3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Mon Mar 3 16:00:28 EST 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 8.3 release.#
10 #A percent sign, "%", marks items that are easier to implement.#
12 Bracketed items, "[]", have more detail.
14 This list contains all known PostgreSQL bugs and feature requests. If
15 you would like to work on an item, please read the Developer's FAQ
16 first. There is also a developer's wiki at
17 http://developer.postgresql.org.
23 * Allow administrators to safely terminate individual sessions either
24 via an SQL function or SIGTERM
26 Lock table corruption following SIGTERM of an individual backend
27 has been reported in 8.0. A possible cause was fixed in 8.1, but
28 it is unknown whether other problems exist. This item mostly
29 requires additional testing rather than of writing any new code.
31 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00174.php
33 * Check for unreferenced table files created by transactions that were
34 in-progress when the server terminated abruptly
36 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
38 * Set proper permissions on non-system schemas during db creation
40 Currently all schemas are owned by the super-user because they are copied
41 from the template1 database. However, since all objects are inherited
42 from the template database, it is not clear that setting schemas to the db
45 * Add function to report the time of the most recent server reload
46 * Allow statistics collector information to be pulled from the collector
47 process directly, rather than requiring the collector to write a
48 filesystem file twice a second?
49 * Allow log_min_messages to be specified on a per-module basis
51 This would allow administrators to see more detailed information from
52 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
53 Another idea is to allow separate configuration files for each module,
54 or allow arbitrary SET commands to be passed to them.
56 * Simplify ability to create partitioned tables
58 This would allow creation of partitioned tables without requiring
59 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
60 for rapid partition selection. Options could include range and hash
63 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
64 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
66 * Allow auto-selection of partitioned tables for min/max() operations
67 * Allow more complex user/database default GUC settings
69 Currently ALTER USER and ALTER DATABASE support per-user and
70 per-database defaults. Consider adding per-user-and-database
71 defaults so things like search_path can be defaulted for a
72 specific user connecting to a specific database.
74 * Allow custom variable classes that can restrict who can set the values
76 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
78 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
79 the privilege granted by the invoking role, and not those granted
82 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
84 * Allow SSL authentication/encryption over unix domain sockets
86 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
90 o Allow pg_hba.conf to specify host names along with IP addresses
92 Host name lookup could occur when the postmaster reads the
93 pg_hba.conf file, or when the backend starts. Another
94 solution would be to reverse lookup the connection IP and
95 check that hostname against the host names in pg_hba.conf.
96 We could also then check that the host name maps to the IP
99 o %Allow postgresql.conf file values to be changed via an SQL
100 API, perhaps using SET GLOBAL
101 o Allow the server to be stopped/restarted via an SQL API
102 o Issue a warning if a change-on-restart-only postgresql.conf value
103 is modified and the server config files are reloaded
108 o Allow a database in tablespace t1 with tables created in
109 tablespace t2 to be used as a template for a new database created
110 with default tablespace t2
112 Currently all objects in the default database tablespace must
113 have default tablespace specifications. This is because new
114 databases are created by copying directories. If you mix default
115 tablespace tables and tablespace-specified tables in the same
116 directory, creating a new database from such a mixed directory
117 would create a new database with tables that had incorrect
118 explicit tablespaces. To fix this would require modifying
119 pg_class in the newly copied database, which we don't currently
122 o Allow reporting of which objects are in which tablespaces
124 This item is difficult because a tablespace can contain objects
125 from multiple databases. There is a server-side function that
126 returns the databases which use a specific tablespace, so this
127 requires a tool that will call that function and connect to each
128 database to find the objects in each database for that tablespace.
130 o Allow WAL replay of CREATE TABLESPACE to work when the directory
131 structure on the recovery computer is different from the original
133 o Allow per-tablespace quotas
136 * Point-In-Time Recovery (PITR)
138 o Allow a warm standby system to also allow read-only statements
141 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
143 o %Create dump tool for write-ahead logs for use in determining
144 transaction id for point-in-time recovery
146 This is useful for checking PITR recovery.
148 o Allow recovery.conf to support the same syntax as
149 postgresql.conf, including quoting
151 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
159 * Change NUMERIC to enforce the maximum precision
160 * Reduce storage space for small NUMERICs
162 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
163 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
165 * Fix data types where equality comparison isn't intuitive, e.g. box
166 * Add support for public SYNONYMs
168 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
170 * Fix CREATE CAST on DOMAINs
172 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
173 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
175 * Add support for SQL-standard GENERATED/IDENTITY columns
177 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
178 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
179 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
180 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
182 * Improve XML support
184 http://developer.postgresql.org/index.php/XML_Support
186 * Consider placing all sequences in a single table, or create a system
189 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
191 * Allow the UUID type to accept non-standard formats
193 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
197 o Allow infinite dates and intervals just like infinite timestamps
198 o Merge hardwired timezone names with the TZ database; allow either
199 kind everywhere a TZ name is currently taken
200 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
201 information, either zone name or offset from UTC [timezone]
203 If the TIMESTAMP value is stored with a time zone name, interval
204 computations should adjust based on the time zone rules.
