3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Mon Mar 24 14:24:24 EDT 2008
6 The most recent version of this document can be viewed at
7 http://www.postgresql.org/docs/faqs.TODO.html.
9 #A hyphen, "-", marks changes that will appear in the upcoming 8.3 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
32 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00218.php
34 * Check for unreferenced table files created by transactions that were
35 in-progress when the server terminated abruptly
37 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
39 * Set proper permissions on non-system schemas during db creation
41 Currently all schemas are owned by the super-user because they are copied
42 from the template1 database. However, since all objects are inherited
43 from the template database, it is not clear that setting schemas to the db
46 * Add function to report the time of the most recent server reload
47 * Allow statistics collector information to be pulled from the collector
48 process directly, rather than requiring the collector to write a
49 filesystem file twice a second?
50 * Reduce file system activity overhead of statistics file pgstat.stat
52 http://archives.postgresql.org/pgsql-general/2007-12/msg00106.php
54 * Allow statistics last vacuum/analyze execution times to be displayed
55 without requiring stats_row_level to be enabled
57 http://archives.postgresql.org/pgsql-docs/2007-04/msg00028.php
59 * Allow log_min_messages to be specified on a per-module basis
61 This would allow administrators to see more detailed information from
62 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
63 Another idea is to allow separate configuration files for each module,
64 or allow arbitrary SET commands to be passed to them.
66 * Simplify ability to create partitioned tables
68 This would allow creation of partitioned tables without requiring
69 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
70 for rapid partition selection. Options could include range and hash
73 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
74 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
75 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00028.php
76 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00248.php
77 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00387.php
78 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00413.php
80 * Allow auto-selection of partitioned tables for min/max() operations
81 * Allow more complex user/database default GUC settings
83 Currently ALTER USER and ALTER DATABASE support per-user and
84 per-database defaults. Consider adding per-user-and-database
85 defaults so things like search_path can be defaulted for a
86 specific user connecting to a specific database.
88 * Allow custom variable classes that can restrict who can set the values
90 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
92 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
93 the privilege granted by the invoking role, and not those granted
96 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
98 * Allow SSL authentication/encryption over unix domain sockets
100 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
102 * Allow SSL key file permission checks to be optionally disabled when
103 sharing SSL keys with other applications
105 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00069.php
107 * Allow client certificate names to be checked against the client
110 This is already implemented in
111 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
114 * Configuration files
116 o Allow pg_hba.conf to specify host names along with IP addresses
118 Host name lookup could occur when the postmaster reads the
119 pg_hba.conf file, or when the backend starts. Another
120 solution would be to reverse lookup the connection IP and
121 check that hostname against the host names in pg_hba.conf.
122 We could also then check that the host name maps to the IP
125 o %Allow postgresql.conf file values to be changed via an SQL
126 API, perhaps using SET GLOBAL
127 o Allow the server to be stopped/restarted via an SQL API
128 o Issue a warning if a change-on-restart-only postgresql.conf value
129 is modified and the server config files are reloaded
130 o Consider normalizing fractions in postgresql.conf, perhaps
133 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00550.php
135 o Allow Kerberos to disable stripping of realms so we can
136 check the username@realm against multiple realms
138 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00009.php
143 o Allow a database in tablespace t1 with tables created in
144 tablespace t2 to be used as a template for a new database created
145 with default tablespace t2
147 Currently all objects in the default database tablespace must
148 have default tablespace specifications. This is because new
149 databases are created by copying directories. If you mix default
150 tablespace tables and tablespace-specified tables in the same
151 directory, creating a new database from such a mixed directory
152 would create a new database with tables that had incorrect
153 explicit tablespaces. To fix this would require modifying
154 pg_class in the newly copied database, which we don't currently
157 o Allow reporting of which objects are in which tablespaces
159 This item is difficult because a tablespace can contain objects
160 from multiple databases. There is a server-side function that
161 returns the databases which use a specific tablespace, so this
162 requires a tool that will call that function and connect to each
163 database to find the objects in each database for that tablespace.
165 o Allow WAL replay of CREATE TABLESPACE to work when the directory
166 structure on the recovery computer is different from the original
168 o Allow per-tablespace quotas
171 * Point-In-Time Recovery (PITR)
