3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Fri Jun 13 12:24:28 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
117 o Add functions to syntax check configuration files
122 o Allow a database in tablespace t1 with tables created in
123 tablespace t2 to be used as a template for a new database created
124 with default tablespace t2
126 Currently all objects in the default database tablespace must
127 have default tablespace specifications. This is because new
128 databases are created by copying directories. If you mix default
129 tablespace tables and tablespace-specified tables in the same
130 directory, creating a new database from such a mixed directory
131 would create a new database with tables that had incorrect
132 explicit tablespaces. To fix this would require modifying
133 pg_class in the newly copied database, which we don't currently
136 o Allow reporting of which objects are in which tablespaces
138 This item is difficult because a tablespace can contain objects
139 from multiple databases. There is a server-side function that
140 returns the databases which use a specific tablespace, so this
141 requires a tool that will call that function and connect to each
142 database to find the objects in each database for that tablespace.
144 o Allow WAL replay of CREATE TABLESPACE to work when the directory
145 structure on the recovery computer is different from the original
147 o Allow per-tablespace quotas
150 * Statistics Collector
152 o Allow statistics collector information to be pulled from the collector
153 process directly, rather than requiring the collector to write a
154 filesystem file twice a second?
155 o Reduce file system activity overhead of statistics file pgstat.stat
157 http://archives.postgresql.org/pgsql-general/2007-12/msg00106.php
159 o Allow statistics last vacuum/analyze execution times to be displayed
160 without requiring stats_row_level to be enabled
162 http://archives.postgresql.org/pgsql-docs/2007-04/msg00028.php
164 o Clear table counters on TRUNCATE
166 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00169.php
169 * Point-In-Time Recovery (PITR)
171 o Allow a warm standby system to also allow read-only statements
173 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
175 o %Create dump tool for write-ahead logs for use in determining
176 transaction id for point-in-time recovery
178 This is useful for checking PITR recovery.
180 o Allow recovery.conf to support the same syntax as
181 postgresql.conf, including quoting
183 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
185 o Fix server restart problem when the server was shutdown during
188 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00800.php
190 o Recreate pg_xlog/archive_status/ if it doesn't exist after
191 restoring from a PITR backup
193 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00487.php
199 * Change NUMERIC to enforce the maximum precision
200 * Reduce storage space for small NUMERICs
202 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
203 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
204 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
206 * Fix data types where equality comparison isn't intuitive, e.g. box
207 * Add support for public SYNONYMs
209 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
211 * Fix CREATE CAST on DOMAINs
213 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
214 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
216 * Allow domains to be cast
218 http://archives.postgresql.org/pgsql-hackers/2003-06/msg01206.php
219 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00289.php
221 * Add support for SQL-standard GENERATED/IDENTITY columns
223 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
224 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
225 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
226 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
227 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
229 * Improve XML support
231 http://developer.postgresql.org/index.php/XML_Support
233 * Consider placing all sequences in a single table, or create a system
236 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
238 * Allow the UUID type to accept non-standard formats
240 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
242 * Consider a special data type for regular expressions
244 http://archives.postgresql.org/pgsql-hackers/2007-08/msg01067.php
246 * Reduce BIT data type overhead using short varlena headers
248 http://archives.postgresql.org/pgsql-general/2007-12/msg00273.php
250 * Allow xml arrays to be cast to other data types
252 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00981.php
253 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00231.php
254 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00471.php
256 * Simplify integer cross-data-type operators
258 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00189.php
260 * Allow XML to accept more liberal DOCTYPE specifications
262 http://archives.postgresql.org/pgsql-general/2008-02/msg00347.php
264 * Allow adding/renaming/removing enumerated values to an existing
267 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01718.php
271 o Allow infinite dates and intervals just like infinite timestamps
272 o Merge hardwired timezone names with the TZ database; allow either
273 kind everywhere a TZ name is currently taken
274 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
275 information, either zone name or offset from UTC
277 If the TIMESTAMP value is stored with a time zone name, interval
278 computations should adjust based on the time zone rules.
279 http://archives.postgresql.org/pgsql-hackers/2004-10/msg00705.php
281 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
282 o Add a GUC variable to allow output of interval values in ISO8601
284 o Have timestamp subtraction not call justify_hours()?
286 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
288 o Improve timestamptz subtraction to be DST-aware
290 Currently subtracting one date from another that crosses a
291 daylight savings time adjustment can return '1 day 1 hour', but
292 adding that back to the first date returns a time one hour in
293 the future. This is caused by the adjustment of '25 hours' to
294 '1 day 1 hour', and '1 day' is the same time the next day, even
295 if daylight savings adjustments are involved.
297 o Fix interval display to support values exceeding 2^31 hours
298 o Add overflow checking to timestamp and interval arithmetic
299 o Extend timezone code to allow 64-bit values so we can
300 represent years beyond 2038
302 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
304 o -Use LC_TIME for localized weekday/month names, rather than
307 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
309 o Add ISO INTERVAL handling
311 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
312 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
314 o Support ISO INTERVAL syntax if units cannot be determined from
315 the string, and are supplied after the string
317 The SQL standard states that the units after the string
318 specify the units of the string, e.g. INTERVAL '2' MINUTE
319 should return '00:02:00'. The current behavior has the units
320 restrict the interval value to the specified unit or unit
321 range, INTERVAL '70' SECOND returns '00:00:10'.
323 For syntax that isn't uniquely ISO or PG syntax, like '1' or
324 '1:30', treat as ISO if there is a range specification clause,
325 and as PG if there no clause is present, e.g. interpret '1:30'
326 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
327 '1:30' as '1 hour, 30 minutes'.
329 This makes common cases like SELECT INTERVAL '1' MONTH
330 SQL-standard results. The SQL standard supports a limited
331 number of unit combinations and doesn't support unit names in
332 the string. The PostgreSQL syntax is more flexible in the
333 range of units supported, e.g. PostgreSQL supports '1 year 1
334 hour', while the SQL standard does not.
336 o Add support for year-month syntax, INTERVAL '50-6' YEAR
338 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
339 year' AS INTERVAL MONTH), and this should return '12 months'
340 o Round or truncate values to the requested precision, e.g.
