privilege information for all objects in psql
1062 o Add auto-expanded mode so expanded output is used if the row
1063 length is wider than the screen width.
1065 Consider using auto-expanded mode for backslash commands like \df+.
1067 o Prevent tab completion of SET TRANSACTION from querying the
1068 database and therefore preventing the transaction isolation
1069 level from being set.
1071 Currently SET <tab> causes a database lookup to check all
1072 supported session variables. This query causes problems
1073 because setting the transaction isolation level must be the
1074 first statement of a transaction.
1076 o Add a \set variable to control whether \s displays line numbers
1078 Another option is to add \# which lists line numbers, and
1079 allows command execution.
1081 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
1083 o Prevent escape string warnings when object names have
1086 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1088 o Have \d show foreign keys that reference a table's primary key
1090 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1092 o Have \d show child tables that inherit from the specified parent
1093 o -Have \l+ show database size, if permissions allow
1095 Ideally it will not generate an error for invalid permissions
1097 o Include the symbolic SQLSTATE name in verbose error reports
1099 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1101 o -Improve display of enums to show valid enum values
1102 o Add prompt escape to display the client and server versions
1105 * pg_dump / pg_restore
1106 o %Add dumping of comments on index columns and composite type columns
1107 o %Add full object name to the tag field. eg. for operators we need
1108 '=(integer, integer)', instead of just '='.
1109 o Add pg_dumpall custom format dumps?
1110 o Allow selection of individual object(s) of all types, not just
1112 o In a selective dump, allow dumping of an object and all its
1114 o Add options like pg_restore -l and -L to pg_dump
1115 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1116 o Allow pg_dump --clean to drop roles that own objects or have
1118 o Change pg_dump so that a comment on the dumped database is
1119 applied to the loaded database, even if the database has a
1120 different name. This will require new backend syntax, perhaps
1121 COMMENT ON CURRENT DATABASE.
1122 o Remove unnecessary function pointer abstractions in pg_dump source
1124 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1125 multiple objects simultaneously
1127 The difficulty with this is getting multiple dump processes to
1128 produce a single dump output file. It also would require
1129 several sessions to share the same snapshot.
1130 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1132 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1133 restoring multiple objects simultaneously
1135 This might require a pg_restore flag to indicate how many
1136 simultaneous operations should be performed. Only pg_dump's
1137 -Fc format has the necessary dependency information.
1138 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1140 o To better utilize resources, allow pg_restore to check foreign
1141 keys simultaneously, where possible
1142 o Allow pg_restore to create all indexes of a table
1143 concurrently, via a single heap scan
1145 This requires a pg_dump -Fc file because that format contains
1146 the required dependency information.
1147 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1149 o Allow pg_restore to load different parts of the COPY data
1151 o Prevent pg_dump/pg_restore from being affected by
1154 Using psql to restore a pg_dump dump is also affected.
1156 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1157 o Allow pre/data/post files when schema and data are dumped
1158 separately, for performance reasons
1160 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1167 Document differences between ecpg and the SQL standard and
1168 information about the Informix-compatibility module.
1170 o Solve cardinality > 1 for input descriptors / variables?
1171 o Add a semantic check level, e.g. check if a table really exists
1172 o fix handling of DB attributes that are arrays
1173 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1175 o Fix nested C comments
1176 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1177 o Make SET CONNECTION thread-aware, non-standard?
1178 o Allow multidimensional arrays
1179 o Add internationalized message strings
1180 o Implement COPY FROM STDIN
1184 o Add PQescapeIdentifierConn()
1185 o Prevent PQfnumber() from lowercasing unquoted the column name
1187 PQfnumber() should never have been doing lowercasing, but
1188 historically it has so we need a way to prevent it
1190 o Allow statement results to be automatically batched to the client
1192 Currently all statement results are transferred to the libpq
1193 client before libpq makes the results available to the
1194 application. This feature would allow the application to make
1195 use of the first result rows while the rest are transferred, or
1196 held on the server waiting for them to be requested by libpq.
1197 One complexity is that a statement like SELECT 1/col could error
1198 out mid-way through the result set.
1200 o Consider disallowing multiple queries in PQexec() as an
1201 additional barrier to SQL injection attacks
1203 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1205 o Add PQexecf() that allows complex parameter substitution
1207 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1209 o Add SQLSTATE severity to PGconn return status
1211 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1217 * Add deferred trigger queue file
1219 Right now all deferred trigger information is stored in backend
1220 memory. This could exhaust memory for very large trigger queues.