173 o Allow a warm standby system to also allow read-only statements
176 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
178 o %Create dump tool for write-ahead logs for use in determining
179 transaction id for point-in-time recovery
181 This is useful for checking PITR recovery.
183 o Allow recovery.conf to support the same syntax as
184 postgresql.conf, including quoting
186 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
188 o Fix server restart problem when the server was shutdown during
191 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00800.php
193 o Recreate pg_xlog/archive_status/ if it doesn't exist after
194 restoring from a PITR backup
196 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00487.php
202 * Change NUMERIC to enforce the maximum precision
203 * Reduce storage space for small NUMERICs
205 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
206 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
207 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
209 * Fix data types where equality comparison isn't intuitive, e.g. box
210 * Add support for public SYNONYMs
212 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
214 * Fix CREATE CAST on DOMAINs
216 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
217 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
219 * Allow domains to be cast
221 http://archives.postgresql.org/pgsql-hackers/2003-06/msg01206.php
222 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00289.php
224 * Add support for SQL-standard GENERATED/IDENTITY columns
226 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
227 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
228 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
229 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
230 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
232 * Improve XML support
234 http://developer.postgresql.org/index.php/XML_Support
236 * Consider placing all sequences in a single table, or create a system
239 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
241 * Allow the UUID type to accept non-standard formats
243 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
245 * Allow text search dictionary to filter out only stop words
247 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
249 * Consider a function-based API for '@@' full text searches
251 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
253 * Improve text search error messages
255 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00966.php
256 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01146.php
258 * Consider a special data type for regular expressions
260 http://archives.postgresql.org/pgsql-hackers/2007-08/msg01067.php
262 * Reduce BIT data type overhead using short varlena headers
264 http://archives.postgresql.org/pgsql-general/2007-12/msg00273.php
269 o Allow infinite dates and intervals just like infinite timestamps
270 o Merge hardwired timezone names with the TZ database; allow either
271 kind everywhere a TZ name is currently taken
272 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
273 information, either zone name or offset from UTC [timezone]
275 If the TIMESTAMP value is stored with a time zone name, interval
276 computations should adjust based on the time zone rules.
278 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
279 o Add a GUC variable to allow output of interval values in ISO8601
281 o Have timestamp subtraction not call justify_hours()?
283 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
285 o Improve timestamptz subtraction to be DST-aware
287 Currently subtracting one date from another that crosses a
288 daylight savings time adjustment can return '1 day 1 hour', but
289 adding that back to the first date returns a time one hour in
290 the future. This is caused by the adjustment of '25 hours' to
291 '1 day 1 hour', and '1 day' is the same time the next day, even
292 if daylight savings adjustments are involved.
294 o Fix interval display to support values exceeding 2^31 hours
295 o Add overflow checking to timestamp and interval arithmetic
296 o Extend timezone code to allow 64-bit values so we can
297 represent years beyond 2038
299 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
301 o Use LC_TIME for localized weekday/month names, rather than
304 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
306 o Add ISO INTERVAL handling
308 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
309 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
311 o Support ISO INTERVAL syntax if units cannot be determined from
312 the string, and are supplied after the string
314 The SQL standard states that the units after the string
315 specify the units of the string, e.g. INTERVAL '2' MINUTE
316 should return '00:02:00'. The current behavior has the units
317 restrict the interval value to the specified unit or unit
318 range, INTERVAL '70' SECOND returns '00:00:10'.
320 For syntax that isn't uniquely ISO or PG syntax, like '1' or
321 '1:30', treat as ISO if there is a range specification clause,
322 and as PG if there no clause is present, e.g. interpret '1:30'
323 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
324 '1:30' as '1 hour, 30 minutes'.
326 This makes common cases like SELECT INTERVAL '1' MONTH
327 SQL-standard results. The SQL standard supports a limited
328 number of unit combinations and doesn't support unit names in
329 the string. The PostgreSQL syntax is more flexible in the
330 range of units supported, e.g. PostgreSQL supports '1 year 1
331 hour', while the SQL standard does not.
333 o Add support for year-month syntax, INTERVAL '50-6' YEAR
335 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
336 year' AS INTERVAL MONTH), and this should return '12 months'
337 o Round or truncate values to the requested precision, e.g.
338 INTERVAL '11 months' AS YEAR should return one or zero
339 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
344 o Delay resolution of array expression's data type so assignment
345 coercion can be performed on empty array expressions
346 o Add support for arrays of domains
348 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
350 o Allow single-byte header storage for array elements
355 o Improve vacuum of large objects, like contrib/vacuumlo?
356 o Add security checking for large objects
357 o Auto-delete large objects when referencing row is deleted
359 contrib/lo offers this functionality.
361 o Allow read/write into TOAST values like large objects
363 This requires the TOAST column to be stored EXTERNAL.
365 o Add API for 64-bit large object access
367 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
371 * Add locale-aware MONEY type, and support multiple currencies
373 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
374 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
376 * MONEY dumps in a locale-specific format making it difficult to
377 restore to a system with a different locale
378 * Allow MONEY to be easily cast to/from other numeric data types
384 * Allow INET subnet tests using non-constants to be indexed
385 * Allow to_date() and to_timestamp() accept localized month names
386 * Fix to_date()-related functions to consistently issue errors
388 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
390 * Add missing parameter handling in to_char()
392 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
394 * Allow substring/replace() to get/set bit values
395 * Allow to_char() on interval values to accumulate the highest unit
398 Some special format flag would be required to request such
399 accumulation. Such functionality could also be added to EXTRACT.
400 Prevent accumulation that crosses the month/day boundary because of
401 the uneven number of days in a month.