341 INTERVAL '11 months' AS YEAR should return one or zero
342 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
347 o Delay resolution of array expression's data type so assignment
348 coercion can be performed on empty array expressions
349 o Add support for arrays of domains
351 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
353 o Allow single-byte header storage for array elements
358 o Improve vacuum of large objects, like contrib/vacuumlo?
359 o Add security checking for large objects
360 o Auto-delete large objects when referencing row is deleted
362 contrib/lo offers this functionality.
364 o Allow read/write into TOAST values like large objects
366 This requires the TOAST column to be stored EXTERNAL.
368 o Add API for 64-bit large object access
370 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
374 * Add locale-aware MONEY type, and support multiple currencies
376 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
377 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
379 * MONEY dumps in a locale-specific format making it difficult to
380 restore to a system with a different locale
381 * Allow MONEY to be easily cast to/from other numeric data types
385 o Allow dictionaries to change the token that is passed on to
388 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
390 o Consider a function-based API for '@@' searches
392 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
394 o Improve text search error messages
396 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00966.php
397 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01146.php
399 o Consider changing error to warning for strings larger than one
402 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00190.php
403 http://archives.postgresql.org/pgsql-patches/2008-03/msg00062.php
410 * Allow INET subnet tests using non-constants to be indexed
411 * Allow to_date() and to_timestamp() accept localized month names
412 * Fix to_date()-related functions to consistently issue errors
414 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
416 * Add missing parameter handling in to_char()
418 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
420 * Allow substring/replace() to get/set bit values
421 * Allow to_char() on interval values to accumulate the highest unit
424 Some special format flag would be required to request such
425 accumulation. Such functionality could also be added to EXTRACT.
426 Prevent accumulation that crosses the month/day boundary because of
427 the uneven number of days in a month.
429 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
430 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
431 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
432 o to_char(INTERVAL '3 years 5 months','MM') => 41
434 * Implement inlining of set-returning functions defined in SQL
435 * Allow SQL-language functions to return results from RETURNING queries
437 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
439 * Allow SQL-language functions to reference parameters by parameter name
441 Currently SQL-language functions can only refer to dollar parameters,
444 * Add SPI_gettypmod() to return the typemod for a TupleDesc
445 * Enforce typmod for function inputs, function results and parameters for
446 spi_prepare'd statements called from PLs
448 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
450 * Allow holdable cursors in SPI
451 * Tighten function permission checks
453 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
455 * Fix IS OF so it matches the ISO specification, and add documentation
457 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
458 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
460 * Add missing operators for geometric data types
462 Some geometric types do not have the full suite of geometric operators,
465 * Implement Boyer-Moore searching in strpos()
467 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
469 * Prevent malicious functions from being executed with the permissions
470 of unsuspecting users
472 Index functions are safe, so VACUUM and ANALYZE are safe too.
473 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
474 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
476 * Reduce memory usage of aggregates in set returning functions
478 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
480 * Add temporal versions of generate_series()
482 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
484 * Add array_accum() and array_to_set() functions for arrays
486 The standards specify array_agg() and UNNEST.
487 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
489 * Fix /contrib/ltree operator
491 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
493 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
495 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
497 * Fix regular expression bug when using complex back-references
499 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00000.php
501 * Have /contrib/dblink reuse unnamed connections
503 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00895.php
505 * Add SQL-standard array_agg() and unnest() array functions
507 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01017.php
509 * Allow calling of a procedure outside a SELECT that can control the
512 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01375.php
516 Multi-Language Support
517 ======================
519 * Add NCHAR (as distinguished from ordinary varchar),
520 * Allow locale to be set at database creation
522 Currently locale can only be set during initdb. No global tables have
523 locale-aware columns. However, the database template used during
524 database creation might have locale-aware indexes. The indexes would
525 need to be reindexed to match the new locale.
527 * Allow encoding on a per-column basis optionally using the ICU library;
530 Right now only one encoding is allowed per database.
532 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
533 http://archives.postgresql.org/pgsql-patches/2005-08/msg00039.php
534 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
535 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00110.php
536 http://archives.postgresql.org/pgsql-patches/2005-09/msg00020.php
537 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01121.php
538 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00767.php
539 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
540 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
541 http://wiki.postgresql.org/wiki/Todo:Collate
542 http://wiki.postgresql.org/wiki/Todo:ICU
544 * Support multiple simultaneous character sets, per SQL92
545 * Improve UTF8 combined character handling?
546 * Add octet_length_server() and octet_length_client()
547 * Make octet_length_client() the same as octet_length()?
548 * Fix problems with wrong runtime encoding conversion for NLS message files
549 * Add URL to more complete multi-byte regression tests
551 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
553 * Fix ILIKE and regular expressions to handle case insensitivity
554 properly in multibyte encodings
556 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
557 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
559 * Set client encoding based on the client operating system encoding
561 Currently client_encoding is set in postgresql.conf, which
562 defaults to the server encoding.
563 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
565 * Change memory allocation for multi-byte functions so memory is
566 allocated inside conversion functions
568 Currently we preallocate memory based on worst-case usage.
575 * Automatically create rules on views so they are updateable, per SQL99
577 We can only auto-create rules for simple views. For more complex
578 cases users will still have to write rules manually.
580 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
581 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
583 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
584 * Allow VIEW/RULE recompilation when the underlying tables change
586 Another issue is whether underlying table changes should be reflected
587 in the view, e.g. should SELECT * show additional columns if they
588 are added after the view is created.
590 * Make it possible to use RETURNING together with conditional DO INSTEAD
591 rules, such as for partitioning setups
593 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
595 * Add the ability to automatically create materialized views
597 Right now materialized views require the user to create triggers on the
598 main table to keep the summary table current. SQL syntax should be able
599 to manager the triggers and summary table automatically. A more
600 sophisticated implementation would automatically retrieve from the
601 summary table when the main table is referenced, if possible.
608 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
609 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
610 * %Allow SET CONSTRAINTS to be qualified by schema/table name
611 * %Add a separate TRUNCATE permission
613 Currently only the owner can TRUNCATE a table because triggers are not
614 called, and the table is locked in exclusive mode.