1221 This item involves dumping large queues into files.
1222 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00876.php
1224 * Allow triggers to be disabled in only the current session.
1226 This is currently possible by starting a multi-statement transaction,
1227 modifying the system tables, performing the desired SQL, restoring the
1228 system tables, and committing the transaction. ALTER TABLE ...
1229 TRIGGER requires a table lock so it is not ideal for this usage.
1231 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1233 If the dump is known to be valid, allow foreign keys to be added
1234 without revalidating the data.
1236 * Allow statement-level triggers to access modified rows
1237 * Support triggers on columns
1239 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1241 * Allow AFTER triggers on system tables
1243 System tables are modified in many places in the backend without going
1244 through the executor and therefore not causing triggers to fire. To
1245 complete this item, the functions that modify system tables will have
1248 * Tighten trigger permission checks
1250 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1252 * Allow BEFORE INSERT triggers on views
1254 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1256 * -Add ability to trigger on TRUNCATE
1257 * Add database and transaction-level triggers
1259 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00451.php
1266 * Add UNIQUE capability to non-btree indexes
1267 * Prevent index uniqueness checks when UPDATE does not modify the column
1269 Uniqueness (index) checks are done when updating a column even if the
1270 column is not modified by the UPDATE.
1272 * Allow the creation of on-disk bitmap indexes which can be quickly
1273 combined with other bitmap indexes
1275 Such indexes could be more compact if there are only a few distinct values.
1276 Such indexes can also be compressed. Keeping such indexes updated can be
1279 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1280 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1281 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00265.php
1282 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01214.php
1283 http://archives.postgresql.org/pgsql-patches/2007-05/msg00013.php
1284 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00741.php
1286 * Allow accurate statistics to be collected on indexes with more than
1287 one column or expression indexes, perhaps using per-index statistics
1289 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1290 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1292 * Consider increasing the default and maximum number of statistics targets,
1293 and reduce statistics target overhead
1295 Also consider having a larger statistics target for indexed columns
1296 and expression indexes.
1297 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1298 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1299 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1300 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00188.php
1302 * Consider smaller indexes that record a range of values per heap page,
1303 rather than having one index entry for every heap row
1305 This is useful if the heap is clustered by the indexed values.
1306 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1307 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1308 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1309 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1310 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00014.php
1311 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00487.php
1312 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01589.php
1314 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1316 This is difficult because you must upgrade to an exclusive table lock
1317 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1318 have this complication. This would allow index compaction without
1320 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1322 * Allow multiple indexes to be created concurrently, ideally via a
1323 single heap scan, and have pg_restore use it
1325 * Consider sorting entries before inserting into btree index
1327 http://archives.postgresql.org/pgsql-general/2008-01/msg01010.php
1329 * Allow index scans to return matching index keys, not just the matching
1332 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01657.php
1338 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1339 and primary/foreign keys
1340 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1341 on inherited table, e.g. INSERT INTO inherit_table
1342 (unique_index_col) VALUES (dup) should fail
1344 The main difficulty with this item is the problem of
1345 creating an index that can span multiple tables.
1347 o Allow SELECT ... FOR UPDATE on inherited tables
1348 o Require all CHECK constraints to be inherited
1350 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1352 o Add checks to prevent a CREATE RULE views on inherited tables
1354 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1355 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1360 o Add more GIST index support for geometric data types
1361 o Allow GIST indexes to create certain complex index types, like
1362 digital trees (see Aoki)
1367 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1369 o Pack hash index buckets onto disk pages more efficiently
1371 Currently only one hash bucket can be stored on a page. Ideally
1372 several hash buckets could be stored on a single page and greater
1373 granularity used for the hash algorithm.
1375 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1377 o Consider sorting hash buckets so entries can be found using a
1378 binary search, rather than a linear scan
1379 o In hash indexes, consider storing the hash value with or instead
1381 o Add WAL logging for crash recovery
1382 o Allow multi-column hash indexes
1383 o -During index creation, pre-sort the tuples to improve build speed
1390 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1392 Ideally this requires a separate test program that can be run
1393 at initdb time or optionally later. Consider O_SYNC when
1396 * Add program to test if fsync has a delay compared to non-fsync
1397 * Consider sorting writes during checkpoint
1399 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1408 We could use a fixed row count and a +/- count to follow MVCC
1409 visibility rules, or a single cached value could be used and
1410 invalidated if anyone modifies the table. Another idea is to
1411 get a count directly from a unique index, but for this to be
1412 faster than a sequential scan it must avoid access to the heap
1413 to obtain tuple visibility information.