403 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
404 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
405 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
406 o to_char(INTERVAL '3 years 5 months','MM') => 41
408 * Implement inlining of set-returning functions defined in SQL
409 * Allow SQL-language functions to return results from RETURNING queries
411 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
413 * Allow SQL-language functions to reference parameters by parameter name
415 Currently SQL-language functions can only refer to dollar parameters,
418 * Add SPI_gettypmod() to return the typemod for a TupleDesc
419 * Enforce typmod for function inputs, function results and parameters for
420 spi_prepare'd statements called from PLs
422 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
424 * Allow holdable cursors in SPI
425 * Tighten function permission checks
427 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
429 * Fix IS OF so it matches the ISO specification, and add documentation
431 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
432 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
434 * Add missing operators for geometric data types
436 Some geometric types do not have the full suite of geometric operators,
439 * Implement Boyer-Moore searching in strpos()
441 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
443 * Prevent malicious functions from being executed with the permissions
444 of unsuspecting users
446 Index functions are safe, so VACUUM and ANALYZE are safe too.
447 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
448 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
450 * Reduce memory usage of aggregates in set returning functions
452 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
454 * Add temporal versions of generate_series()
456 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
458 * Add array_accum() and array_to_set() functions for arrays
460 The standards specify array_agg() and UNNEST.
461 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
463 * Fix /contrib/ltree operator
465 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
467 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
469 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
473 Multi-Language Support
474 ======================
476 * Add NCHAR (as distinguished from ordinary varchar),
477 * Allow locale to be set at database creation
479 Currently locale can only be set during initdb. No global tables have
480 locale-aware columns. However, the database template used during
481 database creation might have locale-aware indexes. The indexes would
482 need to be reindexed to match the new locale.
484 * Allow encoding on a per-column basis optionally using the ICU library:
486 Right now only one encoding is allowed per database. [locale]
488 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
489 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
490 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
491 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
493 * Add CREATE COLLATE? [locale]
494 * Support multiple simultaneous character sets, per SQL92
495 * Improve UTF8 combined character handling?
496 * Add octet_length_server() and octet_length_client()
497 * Make octet_length_client() the same as octet_length()?
498 * Fix problems with wrong runtime encoding conversion for NLS message files
499 * Add URL to more complete multi-byte regression tests
501 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
503 * Fix ILIKE and regular expressions to handle case insensitivity
504 properly in multibyte encodings
506 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
507 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
509 * Set client encoding based on the client operating system encoding
511 Currently client_encoding is set in postgresql.conf, which
512 defaults to the server encoding.
513 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
515 * Change memory allocation for multi-byte functions so memory is
516 allocated inside conversion functions
518 Currently we preallocate memory based on worst-case usage.
525 * Automatically create rules on views so they are updateable, per SQL99
527 We can only auto-create rules for simple views. For more complex
528 cases users will still have to write rules manually.
530 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
531 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
533 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
534 * Allow NOTIFY in rules involving conditionals
535 * Allow VIEW/RULE recompilation when the underlying tables change
537 Another issue is whether underlying table changes should be reflected
538 in the view, e.g. should SELECT * show additional columns if they
539 are added after the view is created.
541 * Make it possible to use RETURNING together with conditional DO INSTEAD
542 rules, such as for partitioning setups
544 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
546 * Add the ability to automatically create materialized views
548 Right now materialized views require the user to create triggers on the
549 main table to keep the summary table current. SQL syntax should be able
550 to manager the triggers and summary table automatically. A more
551 sophisticated implementation would automatically retrieve from the
552 summary table when the main table is referenced, if possible.
559 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
560 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
561 * %Allow SET CONSTRAINTS to be qualified by schema/table name
562 * %Add a separate TRUNCATE permission
564 Currently only the owner can TRUNCATE a table because triggers are not
565 called, and the table is locked in exclusive mode.
567 * Allow PREPARE of cursors
568 * Allow finer control over the caching of prepared query plans
570 Currently queries prepared via the libpq API are planned on first
571 execute using the supplied parameters --- allow SQL PREPARE to do the
572 same. Also, allow control over replanning prepared queries either
573 manually or automatically when statistics for execute parameters
574 differ dramatically from those used during planning.
576 * Improve logging of prepared transactions recovered during startup
578 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
580 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
582 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
583 such information in memory would improve performance.
585 * Add optional textual message to NOTIFY
587 This would allow an informational message to be added to the notify
588 message, perhaps indicating the row modified or other custom
591 * Allow multiple identical NOTIFY events to always be communicated to the
592 client, rather than sent as a single notification to the listener
594 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
596 * Add a GUC variable to warn about non-standard SQL usage in queries
597 * Add SQL-standard MERGE command, typically used to merge two tables
600 This is similar to UPDATE, then for unmatched rows, INSERT.
601 Whether concurrent access allows modifications which could cause
602 row loss is implementation independent.
604 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
607 To implement this cleanly requires that the table have a unique index
608 so duplicate checking can be easily performed. It is possible to
609 do it without a unique index if we require the user to LOCK the table
612 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
613 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
615 * Add NOVICE output level for helpful messages like automatic sequence/index
617 * Add GUC to issue notice about statements that use unjoined tables
618 * Allow EXPLAIN to identify tables that were skipped because of
620 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
621 * Enable standard_conforming_strings
622 * Make standard_conforming_strings the default in 8.5?