616 * Allow PREPARE of cursors
617 * Allow finer control over the caching of prepared query plans
619 Currently queries prepared via the libpq API are planned on first
620 execute using the supplied parameters --- allow SQL PREPARE to do the
621 same. Also, allow control over replanning prepared queries either
622 manually or automatically when statistics for execute parameters
623 differ dramatically from those used during planning.
625 * Improve logging of prepared transactions recovered during startup
627 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
629 * Improve failure message when DROP DATABASE is used on a database that
630 has prepared transactions
631 * Allow prepared transactions with temporary tables created and dropped
632 in the same transaction, and when an ON COMMIT DELETE ROWS temporary
635 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00047.php
637 * Add a GUC variable to warn about non-standard SQL usage in queries
638 * Add SQL-standard MERGE/REPLACE/UPSERT command
640 MERGE is typically used to merge two tables. REPLACE or UPSERT
641 command does UPDATE, or on failure, INSERT. This is similar to UPDATE,
642 then for unmatched rows, INSERT. Whether concurrent access allows
643 modifications which could cause row loss is implementation independent.
644 To implement this cleanly requires that the table have a unique index
645 so duplicate checking can be easily performed. It is possible to do it
646 without a unique index if we require the user to LOCK the table before
649 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
650 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
651 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01475.php
653 * Add NOVICE output level for helpful messages like automatic sequence/index
655 * Add GUC to issue notice about statements that use unjoined tables
656 * Allow EXPLAIN to identify tables that were skipped because of
658 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
659 * Enable standard_conforming_strings
660 * Make standard_conforming_strings the default in 8.5?
662 When this is done, backslash-quote should be prohibited in non-E''
663 strings because of possible confusion over how such strings treat
664 backslashes. Basically, '' is always safe for a literal single
665 quote, while \' might or might not be based on the backslash
668 * Simplify dropping roles that have objects in several databases
669 * Allow COMMENT ON to accept an expression rather than just a string
670 * Allow the count returned by SELECT, etc to be represented as an int64
671 to allow a higher range of values
672 * Add SQL99 WITH clause to SELECT
673 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
675 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
676 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
677 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
678 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
679 http://archives.postgresql.org/pgsql-patches/2008-01/msg00105.php
680 http://archives.postgresql.org/pgsql-patches/2008-03/msg00327.php
682 * Add DEFAULT .. AS OWNER so permission checks are done as the table
685 This would be useful for SERIAL nextval() calls and CHECK constraints.
687 * Allow DISTINCT to work in multiple-argument aggregate calls
688 * Add column to pg_stat_activity that shows the progress of long-running
689 commands like CREATE INDEX and VACUUM
691 http://archives.postgresql.org/pgsql-patches/2008-04/msg00203.php
693 * Implement SQL:2003 window functions
694 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause or
697 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
698 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
699 http://archives.postgresql.org/pgsql-hackers/2008-06/msg00124.php
701 * Increase locking when DROPing objects so dependent objects cannot
702 get dropped while the DROP operation is happening
704 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
706 * -Allow AS in "SELECT col AS label" to be optional in certain cases
708 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
709 RETURNING to supply values to the INSERT
710 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
712 * Add comments on system tables/columns using the information in
715 Ideally the information would be pulled from the SGML file
718 * Improve reporting of UNION type mismatches
720 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
721 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
726 o Allow CREATE TABLE AS to determine column lengths for complex
727 expressions like SELECT col1 || col2
729 o Have WITH CONSTRAINTS also create constraint indexes
731 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
733 o Have CONSTRAINT cname NOT NULL record the contraint name
735 Right now pg_attribute.attnotnull records the NOT NULL status
736 of the column, but does not record the contraint name
738 o Prevent concurrent CREATE TABLE table1 from sometimes returning
739 a cryptic error message
741 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00169.php
743 o Add CREATE SCHEMA ... LIKE that copies a schema
748 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
750 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
751 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
752 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
753 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
755 o Research self-referential UPDATEs that see inconsistent row versions
756 in read-committed mode
758 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
759 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
761 o Allow GLOBAL temporary tables to exist as empty by default in
764 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
769 o Have ALTER TABLE RENAME rename SERIAL sequence names
771 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
773 o Have ALTER SEQUENCE RENAME rename the sequence name stored
774 in the sequence table
776 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
777 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
778 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
780 o Add ALTER DOMAIN to modify the underlying data type
781 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
783 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
785 o %Allow ALTER TABLE to change constraint deferrability and actions
786 o Add missing object types for ALTER ... SET SCHEMA
787 o Allow ALTER TABLESPACE to move to different directories
788 o Allow databases to be moved to different tablespaces
789 o Allow moving system tables to other tablespaces, where possible
791 Currently non-global system tables must be in the default database
792 tablespace. Global system tables can never be moved.
794 o -Prevent parent tables from altering or dropping constraints
795 like CHECK that are inherited by child tables unless CASCADE
797 o -Prevent child tables from altering or dropping constraints
798 like CHECK that were inherited from the parent table
799 o Have ALTER INDEX update the name of a constraint using that index
800 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
801 o Allow column display reordering by recording a display,
802 storage, and permanent id for every column?
804 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
806 o Allow an existing index to be marked as a table's primary key
808 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00500.php
813 o Automatically maintain clustering on a table
815 This might require some background daemon to maintain clustering
816 during periods of low usage. It might also require tables to be only
817 partially filled for easier reorganization. Another idea would
818 be to create a merged heap/index data file so an index lookup would
819 automatically access the heap data too. A third idea would be to
820 store heap rows in hashed groups, perhaps using a user-supplied
822 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
824 o %Add default clustering to system tables
826 To do this, determine the ideal cluster index for each system
827 table and set the cluster setting during initdb.
829 o %Add VERBOSE option to report tables as they are processed,
835 o Allow COPY to report error lines and continue
837 This requires the use of a savepoint before each COPY line is
838 processed, with ROLLBACK on COPY failure.