206 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
207 o Add a GUC variable to allow output of interval values in ISO8601
209 o Have timestamp subtraction not call justify_hours()?
211 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
213 o Improve timestamptz subtraction to be DST-aware
215 Currently subtracting one date from another that crosses a
216 daylight savings time adjustment can return '1 day 1 hour', but
217 adding that back to the first date returns a time one hour in
218 the future. This is caused by the adjustment of '25 hours' to
219 '1 day 1 hour', and '1 day' is the same time the next day, even
220 if daylight savings adjustments are involved.
222 o Fix interval display to support values exceeding 2^31 hours
223 o Add overflow checking to timestamp and interval arithmetic
224 o Extend timezone code to allow 64-bit values so we can
225 represent years beyond 2038
227 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
229 o Use LC_TIME for localized weekday/month names, rather than
232 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
234 o Add ISO INTERVAL handling
236 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
237 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
239 o Support ISO INTERVAL syntax if units cannot be determined from
240 the string, and are supplied after the string
242 The SQL standard states that the units after the string
243 specify the units of the string, e.g. INTERVAL '2' MINUTE
244 should return '00:02:00'. The current behavior has the units
245 restrict the interval value to the specified unit or unit
246 range, INTERVAL '70' SECOND returns '00:00:10'.
248 For syntax that isn't uniquely ISO or PG syntax, like '1' or
249 '1:30', treat as ISO if there is a range specification clause,
250 and as PG if there no clause is present, e.g. interpret '1:30'
251 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
252 '1:30' as '1 hour, 30 minutes'.
254 This makes common cases like SELECT INTERVAL '1' MONTH
255 SQL-standard results. The SQL standard supports a limited
256 number of unit combinations and doesn't support unit names in
257 the string. The PostgreSQL syntax is more flexible in the
258 range of units supported, e.g. PostgreSQL supports '1 year 1
259 hour', while the SQL standard does not.
261 o Add support for year-month syntax, INTERVAL '50-6' YEAR
263 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
264 year' AS INTERVAL MONTH), and this should return '12 months'
265 o Round or truncate values to the requested precision, e.g.
266 INTERVAL '11 months' AS YEAR should return one or zero
267 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
272 o Delay resolution of array expression's data type so assignment
273 coercion can be performed on empty array expressions
274 o Add support for arrays of domains
276 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
278 o Allow single-byte header storage for array elements
283 o Improve vacuum of large objects, like contrib/vacuumlo?
284 o Add security checking for large objects
285 o Auto-delete large objects when referencing row is deleted
287 contrib/lo offers this functionality.
289 o Allow read/write into TOAST values like large objects
291 This requires the TOAST column to be stored EXTERNAL.
293 o Add API for 64-bit large object access
295 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
299 * Add locale-aware MONEY type, and support multiple currencies
301 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
302 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
304 * MONEY dumps in a locale-specific format making it difficult to
305 restore to a system with a different locale
306 * Allow MONEY to be easily cast to/from other numeric data types
313 * Allow INET subnet tests using non-constants to be indexed
314 * Allow to_date() and to_timestamp() accept localized month names
315 * Fix to_date()-related functions to consistently issue errors
317 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
319 * Add missing parameter handling in to_char()
321 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
323 * Allow substring/replace() to get/set bit values
324 * Allow to_char() on interval values to accumulate the highest unit
327 Some special format flag would be required to request such
328 accumulation. Such functionality could also be added to EXTRACT.
329 Prevent accumulation that crosses the month/day boundary because of
330 the uneven number of days in a month.
332 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
333 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
334 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
335 o to_char(INTERVAL '3 years 5 months','MM') => 41
337 * Implement inlining of set-returning functions defined in SQL
338 * Allow SQL-language functions to return results from RETURNING queries
340 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
342 * Allow SQL-language functions to reference parameters by parameter name
344 Currently SQL-language functions can only refer to dollar parameters,
347 * Add SPI_gettypmod() to return the typemod for a TupleDesc
348 * Enforce typmod for function inputs, function results and parameters for
349 spi_prepare'd statements called from PLs
351 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
353 * Allow holdable cursors in SPI
354 * Tighten function permission checks
356 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
358 * Fix IS OF so it matches the ISO specification, and add documentation
360 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
361 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
363 * Add missing operators for geometric data types
365 Some geometric types do not have the full suite of geometric operators,
368 * Implement Boyer-Moore searching in strpos()
370 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
374 Multi-Language Support
375 ======================
377 * Add NCHAR (as distinguished from ordinary varchar),
378 * Allow locale to be set at database creation
380 Currently locale can only be set during initdb. No global tables have
381 locale-aware columns. However, the database template used during
382 database creation might have locale-aware indexes. The indexes would
383 need to be reindexed to match the new locale.