624 When this is done, backslash-quote should be prohibited in non-E''
625 strings because of possible confusion over how such strings treat
626 backslashes. Basically, '' is always safe for a literal single
627 quote, while \' might or might not be based on the backslash
630 * Simplify dropping roles that have objects in several databases
631 * Allow COMMENT ON to accept an expression rather than just a string
632 * Allow the count returned by SELECT, etc to be represented as an int64
633 to allow a higher range of values
634 * Add SQL99 WITH clause to SELECT
635 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
637 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
638 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
639 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
640 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
642 * Add DEFAULT .. AS OWNER so permission checks are done as the table
645 This would be useful for SERIAL nextval() calls and CHECK constraints.
647 * Allow DISTINCT to work in multiple-argument aggregate calls
648 * Add column to pg_stat_activity that shows the progress of long-running
649 commands like CREATE INDEX and VACUUM
650 * Implement SQL:2003 window functions
651 * Improve failure message when DROP DATABASE is used on a database that
652 has prepared transactions
653 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
655 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
656 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
658 * Increase locking when DROPing objects so dependent objects cannot
659 get dropped while the DROP operation is happening
661 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
663 * -Allow AS in "SELECT col AS label" to be optional in certain cases
665 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
666 RETURNING to supply values to the INSERT
667 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
669 * Add comments on system tables/columns using the information in
672 Ideally the information would be pulled from the SGML file
675 * Improve reporting of UNION type mismatches
677 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
678 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
683 o Allow CREATE TABLE AS to determine column lengths for complex
684 expressions like SELECT col1 || col2
686 o Have WITH CONSTRAINTS also create constraint indexes
688 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
690 o Have CONSTRAINT cname NOT NULL record the contraint name
692 Right now pg_attribute.attnotnull records the NOT NULL status
693 of the column, but does not record the contraint name
698 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
700 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
701 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
702 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
703 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
705 o Research self-referential UPDATEs that see inconsistent row versions
706 in read-committed mode
708 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
709 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
711 o Allow GLOBAL temporary tables to exist as empty by default in
714 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
719 o Have ALTER TABLE RENAME rename SERIAL sequence names
721 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
723 o Have ALTER SEQUENCE RENAME rename the sequence name stored
724 in the sequence table
726 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
727 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
728 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
730 o Add ALTER DOMAIN to modify the underlying data type
731 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
733 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
735 o %Allow ALTER TABLE to change constraint deferrability and actions
736 o Add missing object types for ALTER ... SET SCHEMA
737 o Allow ALTER TABLESPACE to move to different directories
738 o Allow databases to be moved to different tablespaces
739 o Allow moving system tables to other tablespaces, where possible
741 Currently non-global system tables must be in the default database
742 tablespace. Global system tables can never be moved.
744 o Prevent parent tables from altering or dropping constraints
745 like CHECK that are inherited by child tables unless CASCADE
747 o %Prevent child tables from altering or dropping constraints
748 like CHECK that were inherited from the parent table
749 o Have ALTER INDEX update the name of a constraint using that index
750 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
751 o Allow column display reordering by recording a display,
752 storage, and permanent id for every column?
754 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
760 o Automatically maintain clustering on a table
762 This might require some background daemon to maintain clustering
763 during periods of low usage. It might also require tables to be only
764 partially filled for easier reorganization. Another idea would
765 be to create a merged heap/index data file so an index lookup would
766 automatically access the heap data too. A third idea would be to
767 store heap rows in hashed groups, perhaps using a user-supplied
769 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
771 o %Add default clustering to system tables
773 To do this, determine the ideal cluster index for each system
774 table and set the cluster setting during initdb.
776 o %Add VERBOSE option to report tables as they are processed,
782 o Allow COPY to report error lines and continue
784 This requires the use of a savepoint before each COPY line is
785 processed, with ROLLBACK on COPY failure.
786 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
788 o Allow COPY on a newly-created table to skip WAL logging
790 On crash recovery, the table involved in the COPY would
791 be removed or have its heap and index files truncated. One
792 issue is that no other backend should be able to add to
793 the table at the same time, which is something that is
794 currently allowed. This currently is done if the table is
795 created inside the same transaction block as the COPY because
796 no other backends can see the table.
798 o Consider using a ring buffer for COPY FROM
800 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
801 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
803 o Allow COPY FROM to create index entries in bulk
805 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
807 o Allow COPY in CSV mode to control whether a quoted zero-length
808 string is treated as NULL
810 Currently this is always treated as a zero-length string,
811 which generates an error when loading into an integer column
812 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
818 o Allow column-level privileges
819 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
822 The proposed syntax is:
823 GRANT SELECT ON ALL TABLES IN public TO phpuser;
824 GRANT SELECT ON NEW TABLES IN public TO phpuser;
826 o Allow GRANT/REVOKE permissions to be inherited by objects based on
829 o Allow SERIAL sequences to inherit permissions from the base table?