839 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
841 o Allow COPY on a newly-created table to skip WAL logging
843 On crash recovery, the table involved in the COPY would
844 be removed or have its heap and index files truncated. One
845 issue is that no other backend should be able to add to
846 the table at the same time, which is something that is
847 currently allowed. This currently is done if the table is
848 created inside the same transaction block as the COPY because
849 no other backends can see the table.
851 o Consider using a ring buffer for COPY FROM
853 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
854 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
856 o Allow COPY FROM to create index entries in bulk
858 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
860 o Allow COPY in CSV mode to control whether a quoted zero-length
861 string is treated as NULL
863 Currently this is always treated as a zero-length string,
864 which generates an error when loading into an integer column
865 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
867 o Impove COPY performance
869 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00954.php
875 o Allow column-level privileges
876 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
879 The proposed syntax is:
880 GRANT SELECT ON ALL TABLES IN public TO phpuser;
881 GRANT SELECT ON NEW TABLES IN public TO phpuser;
883 o Allow GRANT/REVOKE permissions to be inherited by objects based on
886 o Allow SERIAL sequences to inherit permissions from the base table?
891 o Prevent DROP TABLE from dropping a row referenced by its own open
897 o Allow INSERT/UPDATE of the system-generated oid value for a row
898 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
904 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
910 o Allow LISTEN/NOTIFY to store info in memory rather than tables?
912 Currently LISTEN/NOTIFY information is stored in pg_listener.
913 Storing such information in memory would improve performance.
915 o Add optional textual message to NOTIFY
917 This would allow an informational message to be added to the notify
918 message, perhaps indicating the row modified or other custom
921 o Allow multiple identical NOTIFY events to always be communicated
922 to the client, rather than sent as a single notification to the
925 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
927 o Allow NOTIFY in rules involving conditionals
928 o Improve LISTEN concurrency
930 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01106.php
934 Referential Integrity
935 =====================
937 * Add MATCH PARTIAL referential integrity
938 * Change foreign key constraint for array -> element to mean element
940 * Fix problem when cascading referential triggers make changes on
941 cascaded tables, seeing the tables in an intermediate state
943 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
944 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
946 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
948 This would allow UPDATE tab SET col = col + 1 to work if col has
949 a unique index. Currently, uniqueness checks are done while the
950 command is being executed, rather than at the end of the statement
952 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
953 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
955 * Optimize referential integrity checks
957 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
958 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
961 Server-Side Languages
962 =====================
965 o Fix RENAME to work on variables other than OLD/NEW
967 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
968 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
969 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
971 o Allow function parameters to be passed by name,
972 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
973 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
974 o Allow listing of record column names, and access to
975 record columns via variables, e.g. columns := r.(*),
978 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
979 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
980 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
982 o Add support for SCROLL cursors
983 o Add support for WITH HOLD cursors
984 o Allow row and record variables to be set to NULL constants,
985 and allow NULL tests on such variables
987 Because a row is not scalar, do not allow assignment
988 from NULL-valued scalars.
990 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
992 o Review handling of MOVE and FETCH
994 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
996 o Improve logic of determining if an identifier is a a
997 variable or column name
999 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
1001 o Consider keeping seperate cached copies when search_path changes
1003 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01009.php
1005 o -Add CASE capability to language (already in SQL)
1007 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00696.php
1013 o Add table function support to pltcl, plpythonu
1014 o Add support for polymorphic arguments and return types to
1015 languages other than PL/PgSQL
1016 o Add capability to create and call PROCEDURES
1017 o Add support for OUT and INOUT parameters to languages other
1019 o Add PL/PythonU tracebacks
1021 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
1023 o Allow data to be passed in native language formats, rather
1026 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
1028 o Add ability to obfuscate function bodies
1030 http://archives.postgresql.org/pgsql-patches/2008-01/msg00125.php
1037 * Have pg_ctl look at PGHOST in case it is a socket directory?
1038 * Allow pg_ctl to work properly with configuration files located outside
1039 the PGDATA directory
1041 pg_ctl can not read the pid file because it isn't located in the
1042 config directory but in the PGDATA directory. The solution is to
1043 allow pg_ctl to read and understand postgresql.conf to find the
1044 data_directory value.
1046 * Add a function like pg_get_indexdef() that report more detailed index
1049 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
1052 o Have psql show current values for a sequence
1053 o Move psql backslash database information into the backend, use
1056 This would allow non-psql clients to pull the same information out
1057 of the database as psql.
1058 http://archives.postgresql.org/pgsql-hackers/2004-01/msg00191.php
1060 o Make psql's \d commands more consistent
1062 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1063 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1065 o Consistently display privilege information for all objects in psql
1066 o Add auto-expanded mode so expanded output is used if the row
1067 length is wider than the screen width.
1069 Consider using auto-expanded mode for backslash commands like \df+.
1071 o Prevent tab completion of SET TRANSACTION from querying the
1072 database and therefore preventing the transaction isolation
1073 level from being set.
1075 Currently SET <tab> causes a database lookup to check all
1076 supported session variables. This query causes problems
1077 because setting the transaction isolation level must be the
1078 first statement of a transaction.
1080 o Add a \set variable to control whether \s displays line numbers
1082 Another option is to add \# which lists line numbers, and
1083 allows command execution.
1085 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
1087 o Prevent escape string warnings when object names have
1090 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1092 o Have \d show foreign keys that reference a table's primary key
1094 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1096 o Have \d show child tables that inherit from the specified parent
1097 o -Have \l+ show database size, if permissions allow
1099 Ideally it will not generate an error for invalid permissions
1101 o Include the symbolic SQLSTATE name in verbose error reports
1103 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1105 o -Improve display of enums to show valid enum values
1106 o Add prompt escape to display the client and server versions
1109 * pg_dump / pg_restore
1110 o %Add dumping of comments on index columns and composite type columns
1111 o %Add full object name to the tag field. eg. for operators we need
1112 '=(integer, integer)', instead of just '='.
1113 o Add pg_dumpall custom format dumps?