385 * Allow encoding on a per-column basis optionally using the ICU library:
387 Right now only one encoding is allowed per database. [locale]
389 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
390 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
391 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
392 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
394 * Add CREATE COLLATE? [locale]
395 * Support multiple simultaneous character sets, per SQL92
396 * Improve UTF8 combined character handling?
397 * Add octet_length_server() and octet_length_client()
398 * Make octet_length_client() the same as octet_length()?
399 * Fix problems with wrong runtime encoding conversion for NLS message files
400 * Add URL to more complete multi-byte regression tests
402 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
404 * Fix ILIKE and regular expressions to handle case insensitivity
405 properly in multibyte encodings
407 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
408 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
410 * Set client encoding based on the client operating system encoding
412 Currently client_encoding is set in postgresql.conf, which
413 defaults to the server encoding.
415 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
422 * Automatically create rules on views so they are updateable, per SQL99
424 We can only auto-create rules for simple views. For more complex
425 cases users will still have to write rules manually.
427 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
428 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
430 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
431 * Allow NOTIFY in rules involving conditionals
432 * Allow VIEW/RULE recompilation when the underlying tables change
434 Another issue is whether underlying table changes should be reflected
435 in the view, e.g. should SELECT * show additional columns if they
436 are added after the view is created.
438 * Make it possible to use RETURNING together with conditional DO INSTEAD
439 rules, such as for partitioning setups
441 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
443 * Add the ability to automatically create materialized views
445 Right now materialized views require the user to create triggers on the
446 main table to keep the summary table current. SQL syntax should be able
447 to manager the triggers and summary table automatically. A more
448 sophisticated implementation would automatically retrieve from the
449 summary table when the main table is referenced, if possible.
456 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
457 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
458 * %Allow SET CONSTRAINTS to be qualified by schema/table name
459 * %Add a separate TRUNCATE permission
461 Currently only the owner can TRUNCATE a table because triggers are not
462 called, and the table is locked in exclusive mode.
464 * Allow PREPARE of cursors
465 * Allow finer control over the caching of prepared query plans
467 Currently queries prepared via the libpq API are planned on first
468 execute using the supplied parameters --- allow SQL PREPARE to do the
469 same. Also, allow control over replanning prepared queries either
470 manually or automatically when statistics for execute parameters
471 differ dramatically from those used during planning.
473 * Improve logging of prepared transactions recovered during startup
475 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
477 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
479 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
480 such information in memory would improve performance.
482 * Add optional textual message to NOTIFY
484 This would allow an informational message to be added to the notify
485 message, perhaps indicating the row modified or other custom
488 * Allow multiple identical NOTIFY events to always be communicated to the
489 client, rather than sent as a single notification to the listener
490 * Add a GUC variable to warn about non-standard SQL usage in queries
491 * Add SQL-standard MERGE command, typically used to merge two tables
494 This is similar to UPDATE, then for unmatched rows, INSERT.
495 Whether concurrent access allows modifications which could cause
496 row loss is implementation independent.
498 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
501 To implement this cleanly requires that the table have a unique index
502 so duplicate checking can be easily performed. It is possible to
503 do it without a unique index if we require the user to LOCK the table
506 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
507 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
509 * Add NOVICE output level for helpful messages like automatic sequence/index
511 * Add GUC to issue notice about statements that use unjoined tables
512 * Allow EXPLAIN to identify tables that were skipped because of
514 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
515 * Enable standard_conforming_strings
516 * Make standard_conforming_strings the default in 8.5?