834 o Prevent DROP TABLE from dropping a row referenced by its own open
840 o Allow INSERT/UPDATE of the system-generated oid value for a row
841 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
847 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
852 Referential Integrity
853 =====================
855 * Add MATCH PARTIAL referential integrity
856 * Change foreign key constraint for array -> element to mean element
858 * Fix problem when cascading referential triggers make changes on
859 cascaded tables, seeing the tables in an intermediate state
861 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
862 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
864 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
866 This would allow UPDATE tab SET col = col + 1 to work if col has
867 a unique index. Currently, uniqueness checks are done while the
868 command is being executed, rather than at the end of the statement
870 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
871 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
873 * Optimize referential integrity checks
875 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
876 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
879 Server-Side Languages
880 =====================
883 o Fix RENAME to work on variables other than OLD/NEW
885 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
886 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
887 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
889 o Allow function parameters to be passed by name,
890 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
891 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
892 o Allow listing of record column names, and access to
893 record columns via variables, e.g. columns := r.(*),
896 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
897 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
898 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
900 o Add support for SCROLL cursors
901 o Add support for WITH HOLD cursors
902 o Allow row and record variables to be set to NULL constants,
903 and allow NULL tests on such variables
905 Because a row is not scalar, do not allow assignment
906 from NULL-valued scalars.
908 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
910 o Review handling of MOVE and FETCH
912 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
914 o Improve logic of determining if an identifier is a a
915 variable or column name
917 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
921 o Add table function support to pltcl, plpythonu
922 o Add support for polymorphic arguments and return types to
923 languages other than PL/PgSQL
924 o Add capability to create and call PROCEDURES
925 o Add support for OUT and INOUT parameters to languages other
927 o Add PL/PythonU tracebacks
929 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
931 o Allow data to be passed in native language formats, rather
934 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
941 * Have pg_ctl look at PGHOST in case it is a socket directory?
942 * Allow pg_ctl to work properly with configuration files located outside
945 pg_ctl can not read the pid file because it isn't located in the
946 config directory but in the PGDATA directory. The solution is to
947 allow pg_ctl to read and understand postgresql.conf to find the
948 data_directory value.
950 * Add a function like pg_get_indexdef() that report more detailed index
953 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
955 * Prefix command-line utilities like createuser with 'pg_'
957 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00025.php
961 o Have psql show current values for a sequence
962 o Move psql backslash database information into the backend, use
963 mnemonic commands? [psql]
965 This would allow non-psql clients to pull the same information out
966 of the database as psql.
968 o Fix psql's \d commands more consistent
970 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
971 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
973 o Consistently display privilege information for all objects in psql
974 o Add auto-expanded mode so expanded output is used if the row
975 length is wider than the screen width.
977 Consider using auto-expanded mode for backslash commands like \df+.
979 o Prevent tab completion of SET TRANSACTION from querying the
980 database and therefore preventing the transaction isolation
981 level from being set.
983 Currently SET <tab> causes a database lookup to check all
984 supported session variables. This query causes problems
985 because setting the transaction isolation level must be the
986 first statement of a transaction.
988 o Add a \set variable to control whether \s displays line numbers
990 Another option is to add \# which lists line numbers, and
991 allows command execution.
993 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
995 o Prevent escape string warnings when object names have
998 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1000 o Have \d show foreign keys that reference a table's primary key
1002 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1004 o Have \d show child tables that inherit from the specified parent
1005 o Have \l+ show database size, if permissions allow
1007 Ideally it will not generate an error for invalid permissions
1009 o Include the symbolic SQLSTATE name in verbose error reports
1011 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1014 * pg_dump / pg_restore
1015 o %Add dumping of comments on index columns and composite type columns
1016 o %Add full object name to the tag field. eg. for operators we need
1017 '=(integer, integer)', instead of just '='.
1018 o Add pg_dumpall custom format dumps?
1019 o Allow selection of individual object(s) of all types, not just
1021 o In a selective dump, allow dumping of an object and all its
1023 o Add options like pg_restore -l and -L to pg_dump
1024 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1025 o Allow pg_dump --clean to drop roles that own objects or have
1027 o Change pg_dump so that a comment on the dumped database is
1028 applied to the loaded database, even if the database has a
1029 different name. This will require new backend syntax, perhaps
1030 COMMENT ON CURRENT DATABASE.
1031 o Remove unnecessary function pointer abstractions in pg_dump source
1033 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1034 multiple objects simultaneously
1036 The difficulty with this is getting multiple dump processes to
1037 produce a single dump output file. It also would require
1038 several sessions to share the same snapshot.
1039 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1041 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1042 restoring multiple objects simultaneously
1044 This might require a pg_restore flag to indicate how many
1045 simultaneous operations should be performed. Only pg_dump's
1046 -Fc format has the necessary dependency information.
1047 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1049 o To better utilize resources, allow pg_restore to check foreign
1050 keys simultaneously, where possible
1051 o Allow pg_restore to create all indexes of a table
1052 concurrently, via a single heap scan
1054 This requires a pg_dump -Fc file because that format contains
1055 the required dependency information.