1114 o Allow selection of individual object(s) of all types, not just
1116 o In a selective dump, allow dumping of an object and all its
1118 o Add options like pg_restore -l and -L to pg_dump
1119 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1120 o Allow pg_dump --clean to drop roles that own objects or have
1122 o Change pg_dump so that a comment on the dumped database is
1123 applied to the loaded database, even if the database has a
1124 different name. This will require new backend syntax, perhaps
1125 COMMENT ON CURRENT DATABASE.
1126 o Remove unnecessary function pointer abstractions in pg_dump source
1128 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1129 multiple objects simultaneously
1131 The difficulty with this is getting multiple dump processes to
1132 produce a single dump output file. It also would require
1133 several sessions to share the same snapshot.
1134 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1136 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1137 restoring multiple objects simultaneously
1139 This might require a pg_restore flag to indicate how many
1140 simultaneous operations should be performed. Only pg_dump's
1141 -Fc format has the necessary dependency information.
1142 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1144 o To better utilize resources, allow pg_restore to check foreign
1145 keys simultaneously, where possible
1146 o Allow pg_restore to create all indexes of a table
1147 concurrently, via a single heap scan
1149 This requires a pg_dump -Fc file because that format contains
1150 the required dependency information.
1151 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1153 o Allow pg_restore to load different parts of the COPY data
1155 o Prevent pg_dump/pg_restore from being affected by
1158 Using psql to restore a pg_dump dump is also affected.
1160 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1161 o Allow pre/data/post files when schema and data are dumped
1162 separately, for performance reasons
1164 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1171 Document differences between ecpg and the SQL standard and
1172 information about the Informix-compatibility module.
1174 o Solve cardinality > 1 for input descriptors / variables?
1175 o Add a semantic check level, e.g. check if a table really exists
1176 o fix handling of DB attributes that are arrays
1177 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1179 o Fix nested C comments
1180 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1181 o Make SET CONNECTION thread-aware, non-standard?
1182 o Allow multidimensional arrays
1183 o Add internationalized message strings
1184 o Implement COPY FROM STDIN
1188 o Add PQescapeIdentifierConn()
1189 o Prevent PQfnumber() from lowercasing unquoted the column name
1191 PQfnumber() should never have been doing lowercasing, but
1192 historically it has so we need a way to prevent it
1194 o Allow statement results to be automatically batched to the client
1196 Currently all statement results are transferred to the libpq
1197 client before libpq makes the results available to the
1198 application. This feature would allow the application to make
1199 use of the first result rows while the rest are transferred, or
1200 held on the server waiting for them to be requested by libpq.
1201 One complexity is that a statement like SELECT 1/col could error
1202 out mid-way through the result set.
1204 o Consider disallowing multiple queries in PQexec() as an
1205 additional barrier to SQL injection attacks
1207 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1209 o Add PQexecf() that allows complex parameter substitution
1211 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1213 o Add SQLSTATE severity to PGconn return status
1215 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1221 * Add deferred trigger queue file
1223 Right now all deferred trigger information is stored in backend
1224 memory. This could exhaust memory for very large trigger queues.
1225 This item involves dumping large queues into files, or doing some
1226 kind of join to process all the triggers, or some bulk operation.
1227 http://archives.postgresql.org/pgsql-hackers/2008-05/msg00876.php
1229 * Allow triggers to be disabled in only the current session.
1231 This is currently possible by starting a multi-statement transaction,
1232 modifying the system tables, performing the desired SQL, restoring the
1233 system tables, and committing the transaction. ALTER TABLE ...
1234 TRIGGER requires a table lock so it is not ideal for this usage.
1236 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1238 If the dump is known to be valid, allow foreign keys to be added
1239 without revalidating the data.
1241 * Allow statement-level triggers to access modified rows
1242 * Support triggers on columns
1244 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1246 * Allow AFTER triggers on system tables
1248 System tables are modified in many places in the backend without going
1249 through the executor and therefore not causing triggers to fire. To
1250 complete this item, the functions that modify system tables will have
1253 * Tighten trigger permission checks
1255 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1257 * Allow BEFORE INSERT triggers on views
1259 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1261 * -Add ability to trigger on TRUNCATE
1262 * Add database and transaction-level triggers
1264 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00451.php
1271 * Add UNIQUE capability to non-btree indexes
1272 * Prevent index uniqueness checks when UPDATE does not modify the column
1274 Uniqueness (index) checks are done when updating a column even if the
1275 column is not modified by the UPDATE.
1277 * Allow the creation of on-disk bitmap indexes which can be quickly
1278 combined with other bitmap indexes
1280 Such indexes could be more compact if there are only a few distinct values.
1281 Such indexes can also be compressed. Keeping such indexes updated can be
1284 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1285 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1286 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00265.php
1287 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01214.php
1288 http://archives.postgresql.org/pgsql-patches/2007-05/msg00013.php
1289 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00741.php
1291 * Allow accurate statistics to be collected on indexes with more than
1292 one column or expression indexes, perhaps using per-index statistics
1294 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1295 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1297 * Consider increasing the default and maximum number of statistics targets,
1298 and reduce statistics target overhead
1300 Also consider having a larger statistics target for indexed columns
1301 and expression indexes.
1302 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1303 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1304 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1305 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00188.php
1307 * Consider smaller indexes that record a range of values per heap page,
1308 rather than having one index entry for every heap row
1310 This is useful if the heap is clustered by the indexed values.
1311 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1312 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1313 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1314 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1315 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00014.php
1316 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00487.php
1317 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01589.php
1319 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1321 This is difficult because you must upgrade to an exclusive table lock
1322 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1323 have this complication. This would allow index compaction without
1325 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1327 * Allow multiple indexes to be created concurrently, ideally via a
1328 single heap scan, and have pg_restore use it
1330 * Consider sorting entries before inserting into btree index
1332 http://archives.postgresql.org/pgsql-general/2008-01/msg01010.php
1334 * Allow index scans to return matching index keys, not just the matching
1337 http://archives.postgresql.org/pgsql-hackers/2008-04/msg01657.php
1343 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1344 and primary/foreign keys
1345 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1346 on inherited table, e.g. INSERT INTO inherit_table
1347 (unique_index_col) VALUES (dup) should fail
1349 The main difficulty with this item is the problem of
1350 creating an index that can span multiple tables.