1415 * Provide a way to calculate an "estimated COUNT(*)"
1417 Perhaps by using the optimizer's cardinality estimates or random
1420 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1422 * Allow data to be pulled directly from indexes
1424 Currently indexes do not have enough tuple visibility information
1425 to allow data to be pulled from the index without also accessing
1426 the heap. One way to allow this is to set a bit on index tuples
1427 to indicate if a tuple is currently visible to all transactions
1428 when the first valid heap lookup happens. This bit would have to
1429 be cleared when a heap tuple is expired.
1431 Another idea is to maintain a bitmap of heap pages where all rows
1432 are visible to all backends, and allow index lookups to reference
1433 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1434 add someday to determine which heap pages need vacuuming. Frequently
1435 accessed bitmaps would have to be stored in shared memory. One 8k
1436 page of bitmaps could track 512MB of heap pages.
1438 A third idea would be for a heap scan to check if all rows are visible
1439 and if so set a per-table flag which can be checked by index scans.
1440 Any change to the table would have to clear the flag. To detect
1441 changes during the heap scan a counter could be set at the start and
1442 checked at the end --- if it is the same, the table has not been
1443 modified --- any table change would increment the counter.
1445 http://archives.postgresql.org/pgsql-patches/2007-10/msg00166.php
1446 http://archives.postgresql.org/pgsql-patches/2008-01/msg00049.php
1448 * Consider automatic caching of statements at various levels:
1451 o Query execute plan
1453 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00823.php
1455 * Consider increasing internal areas when shared buffers is increased
1457 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1459 * Consider decreasing the amount of memory used by PrivateRefCount
1461 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1462 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1464 * Consider allowing higher priority queries to have referenced buffer
1465 cache pages stay in memory longer
1467 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1474 * Improve speed with indexes
1476 For large table adjustments during VACUUM FULL, it is faster to cluster
1477 or reindex rather than update the index. Also, index updates can bloat
1480 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1481 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1482 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1484 * Auto-fill the free space map by scanning the buffer cache or by
1485 checking pages written by the background writer
1487 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1488 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1490 * Create a bitmap of pages that need vacuuming
1492 Instead of sequentially scanning the entire table, have the background
1493 writer or some other process record pages that have expired rows, then
1494 VACUUM can look at just those pages rather than the entire table. In
1495 the event of a system crash, the bitmap would probably be invalidated.
1496 One complexity is that index entries still have to be vacuumed, and
1497 doing this without an index scan (by using the heap values to find the
1498 index entry) might be slow and unreliable, especially for user-defined
1501 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1502 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1503 http://archives.postgresql.org/pgsql-patches/2007-03/msg00508.php
1504 http://archives.postgresql.org/pgsql-patches/2007-04/msg00347.php
1505 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00156.php
1506 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00546.php
1507 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00416.php
1509 * Allow FSM to return free space toward the beginning of the heap file,
1510 in hopes that empty pages at the end can be truncated by VACUUM
1511 * Allow FSM page return free space based on table clustering, to assist
1512 in maintaining clustering?
1513 * -Improve dead row detection during multi-statement transactions usage
1515 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1517 * Consider a more compact data representation for dead tuples
1519 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1524 o %Issue log message to suggest VACUUM FULL if a table is nearly
1526 o Improve control of auto-vacuum
1528 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1530 o Prevent long-lived temporary tables from causing frozen-xid
1531 advancement starvation
1533 The problem is that autovacuum cannot vacuum them to set frozen xids;
1534 only the session that created them can do that.
1535 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1537 o Store per-table autovacuum settings in pg_class.reloptions.