1056 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1058 o Allow pg_restore to load different parts of the COPY data
1060 o Prevent pg_dump/pg_restore from being affected by
1063 Using psql to restore a pg_dump dump is also affected.
1065 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1066 o Allow pre/data/post files when schema and data are dumped
1067 separately, for performance reasons
1069 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1076 Document differences between ecpg and the SQL standard and
1077 information about the Informix-compatibility module.
1079 o Solve cardinality > 1 for input descriptors / variables?
1080 o Add a semantic check level, e.g. check if a table really exists
1081 o fix handling of DB attributes that are arrays
1082 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1084 o Fix nested C comments
1085 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1086 o Make SET CONNECTION thread-aware, non-standard?
1087 o Allow multidimensional arrays
1088 o Add internationalized message strings
1089 o Implement COPY FROM STDIN
1093 o Add PQescapeIdentifierConn()
1094 o Prevent PQfnumber() from lowercasing unquoted the column name
1096 PQfnumber() should never have been doing lowercasing, but
1097 historically it has so we need a way to prevent it
1099 o Allow statement results to be automatically batched to the client
1101 Currently all statement results are transferred to the libpq
1102 client before libpq makes the results available to the
1103 application. This feature would allow the application to make
1104 use of the first result rows while the rest are transferred, or
1105 held on the server waiting for them to be requested by libpq.
1106 One complexity is that a statement like SELECT 1/col could error
1107 out mid-way through the result set.
1109 o Consider disallowing multiple queries in PQexec() as an
1110 additional barrier to SQL injection attacks
1112 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1114 o Add PQexecf() that allows complex parameter substitution
1116 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1118 o Add SQLSTATE severity to PGconn return status
1120 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1126 * Add deferred trigger queue file
1128 Right now all deferred trigger information is stored in backend
1129 memory. This could exhaust memory for very large trigger queues.
1130 This item involves dumping large queues into files.
1132 * Allow triggers to be disabled in only the current session.
1134 This is currently possible by starting a multi-statement transaction,
1135 modifying the system tables, performing the desired SQL, restoring the
1136 system tables, and committing the transaction. ALTER TABLE ...
1137 TRIGGER requires a table lock so it is not ideal for this usage.
1139 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1141 If the dump is known to be valid, allow foreign keys to be added
1142 without revalidating the data.
1144 * Allow statement-level triggers to access modified rows
1145 * Support triggers on columns
1147 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1149 * Allow AFTER triggers on system tables
1151 System tables are modified in many places in the backend without going
1152 through the executor and therefore not causing triggers to fire. To
1153 complete this item, the functions that modify system tables will have
1156 * Tighten trigger permission checks
1158 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1160 * Allow BEFORE INSERT triggers on views
1162 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1164 * Add ability to trigger on TRUNCATE
1166 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
1173 * Add UNIQUE capability to non-btree indexes
1174 * Prevent index uniqueness checks when UPDATE does not modify the column
1176 Uniqueness (index) checks are done when updating a column even if the
1177 column is not modified by the UPDATE.
1179 * Allow the creation of on-disk bitmap indexes which can be quickly
1180 combined with other bitmap indexes
1182 Such indexes could be more compact if there are only a few distinct values.
1183 Such indexes can also be compressed. Keeping such indexes updated can be
1186 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1187 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1189 * Allow accurate statistics to be collected on indexes with more than
1190 one column or expression indexes, perhaps using per-index statistics
1192 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1193 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1195 * Consider increasing the number of default statistics target, and
1196 reduce statistics target overhead
1198 Also consider having a larger statistics target for indexed columns
1199 and expression indexes
1200 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1201 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1202 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1204 * Consider compressing indexes by storing key values duplicated in
1205 several rows as a single index entry
1207 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1208 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1209 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1210 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1212 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1214 This is difficult because you must upgrade to an exclusive table lock
1215 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1216 have this complication. This would allow index compaction without
1218 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1220 * Allow multiple indexes to be created concurrently, ideally via a
1221 single heap scan, and have pg_restore use it
1227 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1228 and primary/foreign keys
1229 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1230 on inherited table, e.g. INSERT INTO inherit_table
1231 (unique_index_col) VALUES (dup) should fail
1233 The main difficulty with this item is the problem of
1234 creating an index that can span multiple tables.
1236 o Allow SELECT ... FOR UPDATE on inherited tables
1237 o Require all CHECK constraints to be inherited
1239 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1241 o Add checks to prevent a CREATE RULE views on inherited tables
1243 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1244 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1249 o Add more GIST index support for geometric data types
1250 o Allow GIST indexes to create certain complex index types, like
1251 digital trees (see Aoki)
1256 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1258 o Pack hash index buckets onto disk pages more efficiently
1260 Currently only one hash bucket can be stored on a page. Ideally
1261 several hash buckets could be stored on a single page and greater
1262 granularity used for the hash algorithm.