1352 o Allow SELECT ... FOR UPDATE on inherited tables
1353 o Require all CHECK constraints to be inherited
1355 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1357 o Add checks to prevent a CREATE RULE views on inherited tables
1359 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1360 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1365 o Add more GIST index support for geometric data types
1366 o Allow GIST indexes to create certain complex index types, like
1367 digital trees (see Aoki)
1372 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1374 o Pack hash index buckets onto disk pages more efficiently
1376 Currently only one hash bucket can be stored on a page. Ideally
1377 several hash buckets could be stored on a single page and greater
1378 granularity used for the hash algorithm.
1380 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1382 o Consider sorting hash buckets so entries can be found using a
1383 binary search, rather than a linear scan
1384 o In hash indexes, consider storing the hash value with or instead
1386 o Add WAL logging for crash recovery
1387 o Allow multi-column hash indexes
1388 o -During index creation, pre-sort the tuples to improve build speed
1395 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1397 Ideally this requires a separate test program that can be run
1398 at initdb time or optionally later. Consider O_SYNC when
1401 * Add program to test if fsync has a delay compared to non-fsync
1402 * Consider sorting writes during checkpoint
1404 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1413 We could use a fixed row count and a +/- count to follow MVCC
1414 visibility rules, or a single cached value could be used and
1415 invalidated if anyone modifies the table. Another idea is to
1416 get a count directly from a unique index, but for this to be
1417 faster than a sequential scan it must avoid access to the heap
1418 to obtain tuple visibility information.
1420 * Provide a way to calculate an "estimated COUNT(*)"
1422 Perhaps by using the optimizer's cardinality estimates or random
1425 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1427 * Allow data to be pulled directly from indexes
1429 Currently indexes do not have enough tuple visibility information
1430 to allow data to be pulled from the index without also accessing
1431 the heap. One way to allow this is to set a bit on index tuples
1432 to indicate if a tuple is currently visible to all transactions
1433 when the first valid heap lookup happens. This bit would have to
1434 be cleared when a heap tuple is expired.
1436 Another idea is to maintain a bitmap of heap pages where all rows
1437 are visible to all backends, and allow index lookups to reference
1438 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1439 add someday to determine which heap pages need vacuuming. Frequently
1440 accessed bitmaps would have to be stored in shared memory. One 8k
1441 page of bitmaps could track 512MB of heap pages.
1443 A third idea would be for a heap scan to check if all rows are visible
1444 and if so set a per-table flag which can be checked by index scans.
1445 Any change to the table would have to clear the flag. To detect
1446 changes during the heap scan a counter could be set at the start and
1447 checked at the end --- if it is the same, the table has not been
1448 modified --- any table change would increment the counter.
1450 http://archives.postgresql.org/pgsql-patches/2007-10/msg00166.php
1451 http://archives.postgresql.org/pgsql-patches/2008-01/msg00049.php
1453 * Consider automatic caching of statements at various levels:
1456 o Query execute plan
1458 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00823.php
1460 * Consider increasing internal areas when shared buffers is increased
1462 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1464 * Consider decreasing the amount of memory used by PrivateRefCount
1466 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1467 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1469 * Consider allowing higher priority queries to have referenced buffer
1470 cache pages stay in memory longer
1472 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1479 * Improve speed with indexes
1481 For large table adjustments during VACUUM FULL, it is faster to cluster
1482 or reindex rather than update the index. Also, index updates can bloat
1485 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1486 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1487 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1489 * Auto-fill the free space map by scanning the buffer cache or by
1490 checking pages written by the background writer
1492 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1493 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1495 * Create a bitmap of pages that need vacuuming
1497 Instead of sequentially scanning the entire table, have the background
1498 writer or some other process record pages that have expired rows, then
1499 VACUUM can look at just those pages rather than the entire table. In
1500 the event of a system crash, the bitmap would probably be invalidated.
1501 One complexity is that index entries still have to be vacuumed, and
1502 doing this without an index scan (by using the heap values to find the
1503 index entry) might be slow and unreliable, especially for user-defined
1506 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1507 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1508 http://archives.postgresql.org/pgsql-patches/2007-03/msg00508.php
1509 http://archives.postgresql.org/pgsql-patches/2007-04/msg00347.php
1510 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00156.php
1511 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00546.php
1512 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00416.php
1514 * Allow FSM to return free space toward the beginning of the heap file,
1515 in hopes that empty pages at the end can be truncated by VACUUM
1516 * Allow FSM page return free space based on table clustering, to assist
1517 in maintaining clustering?
1518 * -Improve dead row detection during multi-statement transactions usage
1520 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1522 * Consider a more compact data representation for dead tuples
1524 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1529 o %Issue log message to suggest VACUUM FULL if a table is nearly
1531 o Improve control of auto-vacuum
1533 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1535 o Prevent long-lived temporary tables from causing frozen-xid
1536 advancement starvation
1538 The problem is that autovacuum cannot vacuum them to set frozen xids;
1539 only the session that created them can do that.
1540 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1542 o Store per-table autovacuum settings in pg_class.reloptions.
1544 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1545 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1547 o Prevent autovacuum from running if an old transaction is still
1548 running from the last vacuum
1550 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1557 * Fix priority ordering of read and write light-weight locks (Neil)
1559 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1560 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1562 * Fix problem when multiple subtransactions of the same outer transaction
1563 hold different types of locks, and one subtransaction aborts
1565 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1566 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1567 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1568 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1570 * Allow UPDATEs on only non-referential integrity columns not to conflict
1571 with referential integrity locks
1573 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1575 * Add idle_in_transaction_timeout GUC so locks are not held for long
1578 * Improve deadlock detection when a page cleaning lock conflicts
1579 with a shared buffer that is pinned
1581 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1582 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1583 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1585 * Detect deadlocks involving LockBufferForCleanup()
1587 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1591 Startup Time Improvements
1592 =========================
1594 * Experiment with multi-threaded backend for backend creation
1596 This would prevent the overhead associated with process creation. Most
1597 operating systems have trivial process creation time compared to
1598 database startup overhead, but a few operating systems (Win32,
1599 Solaris) might benefit from threading. Also explore the idea of
1600 a single session using multiple threads to execute a statement faster.