1539 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1540 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1542 o Prevent autovacuum from running if an old transaction is still
1543 running from the last vacuum
1545 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1552 * Fix priority ordering of read and write light-weight locks (Neil)
1554 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1555 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1557 * Fix problem when multiple subtransactions of the same outer transaction
1558 hold different types of locks, and one subtransaction aborts
1560 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1561 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1562 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1563 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1565 * Allow UPDATEs on only non-referential integrity columns not to conflict
1566 with referential integrity locks
1568 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1570 * Add idle_in_transaction_timeout GUC so locks are not held for long
1573 * Improve deadlock detection when a page cleaning lock conflicts
1574 with a shared buffer that is pinned
1576 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1577 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1578 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1580 * Detect deadlocks involving LockBufferForCleanup()
1582 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1586 Startup Time Improvements
1587 =========================
1589 * Experiment with multi-threaded backend for backend creation
1591 This would prevent the overhead associated with process creation. Most
1592 operating systems have trivial process creation time compared to
1593 database startup overhead, but a few operating systems (Win32,
1594 Solaris) might benefit from threading. Also explore the idea of
1595 a single session using multiple threads to execute a statement faster.
1602 * Eliminate need to write full pages to WAL before page modification
1604 Currently, to protect against partial disk page writes, we write
1605 full page images to WAL before they are modified so we can correct any
1606 partial page writes during recovery. These pages can also be
1607 eliminated from point-in-time archive files.
1608 http://archives.postgresql.org/pgsql-hackers/2002-06/msg00655.php
1610 o When off, write CRC to WAL and check file system blocks
1613 If CRC check fails during recovery, remember the page in case
1614 a later CRC for that page properly matches.
1616 o Write full pages during file system write and not when
1617 the page is modified in the buffer cache
1619 This allows most full page writes to happen in the background
1620 writer. It might cause problems for applying WAL on recovery
1621 into a partially-written page, but later the full page will be
1624 * Allow WAL traffic to be streamed to another server for stand-by
1626 * Reduce WAL traffic so only modified values are written rather than
1629 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1631 * Allow WAL information to recover corrupted pg_controldata
1633 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1635 * Find a way to reduce rotational delay when repeatedly writing
1638 Currently fsync of WAL requires the disk platter to perform a full
1639 rotation to fsync again. One idea is to write the WAL to different
1640 offsets that might reduce the rotational delay.
1641 http://archives.postgresql.org/pgsql-hackers/2002-11/msg00483.php
1643 * Allow WAL logging to be turned off for a table, but the table
1644 might be dropped or truncated during crash recovery
1646 Allow tables to bypass WAL writes and just fsync() dirty pages on
1647 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1648 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1649 non-default logging should not use referential integrity with
1650 default-logging tables. A table without dirty buffers during a
1651 crash could perhaps avoid the drop/truncate.
1652 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1654 * Allow WAL logging to be turned off for a table, but the table would
1655 avoid being truncated/dropped
1657 To do this, only a single writer can modify the table, and writes
1658 must happen only on new pages so the new pages can be removed during
1659 crash recovery. Readers can continue accessing the table. Such
1660 tables probably cannot have indexes. One complexity is the handling
1661 of indexes on TOAST tables.
1662 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1664 * Speed WAL recovery by allowing more than one page to be prefetched
1666 This should be done utilizing the same infrastructure used for
1667 prefetching in general to avoid introducing complex error-prone code
1669 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1670 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1671 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1673 * Improve WAL concurrency by increasing lock granularity
1675 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1677 * Be more aggressive about creating WAL files
1679 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01325.php
1680 http://archives.postgresql.org/pgsql-hackers/2004-07/msg01075.php
1681 http://archives.postgresql.org/pgsql-hackers/2005-04/msg00556.php
1683 * Have resource managers report the duration of their status changes
1685 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01468.php
1687 * Move pgfoundry's xlogdump to /contrib and have it rely more closely
1688 on the WAL backend code
1690 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00035.php
1695 Optimizer / Executor
1696 ====================
1698 * Improve selectivity functions for geometric operators
1699 * Precompile SQL functions to avoid overhead
1700 * Create utility to compute accurate random_page_cost value
1701 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1702 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1703 actual row counts differ by a specified percentage
1704 * Improve how ANALYZE computes in-doubt tuples
1706 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00771.php
1708 * Consider using hash buckets to do DISTINCT, rather than sorting
1710 This would be beneficial when there are few distinct values. This is
1711 already used by GROUP BY.
1713 * Log statements where the optimizer row estimates were dramatically
1714 different from the number of rows actually found?