1264 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1266 o Consider sorting hash buckets so entries can be found using a
1267 binary search, rather than a linear scan
1268 o In hash indexes, consider storing the hash value with or instead
1270 o Add WAL logging for crash recovery
1271 o Allow multi-column hash indexes
1272 o -During index creation, pre-sort the tuples to improve build speed
1279 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1281 Ideally this requires a separate test program that can be run
1282 at initdb time or optionally later. Consider O_SYNC when
1285 * Add program to test if fsync has a delay compared to non-fsync
1286 * Consider sorting writes during checkpoint
1288 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1297 We could use a fixed row count and a +/- count to follow MVCC
1298 visibility rules, or a single cached value could be used and
1299 invalidated if anyone modifies the table. Another idea is to
1300 get a count directly from a unique index, but for this to be
1301 faster than a sequential scan it must avoid access to the heap
1302 to obtain tuple visibility information.
1304 * Provide a way to calculate an "estimated COUNT(*)"
1306 Perhaps by using the optimizer's cardinality estimates or random
1309 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1311 * Allow data to be pulled directly from indexes
1313 Currently indexes do not have enough tuple visibility information
1314 to allow data to be pulled from the index without also accessing
1315 the heap. One way to allow this is to set a bit on index tuples
1316 to indicate if a tuple is currently visible to all transactions
1317 when the first valid heap lookup happens. This bit would have to
1318 be cleared when a heap tuple is expired.
1320 Another idea is to maintain a bitmap of heap pages where all rows
1321 are visible to all backends, and allow index lookups to reference
1322 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1323 add someday to determine which heap pages need vacuuming. Frequently
1324 accessed bitmaps would have to be stored in shared memory. One 8k
1325 page of bitmaps could track 512MB of heap pages.
1327 A third idea would be for a heap scan to check if all rows are visible
1328 and if so set a per-table flag which can be checked by index scans.
1329 Any change to the table would have to clear the flag. To detect
1330 changes during the heap scan a counter could be set at the start and
1331 checked at the end --- if it is the same, the table has not been
1332 modified --- any table change would increment the counter.
1334 * Consider automatic caching of statements at various levels:
1337 o Query execute plan
1340 * Consider increasing internal areas when shared buffers is increased
1342 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1344 * Consider decreasing the amount of memory used by PrivateRefCount
1346 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1347 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1349 * Consider allowing higher priority queries to have referenced buffer
1350 cache pages stay in memory longer
1352 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1359 * Improve speed with indexes
1361 For large table adjustments during VACUUM FULL, it is faster to cluster
1362 or reindex rather than update the index. Also, index updates can bloat
1365 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1366 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1367 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1369 * Auto-fill the free space map by scanning the buffer cache or by
1370 checking pages written by the background writer
1372 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1373 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1375 * Create a bitmap of pages that need vacuuming
1377 Instead of sequentially scanning the entire table, have the background
1378 writer or some other process record pages that have expired rows, then
1379 VACUUM can look at just those pages rather than the entire table. In
1380 the event of a system crash, the bitmap would probably be invalidated.
1381 One complexity is that index entries still have to be vacuumed, and
1382 doing this without an index scan (by using the heap values to find the
1383 index entry) might be slow and unreliable, especially for user-defined
1386 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1387 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1389 * Allow FSM to return free space toward the beginning of the heap file,
1390 in hopes that empty pages at the end can be truncated by VACUUM
1391 * Allow FSM page return free space based on table clustering, to assist
1392 in maintaining clustering?
1393 * Improve dead row detection during multi-statement transactions usage
1395 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1397 * Consider a more compact data representation for dead tuples
1399 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1404 o %Issue log message to suggest VACUUM FULL if a table is nearly
1406 o Improve control of auto-vacuum
1408 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1410 o Prevent long-lived temporary tables from causing frozen-xid
1411 advancement starvation
1413 The problem is that autovacuum cannot vacuum them to set frozen xids;
1414 only the session that created them can do that.
1415 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1417 o Store per-table autovacuum settings in pg_class.reloptions.
1419 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1420 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1422 o Prevent autovacuum from running if an old transaction is still
1423 running from the last vacuum
1425 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1432 * Fix priority ordering of read and write light-weight locks (Neil)
1434 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1435 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1437 * Fix problem when multiple subtransactions of the same outer transaction
1438 hold different types of locks, and one subtransaction aborts
1440 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1441 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1442 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1443 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1445 * Allow UPDATEs on only non-referential integrity columns not to conflict
1446 with referential integrity locks
1448 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1450 * Add idle_in_transaction_timeout GUC so locks are not held for long
1453 * Improve deadlock detection when a page cleaning lock conflicts
1454 with a shared buffer that is pinned
1456 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1457 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1458 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1463 Startup Time Improvements
1464 =========================
1466 * Experiment with multi-threaded backend for backend creation [thread]
1468 This would prevent the overhead associated with process creation. Most
1469 operating systems have trivial process creation time compared to
1470 database startup overhead, but a few operating systems (Win32,
1471 Solaris) might benefit from threading. Also explore the idea of
1472 a single session using multiple threads to execute a statement faster.