518 When this is done, backslash-quote should be prohibited in non-E''
519 strings because of possible confusion over how such strings treat
520 backslashes. Basically, '' is always safe for a literal single
521 quote, while \' might or might not be based on the backslash
524 * Simplify dropping roles that have objects in several databases
525 * Allow COMMENT ON to accept an expression rather than just a string
526 * Allow the count returned by SELECT, etc to be represented as an int64
527 to allow a higher range of values
528 * Add SQL99 WITH clause to SELECT
529 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
530 * Add DEFAULT .. AS OWNER so permission checks are done as the table
533 This would be useful for SERIAL nextval() calls and CHECK constraints.
535 * Allow DISTINCT to work in multiple-argument aggregate calls
536 * Add column to pg_stat_activity that shows the progress of long-running
537 commands like CREATE INDEX and VACUUM
538 * Implement SQL:2003 window functions
539 * Improve failure message when DROP DATABASE is used on a database that
540 has prepared transactions
541 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
543 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
544 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
546 * Increase locking when DROPing objects so dependent objects cannot
547 get dropped while the DROP operation is happening
549 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
551 * -Allow AS in "SELECT col AS label" to be optional in certain cases
555 o Allow CREATE TABLE AS to determine column lengths for complex
556 expressions like SELECT col1 || col2
558 o Have WITH CONSTRAINTS also create constraint indexes
560 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
563 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
565 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
566 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
568 o Research self-referential UPDATEs that see inconsistent row versions
569 in read-committed mode
571 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
572 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
574 o Allow GLOBAL temporary tables to exist as empty by default in
577 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
582 o Have ALTER TABLE RENAME rename SERIAL sequence names
584 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
586 o Have ALTER SEQUENCE RENAME rename the sequence name stored
587 in the sequence table
589 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
590 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
591 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
593 o Add ALTER DOMAIN to modify the underlying data type
594 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
596 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
598 o %Allow ALTER TABLE to change constraint deferrability and actions
599 o Add missing object types for ALTER ... SET SCHEMA
600 o Allow ALTER TABLESPACE to move to different directories
601 o Allow databases to be moved to different tablespaces
602 o Allow moving system tables to other tablespaces, where possible
604 Currently non-global system tables must be in the default database
605 tablespace. Global system tables can never be moved.
607 o Prevent parent tables from altering or dropping constraints
608 like CHECK that are inherited by child tables unless CASCADE
610 o %Prevent child tables from altering or dropping constraints
611 like CHECK that were inherited from the parent table
612 o Have ALTER INDEX update the name of a constraint using that index
613 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
614 o Allow column display reordering by recording a display,
615 storage, and permanent id for every column?
617 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
623 o Automatically maintain clustering on a table
625 This might require some background daemon to maintain clustering
626 during periods of low usage. It might also require tables to be only
627 partially filled for easier reorganization. Another idea would
628 be to create a merged heap/index data file so an index lookup would
629 automatically access the heap data too. A third idea would be to
630 store heap rows in hashed groups, perhaps using a user-supplied
632 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
634 o %Add default clustering to system tables
636 To do this, determine the ideal cluster index for each system
637 table and set the cluster setting during initdb.
639 o %Add VERBOSE option to report tables as they are processed,
645 o Allow COPY to report error lines and continue
647 This requires the use of a savepoint before each COPY line is
648 processed, with ROLLBACK on COPY failure.
650 o Allow COPY on a newly-created table to skip WAL logging
652 On crash recovery, the table involved in the COPY would
653 be removed or have its heap and index files truncated. One
654 issue is that no other backend should be able to add to
655 the table at the same time, which is something that is
656 currently allowed. This currently is done if the table is
657 created inside the same transaction block as the COPY because
658 no other backends can see the table.
660 o Consider using a ring buffer for COPY FROM
662 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
663 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
668 o Allow column-level privileges
669 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
672 The proposed syntax is:
673 GRANT SELECT ON ALL TABLES IN public TO phpuser;
674 GRANT SELECT ON NEW TABLES IN public TO phpuser;
676 o Allow GRANT/REVOKE permissions to be inherited by objects based on
679 o Allow SERIAL sequences to inherit permissions from the base table?
684 o Prevent DROP TABLE from dropping a row referenced by its own open
690 o Allow INSERT/UPDATE of the system-generated oid value for a row
691 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
697 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
701 * Referential Integrity
703 o Add MATCH PARTIAL referential integrity
704 o Change foreign key constraint for array -> element to mean element
706 o Fix problem when cascading referential triggers make changes on
707 cascaded tables, seeing the tables in an intermediate state
709 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
710 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
712 o Allow DEFERRABLE and end-of-statement UNIQUE constraints?
714 This would allow UPDATE tab SET col = col + 1 to work if col has
715 a unique index. Currently, uniqueness checks are done while the
716 command is being executed, rather than at the end of the statement
719 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
720 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
723 * Server-Side Languages
726 o Fix RENAME to work on variables other than OLD/NEW
728 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
729 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
730 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
732 o Allow function parameters to be passed by name,
733 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
734 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
735 o Allow listing of record column names, and access to
736 record columns via variables, e.g. columns := r.(*),
739 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
740 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
741 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
743 o Add support for SCROLL cursors
744 o Add support for WITH HOLD cursors
745 o Allow row and record variables to be set to NULL constants,
746 and allow NULL tests on such variables
748 Because a row is not scalar, do not allow assignment
749 from NULL-valued scalars.