1715 * Consider compressed annealing to search for query plans
1717 This might replace GEQO, http://sixdemonbag.org/Djinni.
1719 * Improve merge join performance by allowing mark/restore of
1722 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1724 * Consider using a hash for joining to a large IN (VALUES ...) list
1726 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1733 * Consider having the background writer update the transaction status
1734 hint bits before writing out the page
1736 Implementing this requires the background writer to have access to system
1737 catalogs and the transaction status log.
1739 * Consider adding buffers the background writer finds reusable to the
1742 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1744 * Automatically tune bgwriter_delay based on activity rather then using a
1747 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1749 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1751 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1753 * Test to see if calling PreallocXlogFiles() from the background writer
1754 will help with WAL segment creation latency
1756 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1760 Miscellaneous Performance
1761 =========================
1763 * Do async I/O for faster random read-ahead of data
1765 Async I/O allows multiple I/O requests to be sent to the disk with
1766 results coming back asynchronously.
1768 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1769 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1770 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1771 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1773 * Use mmap() rather than SYSV shared memory or to write WAL files?
1775 This would remove the requirement for SYSV SHM but would introduce
1776 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1777 to prevent I/O overhead.
1779 * Consider mmap()'ing files into a backend?
1781 Doing I/O to large tables would consume a lot of address space or
1782 require frequent mapping/unmapping. Extending the file also causes
1783 mapping problems that might require mapping only individual pages,
1784 leading to thousands of mappings. Another problem is that there is no
1785 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1786 could hit disk before WAL is written.
1788 * Add a script to ask system configuration questions and tune postgresql.conf
1789 * Consider ways of storing rows more compactly on disk
1791 o Reduce the row header size?
1792 o Consider reducing on-disk varlena length from four bytes to
1793 two because a heap row cannot be more than 64k in length
1795 * Consider increasing NUM_CLOG_BUFFERS
1797 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1798 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1800 * Consider transaction start/end performance improvements
1802 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00948.php
1803 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1805 * Allow user configuration of TOAST thresholds
1807 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1808 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1810 * Allow configuration of backend priorities via the operating system
1812 Though backend priorities make priority inversion during lock
1813 waits possible, research shows that this is not a huge problem.
1815 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1817 * Experiment with multi-threaded backend better I/O utilization
1819 This would allow a single query to make use of multiple I/O channels
1820 simultaneously. One idea is to create a background reader that can
1821 pre-fetch sequential and index scan pages needed by other backends.
1822 This could be expanded to allow concurrent reads from multiple devices
1823 in a partitioned table.
1825 * Experiment with multi-threaded backend better CPU utilization
1827 This would allow several CPUs to be used for a single query, such as
1828 for sorting or query execution.
1830 * Consider increasing the minimum allowed number of shared buffers
1832 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1834 * Expire published xmin for read-only and idle transactions
1836 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1838 * Consider if CommandCounterIncrement() can avoid its
1839 AcceptInvalidationMessages() call
1841 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1843 * Improve performance of shared invalidation queue for multiple CPUs
1845 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1847 * Consider Cartesian joins when both relations are needed to form an
1848 indexscan qualification for a third relation
1850 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1852 * Consider not storing a NULL bitmap on disk if all the NULLs are
1855 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1856 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1858 * Sort large UPDATE/DELETEs so it is done in heap order
1860 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1862 * -Avoid tuple some tuple copying in sort routines
1863 * SMP scalability improvements
1865 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1866 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1867 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1869 * Research reducing deTOASTing in more places
1871 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1873 * Consider being smarter about memory and external files used during