1607 * Eliminate need to write full pages to WAL before page modification
1609 Currently, to protect against partial disk page writes, we write
1610 full page images to WAL before they are modified so we can correct any
1611 partial page writes during recovery. These pages can also be
1612 eliminated from point-in-time archive files.
1613 http://archives.postgresql.org/pgsql-hackers/2002-06/msg00655.php
1615 o When off, write CRC to WAL and check file system blocks
1618 If CRC check fails during recovery, remember the page in case
1619 a later CRC for that page properly matches.
1621 o Write full pages during file system write and not when
1622 the page is modified in the buffer cache
1624 This allows most full page writes to happen in the background
1625 writer. It might cause problems for applying WAL on recovery
1626 into a partially-written page, but later the full page will be
1629 * Allow WAL traffic to be streamed to another server for stand-by
1631 * Reduce WAL traffic so only modified values are written rather than
1634 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1636 * Allow WAL information to recover corrupted pg_controldata
1638 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1640 * Find a way to reduce rotational delay when repeatedly writing
1643 Currently fsync of WAL requires the disk platter to perform a full
1644 rotation to fsync again. One idea is to write the WAL to different
1645 offsets that might reduce the rotational delay.
1646 http://archives.postgresql.org/pgsql-hackers/2002-11/msg00483.php
1648 * Allow WAL logging to be turned off for a table, but the table
1649 might be dropped or truncated during crash recovery
1651 Allow tables to bypass WAL writes and just fsync() dirty pages on
1652 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1653 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1654 non-default logging should not use referential integrity with
1655 default-logging tables. A table without dirty buffers during a
1656 crash could perhaps avoid the drop/truncate.
1657 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1659 * Allow WAL logging to be turned off for a table, but the table would
1660 avoid being truncated/dropped
1662 To do this, only a single writer can modify the table, and writes
1663 must happen only on new pages so the new pages can be removed during
1664 crash recovery. Readers can continue accessing the table. Such
1665 tables probably cannot have indexes. One complexity is the handling
1666 of indexes on TOAST tables.
1667 http://archives.postgresql.org/pgsql-hackers/2005-12/msg01016.php
1669 * Speed WAL recovery by allowing more than one page to be prefetched
1671 This should be done utilizing the same infrastructure used for
1672 prefetching in general to avoid introducing complex error-prone code
1674 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1675 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1676 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1678 * Improve WAL concurrency by increasing lock granularity
1680 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1682 * Be more aggressive about creating WAL files
1684 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01325.php
1685 http://archives.postgresql.org/pgsql-hackers/2004-07/msg01075.php
1686 http://archives.postgresql.org/pgsql-hackers/2005-04/msg00556.php
1688 * Have resource managers report the duration of their status changes
1690 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01468.php
1692 * Move pgfoundry's xlogdump to /contrib and have it rely more closely
1693 on the WAL backend code
1695 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00035.php
1700 Optimizer / Executor
1701 ====================
1703 * Improve selectivity functions for geometric operators
1704 * Precompile SQL functions to avoid overhead
1705 * Create utility to compute accurate random_page_cost value
1706 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1707 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1708 actual row counts differ by a specified percentage
1709 * Improve how ANALYZE computes in-doubt tuples
1711 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00771.php
1713 * Consider using hash buckets to do DISTINCT, rather than sorting
1715 This would be beneficial when there are few distinct values. This is
1716 already used by GROUP BY.
1718 * Log statements where the optimizer row estimates were dramatically
1719 different from the number of rows actually found?
1720 * Consider compressed annealing to search for query plans
1722 This might replace GEQO, http://sixdemonbag.org/Djinni.
1724 * Improve merge join performance by allowing mark/restore of
1727 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1729 * Consider using a hash for joining to a large IN (VALUES ...) list
1731 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1738 * Consider having the background writer update the transaction status
1739 hint bits before writing out the page
1741 Implementing this requires the background writer to have access to system
1742 catalogs and the transaction status log.
1744 * Consider adding buffers the background writer finds reusable to the
1747 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1749 * Automatically tune bgwriter_delay based on activity rather then using a
1752 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1754 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1756 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1758 * Test to see if calling PreallocXlogFiles() from the background writer
1759 will help with WAL segment creation latency
1761 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1765 Miscellaneous Performance
1766 =========================
1768 * Do async I/O for faster random read-ahead of data
1770 Async I/O allows multiple I/O requests to be sent to the disk with
1771 results coming back asynchronously.
1773 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1774 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1775 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1776 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1778 * Use mmap() rather than SYSV shared memory or to write WAL files?
1780 This would remove the requirement for SYSV SHM but would introduce
1781 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1782 to prevent I/O overhead.
1784 * Consider mmap()'ing files into a backend?
1786 Doing I/O to large tables would consume a lot of address space or
1787 require frequent mapping/unmapping. Extending the file also causes
1788 mapping problems that might require mapping only individual pages,
1789 leading to thousands of mappings. Another problem is that there is no
1790 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1791 could hit disk before WAL is written.
1793 * Add a script to ask system configuration questions and tune postgresql.conf
1794 * Consider ways of storing rows more compactly on disk
1796 o Reduce the row header size?
1797 o Consider reducing on-disk varlena length from four bytes to
1798 two because a heap row cannot be more than 64k in length
1800 * Consider increasing NUM_CLOG_BUFFERS
1802 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1803 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1805 * Consider transaction start/end performance improvements
1807 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00948.php
1808 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1810 * Allow user configuration of TOAST thresholds
1812 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1813 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1815 * Allow configuration of backend priorities via the operating system
1817 Though backend priorities make priority inversion during lock
1818 waits possible, research shows that this is not a huge problem.
1820 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1822 * Experiment with multi-threaded backend better I/O utilization
1824 This would allow a single query to make use of multiple I/O channels
1825 simultaneously. One idea is to create a background reader that can
1826 pre-fetch sequential and index scan pages needed by other backends.
1827 This could be expanded to allow concurrent reads from multiple devices
1828 in a partitioned table.