751 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
754 o Add table function support to pltcl, plpythonu
755 o Add support for polymorphic arguments and return types to
756 languages other than PL/PgSQL
757 o Add capability to create and call PROCEDURES
758 o Add support for OUT and INOUT parameters to languages other
760 o Add PL/PythonU tracebacks
762 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
764 o Allow data to be passed in native language formats, rather
767 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
773 * Have pg_ctl look at PGHOST in case it is a socket directory?
774 * Allow pg_ctl to work properly with configuration files located outside
777 pg_ctl can not read the pid file because it isn't located in the
778 config directory but in the PGDATA directory. The solution is to
779 allow pg_ctl to read and understand postgresql.conf to find the
780 data_directory value.
783 o Have psql show current values for a sequence
784 o Move psql backslash database information into the backend, use
785 mnemonic commands? [psql]
787 This would allow non-psql clients to pull the same information out
788 of the database as psql.
790 o Fix psql's \d commands more consistent
792 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
793 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
795 o Consistently display privilege information for all objects in psql
796 o Add auto-expanded mode so expanded output is used if the row
797 length is wider than the screen width.
799 Consider using auto-expanded mode for backslash commands like \df+.
801 o Prevent tab completion of SET TRANSACTION from querying the
802 database and therefore preventing the transaction isolation
803 level from being set.
805 Currently SET <tab> causes a database lookup to check all
806 supported session variables. This query causes problems
807 because setting the transaction isolation level must be the
808 first statement of a transaction.
810 o Add a \set variable to control whether \s displays line numbers
812 Another option is to add \# which lists line numbers, and
813 allows command execution.
815 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
819 o %Add dumping of comments on index columns and composite type columns
820 o %Add full object name to the tag field. eg. for operators we need
821 '=(integer, integer)', instead of just '='.
822 o Add pg_dumpall custom format dumps?
823 o Allow selection of individual object(s) of all types, not just
825 o In a selective dump, allow dumping of an object and all its
827 o Add options like pg_restore -l and -L to pg_dump
828 o Stop dumping CASCADE on DROP TYPE commands in clean mode
829 o Allow pg_dump --clean to drop roles that own objects or have
831 o Change pg_dump so that a comment on the dumped database is
832 applied to the loaded database, even if the database has a
833 different name. This will require new backend syntax, perhaps
834 COMMENT ON CURRENT DATABASE.
835 o Remove unnecessary function pointer abstractions in pg_dump source
842 Document differences between ecpg and the SQL standard and
843 information about the Informix-compatibility module.
845 o Solve cardinality > 1 for input descriptors / variables?
846 o Add a semantic check level, e.g. check if a table really exists
847 o fix handling of DB attributes that are arrays
848 o Use backend PREPARE/EXECUTE facility for ecpg where possible
850 o Fix nested C comments
851 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
852 o Make SET CONNECTION thread-aware, non-standard?
853 o Allow multidimensional arrays
854 o Add internationalized message strings
855 o Implement COPY FROM STDIN
859 o Add PQescapeIdentifierConn()
860 o Prevent PQfnumber() from lowercasing unquoted the column name
862 PQfnumber() should never have been doing lowercasing, but
863 historically it has so we need a way to prevent it
865 o Allow statement results to be automatically batched to the client
867 Currently all statement results are transferred to the libpq
868 client before libpq makes the results available to the
869 application. This feature would allow the application to make
870 use of the first result rows while the rest are transferred, or
871 held on the server waiting for them to be requested by libpq.
872 One complexity is that a statement like SELECT 1/col could error
873 out mid-way through the result set.
875 o Consider disallowing multiple queries in PQexec() as an
876 additional barrier to SQL injection attacks
878 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
880 o Add PQexecf() that allows complex parameter substitution
882 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
887 * Add deferred trigger queue file
889 Right now all deferred trigger information is stored in backend
890 memory. This could exhaust memory for very large trigger queues.
891 This item involves dumping large queues into files.
893 * Allow triggers to be disabled in only the current session.
895 This is currently possible by starting a multi-statement transaction,
896 modifying the system tables, performing the desired SQL, restoring the
897 system tables, and committing the transaction. ALTER TABLE ...
898 TRIGGER requires a table lock so it is not ideal for this usage.
900 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
902 If the dump is known to be valid, allow foreign keys to be added
903 without revalidating the data.
905 * Allow statement-level triggers to access modified rows
906 * Support triggers on columns
908 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
910 * Allow AFTER triggers on system tables
912 System tables are modified in many places in the backend without going
913 through the executor and therefore not causing triggers to fire. To
914 complete this item, the functions that modify system tables will have
917 * Tighten trigger permission checks
919 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
921 * Allow BEFORE INSERT triggers on views
923 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
925 * Add ability to trigger on TRUNCATE
927 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
934 * Add UNIQUE capability to non-btree indexes
935 * Prevent index uniqueness checks when UPDATE does not modify the column
937 Uniqueness (index) checks are done when updating a column even if the
938 column is not modified by the UPDATE.