1876 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01101.php
1877 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00045.php
1879 * Allow one transaction to see tuples using the snapshot of another
1882 This would assist multiple backends in working together.
1883 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00400.php
1889 * Add use of 'const' for variables in source tree
1890 * Move some things from contrib into main tree
1891 * %Remove warnings created by -Wcast-align
1892 * Move platform-specific ps status display info from ps_status.c to ports
1893 * Add optional CRC checksum to heap and index pages
1894 * Improve documentation to build only interfaces (Marc)
1895 * Remove or relicense modules that are not under the BSD license, if possible
1896 * Acquire lock on a relation before building a relcache entry for it
1897 * Allow cross-compiling by generating the zic database on the target system
1898 * Improve NLS maintenance of libpgport messages linked onto applications
1899 * Clean up casting in contrib/isn
1901 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1903 * Improve the /contrib installation experience
1905 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00132.php
1907 * Use UTF8 encoding for NLS messages so all server encodings can
1909 * Update Bonjour to work with newer cross-platform SDK
1911 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1912 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1914 * Consider detoasting keys before sorting
1915 * Consider GnuTLS if OpenSSL license becomes a problem
1917 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1918 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1920 * Consider changing documentation format from SGML to XML
1922 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1924 * Consider making NAMEDATALEN more configurable in future releases
1925 * Update our code to handle 64-bit timezone files to match the zic
1926 source code, which now uses them
1927 * Have configure choose integer datetimes by default
1929 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1931 * Support scoped IPv6 addresses
1933 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1935 * Consider allowing 64-bit integers and floats to be passed by value on
1938 Also change 32-bit floats (float4) to be passed by value at the same
1941 * Research use of signals and sleep wake ups
1943 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1945 * Add automated check for invalid C++ source code constructs
1947 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1949 * Consider simplifying how memory context resets handle child contexts
1951 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1953 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1955 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1957 * Convert single quotes to apostrophes in the PDF documentation
1959 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
1961 * Create three versions of libpgport to simplify client code
1963 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00154.php
1965 * Remove old-style routines for manipulating tuples
1967 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00851.php
1969 * Improve detection of shared memory segments being used by others
1970 by checking the SysV shared memory field 'nattch'
1972 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00656.php
1973 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00673.php
1975 * Implement the non-threaded Avahi service discovery protocol
1977 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00939.php
1978 http://archives.postgresql.org/pgsql-patches/2008-02/msg00097.php
1979 http://archives.postgresql.org/pgsql-hackers/2008-03/msg01211.php
1980 http://archives.postgresql.org/pgsql-patches/2008-04/msg00001.php
1984 o Remove configure.in check for link failure when cause is found
1985 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1987 o Remove psql newline patch when we find out why mingw outputs an
1989 o Allow psql to use readline once non-US code pages work with
1991 o Fix problem with shared memory on the Win32 Terminal Server
1992 o Diagnose problem where shared memory can sometimes not be
1993 attached by postmaster children
1995 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
1997 o Improve signal handling
1999 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
2001 o Convert MSVC build system to remove most batch files
2003 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
2005 o Prevent SSL from sending network packets to avoid interference
2006 with Win32 signal emulation
2008 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
2010 o Support pgxs when using MSVC
2012 o Fix MSVC NLS support, like for to_char()
2014 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
2015 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
2017 o Find a correct rint() substitute on Windows
2019 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00808.php
2023 * Wire Protocol Changes
2025 o Allow dynamic character set handling
2026 o Add decoded type, length, precision
2028 o Update clients to use data types, typmod, schema.table.column names
2029 of result sets using new statement protocol
2035 * Add pre-parsing phase that converts non-ISO syntax to supported
2038 This could allow SQL written for other databases to run without
2041 * Allow plug-in modules to emulate features from other databases
2042 * Add features of Oracle-style packages (Pavel)
2044 A package would be a schema with session-local variables,
2045 public/private functions, and initialization functions. It
2046 is also possible to implement these capabilities
2047 in any schema and not use a separate "packages"
2050 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
2052 * Consider allowing control of upper/lower case folding of unquoted
2055 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
2056 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
2057 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00849.php
2059 * Add autonomous transactions
2061 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
2065 Features We Do _Not_ Want
2066 =========================
2068 * All backends running as threads in a single process (not wanted)
2070 This eliminates the process protection we get from the current setup.
2071 Thread creation is usually the same overhead as process creation on
2072 modern systems, so it seems unwise to use a pure threaded model.
2074 * Optimizer hints (not wanted)
2076 Optimizer hints are used to work around problems in the optimizer. We
2077 would rather have the problems reported and fixed.
2079 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
2080 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
2081 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
2084 Because we support postfix operators, it isn't possible to make AS
2085 optional and continue to use bison.
2086 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
2088 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
2090 * Embedded server (not wanted)
2092 While PostgreSQL clients runs fine in limited-resource environments, the
2093 server requires multiple processes and a stable pool of resources to
2094 run reliabily and efficiently. Stripping down the PostgreSQL server
2095 to run in the same process address space as the client application
2096 would add too much complexity and failure cases.