1830 * Experiment with multi-threaded backend better CPU utilization
1832 This would allow several CPUs to be used for a single query, such as
1833 for sorting or query execution.
1835 * Consider increasing the minimum allowed number of shared buffers
1837 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1839 * Expire published xmin for read-only and idle transactions
1841 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1843 * Consider if CommandCounterIncrement() can avoid its
1844 AcceptInvalidationMessages() call
1846 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1848 * Improve performance of shared invalidation queue for multiple CPUs
1850 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1852 * Consider Cartesian joins when both relations are needed to form an
1853 indexscan qualification for a third relation
1855 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1857 * Consider not storing a NULL bitmap on disk if all the NULLs are
1860 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1861 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1863 * Sort large UPDATE/DELETEs so it is done in heap order
1865 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1867 * -Avoid tuple some tuple copying in sort routines
1868 * SMP scalability improvements
1870 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1871 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1872 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1874 * Research reducing deTOASTing in more places
1876 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1878 * Consider being smarter about memory and external files used during
1881 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01101.php
1882 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00045.php
1884 * Allow one transaction to see tuples using the snapshot of another
1887 This would assist multiple backends in working together.
1888 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00400.php
1894 * Add use of 'const' for variables in source tree
1895 * Move some things from contrib into main tree
1896 * %Remove warnings created by -Wcast-align
1897 * Move platform-specific ps status display info from ps_status.c to ports
1898 * Add optional CRC checksum to heap and index pages
1899 * Improve documentation to build only interfaces (Marc)
1900 * Remove or relicense modules that are not under the BSD license, if possible
1901 * Acquire lock on a relation before building a relcache entry for it
1902 * Allow cross-compiling by generating the zic database on the target system
1903 * Improve NLS maintenance of libpgport messages linked onto applications
1904 * Clean up casting in contrib/isn
1906 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1908 * Improve the /contrib installation experience
1910 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00132.php
1912 * Use UTF8 encoding for NLS messages so all server encodings can
1914 * Update Bonjour to work with newer cross-platform SDK
1916 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1917 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1919 * Consider detoasting keys before sorting
1920 * Consider GnuTLS if OpenSSL license becomes a problem
1922 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1923 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1925 * Consider changing documentation format from SGML to XML
1927 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1929 * Consider making NAMEDATALEN more configurable in future releases
1930 * Update our code to handle 64-bit timezone files to match the zic
1931 source code, which now uses them
1932 * Have configure choose integer datetimes by default
1934 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1936 * Support scoped IPv6 addresses
1938 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1940 * Consider allowing 64-bit integers and floats to be passed by value on
1943 Also change 32-bit floats (float4) to be passed by value at the same
1946 * Research use of signals and sleep wake ups
1948 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1950 * Add automated check for invalid C++ source code constructs
1952 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1954 * Consider simplifying how memory context resets handle child contexts
1956 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1958 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1960 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1962 * Convert single quotes to apostrophes in the PDF documentation
1964 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
1966 * Create three versions of libpgport to simplify client code
1968 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00154.php
1970 * Remove old-style routines for manipulating tuples
1972 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00851.php
1974 * Improve detection of shared memory segments being used by others
1975 by checking the SysV shared memory field 'nattch'
1977 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00656.php
1978 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00673.php
1980 * Implement the non-threaded Avahi service discovery protocol
1982 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00939.php
1983 http://archives.postgresql.org/pgsql-patches/2008-02/msg00097.php
1984 http://archives.postgresql.org/pgsql-hackers/2008-03/msg01211.php
1985 http://archives.postgresql.org/pgsql-patches/2008-04/msg00001.php
1989 o Remove configure.in check for link failure when cause is found
1990 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1992 o -Remove psql newline patch when we find out why mingw outputs an
1994 o Allow psql to use readline once non-US code pages work with
1996 o Fix problem with shared memory on the Win32 Terminal Server
1997 o Diagnose problem where shared memory can sometimes not be
1998 attached by postmaster children
2000 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
2002 o Improve signal handling
2004 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
2006 o Convert MSVC build system to remove most batch files
2008 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
2010 o Prevent SSL from sending network packets to avoid interference
2011 with Win32 signal emulation
2013 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
2015 o Support pgxs when using MSVC
2017 o Fix MSVC NLS support, like for to_char()
2019 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
2020 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
2022 o Find a correct rint() substitute on Windows
2024 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00808.php
2028 * Wire Protocol Changes
2030 o Allow dynamic character set handling
2031 o Add decoded type, length, precision
2033 o Update clients to use data types, typmod, schema.table.column names
2034 of result sets using new statement protocol
2040 * Add pre-parsing phase that converts non-ISO syntax to supported
2043 This could allow SQL written for other databases to run without
2046 * Allow plug-in modules to emulate features from other databases
2047 * Add features of Oracle-style packages (Pavel)
2049 A package would be a schema with session-local variables,
2050 public/private functions, and initialization functions. It
2051 is also possible to implement these capabilities
2052 in any schema and not use a separate "packages"
2055 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
2057 * Consider allowing control of upper/lower case folding of unquoted
2060 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
2061 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
2062 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00849.php
2064 * Add autonomous transactions
2066 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
2070 Features We Do _Not_ Want
2071 =========================
2073 * All backends running as threads in a single process (not wanted)
2075 This eliminates the process protection we get from the current setup.
2076 Thread creation is usually the same overhead as process creation on
2077 modern systems, so it seems unwise to use a pure threaded model.
2079 * Optimizer hints (not wanted)
2081 Optimizer hints are used to work around problems in the optimizer. We
2082 would rather have the problems reported and fixed.
2084 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
2085 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
2086 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
2089 Because we support postfix operators, it isn't possible to make AS
2090 optional and continue to use bison.
2091 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
2093 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
2095 * Embedded server (not wanted)
2097 While PostgreSQL clients runs fine in limited-resource environments, the
2098 server requires multiple processes and a stable pool of resources to
2099 run reliabily and efficiently. Stripping down the PostgreSQL server
2100 to run in the same process address space as the client application
2101 would add too much complexity and failure cases.