940 * Allow the creation of on-disk bitmap indexes which can be quickly
941 combined with other bitmap indexes
943 Such indexes could be more compact if there are only a few distinct values.
944 Such indexes can also be compressed. Keeping such indexes updated can be
947 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
948 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
950 * Allow accurate statistics to be collected on indexes with more than
951 one column or expression indexes, perhaps using per-index statistics
953 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
954 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
956 * Consider compressing indexes by storing key values duplicated in
957 several rows as a single index entry
958 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
960 This is difficult because you must upgrade to an exclusive table lock
961 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
962 have this complication. This would allow index compaction without
965 * Allow multiple indexes to be created concurrently, ideally via a
966 single heap scan, and have a restore of a pg_dump somehow use it
968 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
973 o Allow inherited tables to inherit indexes, UNIQUE constraints,
974 and primary/foreign keys
975 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
976 on inherited table, e.g. INSERT INTO inherit_table
977 (unique_index_col) VALUES (dup) should fail
979 The main difficulty with this item is the problem of
980 creating an index that can span multiple tables.
982 o Allow SELECT ... FOR UPDATE on inherited tables
987 o Add more GIST index support for geometric data types
988 o Allow GIST indexes to create certain complex index types, like
989 digital trees (see Aoki)
994 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
996 o Pack hash index buckets onto disk pages more efficiently
998 Currently only one hash bucket can be stored on a page. Ideally
999 several hash buckets could be stored on a single page and greater
1000 granularity used for the hash algorithm.
1002 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1004 o Consider sorting hash buckets so entries can be found using a
1005 binary search, rather than a linear scan
1006 o In hash indexes, consider storing the hash value with or instead
1008 o Add WAL logging for crash recovery
1009 o Allow multi-column hash indexes
1010 o During index creation, pre-sort the tuples to improve build speed
1012 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
1019 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1021 Ideally this requires a separate test program that can be run
1022 at initdb time or optionally later. Consider O_SYNC when
1025 * Add program to test if fsync has a delay compared to non-fsync
1033 We could use a fixed row count and a +/- count to follow MVCC
1034 visibility rules, or a single cached value could be used and
1035 invalidated if anyone modifies the table. Another idea is to
1036 get a count directly from a unique index, but for this to be
1037 faster than a sequential scan it must avoid access to the heap
1038 to obtain tuple visibility information.
1040 * Provide a way to calculate an "estimated COUNT(*)"
1042 Perhaps by using the optimizer's cardinality estimates or random
1045 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1047 * Allow data to be pulled directly from indexes
1049 Currently indexes do not have enough tuple visibility information
1050 to allow data to be pulled from the index without also accessing
1051 the heap. One way to allow this is to set a bit on index tuples
1052 to indicate if a tuple is currently visible to all transactions
1053 when the first valid heap lookup happens. This bit would have to
1054 be cleared when a heap tuple is expired.
1056 Another idea is to maintain a bitmap of heap pages where all rows
1057 are visible to all backends, and allow index lookups to reference
1058 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1059 add someday to determine which heap pages need vacuuming. Frequently
1060 accessed bitmaps would have to be stored in shared memory. One 8k
1061 page of bitmaps could track 512MB of heap pages.
1063 A third idea would be for a heap scan to check if all rows are visible
1064 and if so set a per-table flag which can be checked by index scans.
1065 Any change to the table would have to clear the flag. To detect
1066 changes during the heap scan a counter could be set at the start and
1067 checked at the end --- if it is the same, the table has not been
1068 modified --- any table change would increment the counter.
1070 * Consider automatic caching of statements at various levels:
1073 o Query execute plan
1076 * Consider increasing internal areas when shared buffers is increased
1078 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1080 * Consider decreasing the amount of memory used by PrivateRefCount
1082 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1083 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1091 * Improve speed with indexes
1093 For large table adjustments during VACUUM FULL, it is faster to cluster
1094 or reindex rather than update the index. Also, index updates can bloat
1097 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1098 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1100 * Auto-fill the free space map by scanning the buffer cache or by
1101 checking pages written by the background writer
1103 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1104 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1106 * Create a bitmap of pages that need vacuuming
1108 Instead of sequentially scanning the entire table, have the background
1109 writer or some other process record pages that have expired rows, then
1110 VACUUM can look at just those pages rather than the entire table. In
1111 the event of a system crash, the bitmap would probably be invalidated.
1112 One complexity is that index entries still have to be vacuumed, and
1113 doing this without an index scan (by using the heap values to find the
1114 index entry) might be slow and unreliable, especially for user-defined
1117 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1118 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1120 * Allow FSM to return free space toward the beginning of the heap file,
1121 in hopes that empty pages at the end can be truncated by VACUUM
1122 * Allow FSM page return free space based on table clustering, to assist
1123 in maintaining clustering?
1124 * Improve dead row detection during multi-statement transactions usage
1126 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1130 o %Issue log message to suggest VACUUM FULL if a table is nearly
1132 o Improve control of auto-vacuum
1134 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1136 o Prevent long-lived temporary tables from causing frozen-xid
1137 advancement starvation
1139 The problem is that autovacuum cannot vacuum them to set frozen xids;
1140 only the session that created them can do that.
1141 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1147 * Fix priority ordering of read and write light-weight locks (Neil)
1149 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1150 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1152 * Fix problem when multiple subtransactions of the same outer transaction
1153 hold different types of locks, and one subtransaction aborts
1155 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1156 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1157 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1159 * Allow UPDATEs on only non-referential integrity columns not to conflict
1160 with referential integrity locks
1162 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1164 * Add idle_in_transaction_timeout GUC so locks are not held for long
1167 * Improve deadlock detection when a page cleaning lock conflicts
1168 with a shared buffer that is pinned
1170 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1171 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1172 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1176 Startup Time Improvements
1177 =========================
1179 * Experiment with multi-threaded backend for backend creation [thread]
1181 This would prevent the overhead associated with process creation. Most
1182 operating systems have trivial process creation time compared to
1183 database startup overhead, but a few operating systems (Win32,
1184 Solaris) might benefit from threading. Also explore the idea of
1185 a single session using multiple threads to execute a statement faster.
1191 * Eliminate need to write full pages to WAL before page modification [wal]
1193 Currently, to protect against partial disk page writes, we write
1194 full page images to WAL before they are modified so we can correct any
1195 partial page writes during recovery. These pages can also be
1196 eliminated from point-in-time archive files.
1198 o When off, write CRC to WAL and check file system blocks
1201 If CRC check fails during recovery, remember the page in case
1202 a later CRC for that page properly matches.
1204 o Write full pages during file system write and not when
1205 the page is modified in the buffer cache
1207 This allows most full page writes to happen in the background
1208 writer. It might cause problems for applying WAL on recovery
1209 into a partially-written page, but later the full page will be
1212 * Allow WAL traffic to be streamed to another server for stand-by
1214 * Reduce WAL traffic so only modified values are written rather than
1217 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1219 * Allow WAL information to recover corrupted pg_controldata
1221 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1223 * Find a way to reduce rotational delay when repeatedly writing
1226 Currently fsync of WAL requires the disk platter to perform a full
1227 rotation to fsync again. One idea is to write the WAL to different
1228 offsets that might reduce the rotational delay.
1230 * Allow WAL logging to be turned off for a table, but the table
1231 might be dropped or truncated during crash recovery [walcontrol]
1233 Allow tables to bypass WAL writes and just fsync() dirty pages on
1234 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1235 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1236 non-default logging should not use referential integrity with
1237 default-logging tables. A table without dirty buffers during a
1238 crash could perhaps avoid the drop/truncate.
1240 * Allow WAL logging to be turned off for a table, but the table would
1241 avoid being truncated/dropped [walcontrol]
1243 To do this, only a single writer can modify the table, and writes
1244 must happen only on new pages so the new pages can be removed during
1245 crash recovery. Readers can continue accessing the table. Such
1246 tables probably cannot have indexes. One complexity is the handling
1247 of indexes on TOAST tables.
1249 * Speed WAL recovery by allowing more than one page to be prefetched
1251 This involves having a separate process that can be told which pages
1252 the recovery process will need in the near future.
1253 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1257 Optimizer / Executor
1258 ====================
1260 * Improve selectivity functions for geometric operators
1261 * Precompile SQL functions to avoid overhead
1262 * Create utility to compute accurate random_page_cost value
1263 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1264 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1265 actual row counts differ by a specified percentage
1266 * Consider using hash buckets to do DISTINCT, rather than sorting
1268 This would be beneficial when there are few distinct values. This is
1269 already used by GROUP BY.
1271 * Log statements where the optimizer row estimates were dramatically
1272 different from the number of rows actually found?
1273 * Consider compressed annealing to search for query plans
1275 This might replace GEQO, http://sixdemonbag.org/Djinni.
1277 * Improve merge join performance by allowing mark/restore of
1280 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1284 Miscellaneous Performance
1285 =========================
1287 * Do async I/O for faster random read-ahead of data
1289 Async I/O allows multiple I/O requests to be sent to the disk with
1290 results coming back asynchronously.
1292 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1293 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1295 * Use mmap() rather than SYSV shared memory or to write WAL files?
1297 This would remove the requirement for SYSV SHM but would introduce
1298 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1299 to prevent I/O overhead.
1301 * Consider mmap()'ing files into a backend?
1303 Doing I/O to large tables would consume a lot of address space or
1304 require frequent mapping/unmapping. Extending the file also causes
1305 mapping problems that might require mapping only individual pages,
1306 leading to thousands of mappings. Another problem is that there is no
1307 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1308 could hit disk before WAL is written.
1310 * Add a script to ask system configuration questions and tune postgresql.conf
1311 * Consider ways of storing rows more compactly on disk
1313 o Reduce the row header size?
1314 o Consider reducing on-disk varlena length from four bytes to
1315 two because a heap row cannot be more than 64k in length
1317 * Consider increasing NUM_CLOG_BUFFERS
1318 * Consider having the background writer update the transaction status
1319 hint bits before writing out the page
1321 Implementing this requires the background writer to have access to system
1322 catalogs and the transaction status log.
1324 * Allow user configuration of TOAST thresholds
1326 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1328 * Allow configuration of backend priorities via the operating system
1330 Though backend priorities make priority inversion during lock
1331 waits possible, research shows that this is not a huge problem.
1333 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1335 * Experiment with multi-threaded backend better resource utilization
1337 This would allow a single query to make use of multiple CPU's or
1338 multiple I/O channels simultaneously. One idea is to create a
1339 background reader that can pre-fetch sequential and index scan
1340 pages needed by other backends. This could be expanded to allow
1341 concurrent reads from multiple devices in a partitioned table.
1347 * Add use of 'const' for variables in source tree
1348 * Move some things from contrib into main tree
1349 * %Remove warnings created by -Wcast-align
1350 * Move platform-specific ps status display info from ps_status.c to ports
1351 * Add optional CRC checksum to heap and index pages
1352 * Improve documentation to build only interfaces (Marc)
1353 * Remove or relicense modules that are not under the BSD license, if possible
1354 * Acquire lock on a relation before building a relcache entry for it
1355 * Allow cross-compiling by generating the zic database on the target system
1356 * Improve NLS maintenance of libpgport messages linked onto applications
1357 * Clean up casting in contrib/isn
1359 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1361 * Use UTF8 encoding for NLS messages so all server encodings can
1363 * Update Bonjour to work with newer cross-platform SDK
1365 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1366 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1368 * Consider detoasting keys before sorting
1369 * Consider GnuTLS if OpenSSL license becomes a problem
1371 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1372 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1374 * Consider changing documentation format from SGML to XML
1376 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1378 * Consider making NAMEDATALEN more configurable in future releases
1379 * Update our code to handle 64-bit timezone files to match the zic
1380 source code, which now uses them
1381 * Have configure choose integer datetimes by default
1383 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1385 * Support scoped IPv6 addresses
1387 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1389 * Consider allowing 64-bit integers and floats to be passed by value on
1392 Also change 32-bit floats (float4) to be passed by value at the same
1399 o Remove configure.in check for link failure when cause is found
1400 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1402 o Remove psql newline patch when we find out why mingw outputs an
1404 o Allow psql to use readline once non-US code pages work with
1406 o Fix problem with shared memory on the Win32 Terminal Server
1407 o Improve signal handling
1409 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1413 * Wire Protocol Changes
1415 o Allow dynamic character set handling
1416 o Add decoded type, length, precision
1418 o Update clients to use data types, typmod, schema.table.column names
1419 of result sets using new statement protocol
1425 * Add pre-parsing phase that converts non-ISO syntax to supported
1428 This could allow SQL written for other databases to run without
1431 * Allow plug-in modules to emulate features from other databases
1432 * Add features of Oracle-style packages (Pavel)
1434 A package would be a schema with session-local variables,
1435 public/private functions, and initialization functions. It
1436 is also possible to implement these capabilities
1437 in any schema and not use a separate "packages"
1440 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1442 * Consider allowing control of upper/lower case folding of unquoted
1445 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1446 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1448 * Add autonomous transactions
1450 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1454 Features We Do _Not_ Want
1455 =========================
1457 * All backends running as threads in a single process (not wanted)
1459 This eliminates the process protection we get from the current setup.
1460 Thread creation is usually the same overhead as process creation on
1461 modern systems, so it seems unwise to use a pure threaded model.
1463 * Optimizer hints (not wanted)
1465 Optimizer hints are used to work around problems in the optimizer. We
1466 would rather have the problems reported and fixed.
1468 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1469 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1470 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1473 Because we support postfix operators, it isn't possible to make AS
1474 optional and continue to use bison.
1475 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1477 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1479 * Embedded server (not wanted)
1481 While PostgreSQL clients runs fine in limited-resource environments, the
1482 server requires multiple processes and a stable pool of resources to
1483 run reliabily and efficiently. Stripping down the PostgreSQL server
1484 to run in the same process address space as the client application
1485 would add too much complexity and failure cases.