1479 * Eliminate need to write full pages to WAL before page modification [wal]
1481 Currently, to protect against partial disk page writes, we write
1482 full page images to WAL before they are modified so we can correct any
1483 partial page writes during recovery. These pages can also be
1484 eliminated from point-in-time archive files.
1486 o When off, write CRC to WAL and check file system blocks
1489 If CRC check fails during recovery, remember the page in case
1490 a later CRC for that page properly matches.
1492 o Write full pages during file system write and not when
1493 the page is modified in the buffer cache
1495 This allows most full page writes to happen in the background
1496 writer. It might cause problems for applying WAL on recovery
1497 into a partially-written page, but later the full page will be
1500 * Allow WAL traffic to be streamed to another server for stand-by
1502 * Reduce WAL traffic so only modified values are written rather than
1505 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1507 * Allow WAL information to recover corrupted pg_controldata
1509 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1511 * Find a way to reduce rotational delay when repeatedly writing
1514 Currently fsync of WAL requires the disk platter to perform a full
1515 rotation to fsync again. One idea is to write the WAL to different
1516 offsets that might reduce the rotational delay.
1518 * Allow WAL logging to be turned off for a table, but the table
1519 might be dropped or truncated during crash recovery [walcontrol]
1521 Allow tables to bypass WAL writes and just fsync() dirty pages on
1522 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1523 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1524 non-default logging should not use referential integrity with
1525 default-logging tables. A table without dirty buffers during a
1526 crash could perhaps avoid the drop/truncate.
1528 * Allow WAL logging to be turned off for a table, but the table would
1529 avoid being truncated/dropped [walcontrol]
1531 To do this, only a single writer can modify the table, and writes
1532 must happen only on new pages so the new pages can be removed during
1533 crash recovery. Readers can continue accessing the table. Such
1534 tables probably cannot have indexes. One complexity is the handling
1535 of indexes on TOAST tables.
1537 * Speed WAL recovery by allowing more than one page to be prefetched
1539 This should be done utilizing the same infrastructure used for
1540 prefetching in general to avoid introducing complex error-prone code
1542 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1543 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1544 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1546 * Improve WAL concurrency by increasing lock granularity
1548 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1552 Optimizer / Executor
1553 ====================
1555 * Improve selectivity functions for geometric operators
1556 * Precompile SQL functions to avoid overhead
1557 * Create utility to compute accurate random_page_cost value
1558 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1559 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1560 actual row counts differ by a specified percentage
1561 * Consider using hash buckets to do DISTINCT, rather than sorting
1563 This would be beneficial when there are few distinct values. This is
1564 already used by GROUP BY.
1566 * Log statements where the optimizer row estimates were dramatically
1567 different from the number of rows actually found?
1568 * Consider compressed annealing to search for query plans
1570 This might replace GEQO, http://sixdemonbag.org/Djinni.
1572 * Improve merge join performance by allowing mark/restore of
1575 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1577 * Consider using a hash for joining to a large IN (VALUES ...) list
1579 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1586 * Consider having the background writer update the transaction status
1587 hint bits before writing out the page
1589 Implementing this requires the background writer to have access to system
1590 catalogs and the transaction status log.
1592 * Consider adding buffers the background writer finds reusable to the
1595 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1597 * Automatically tune bgwriter_delay based on activity rather then using a
1600 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1602 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1604 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1606 * Test to see if calling PreallocXlogFiles() from the background writer
1607 will help with WAL segment creation latency
1609 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1613 Miscellaneous Performance
1614 =========================
1616 * Do async I/O for faster random read-ahead of data
1618 Async I/O allows multiple I/O requests to be sent to the disk with
1619 results coming back asynchronously.
1621 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1622 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1623 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1624 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1626 * Use mmap() rather than SYSV shared memory or to write WAL files?
1628 This would remove the requirement for SYSV SHM but would introduce
1629 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1630 to prevent I/O overhead.
1632 * Consider mmap()'ing files into a backend?
1634 Doing I/O to large tables would consume a lot of address space or
1635 require frequent mapping/unmapping. Extending the file also causes
1636 mapping problems that might require mapping only individual pages,
1637 leading to thousands of mappings. Another problem is that there is no
1638 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1639 could hit disk before WAL is written.
1641 * Add a script to ask system configuration questions and tune postgresql.conf
1642 * Consider ways of storing rows more compactly on disk
1644 o Reduce the row header size?
1645 o Consider reducing on-disk varlena length from four bytes to
1646 two because a heap row cannot be more than 64k in length
1648 * Consider increasing NUM_CLOG_BUFFERS
1650 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1651 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1653 * Allow user configuration of TOAST thresholds
1655 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1656 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1658 * Allow configuration of backend priorities via the operating system
1660 Though backend priorities make priority inversion during lock
1661 waits possible, research shows that this is not a huge problem.
1663 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1665 * Experiment with multi-threaded backend better I/O utilization
1667 This would allow a single query to make use of multiple I/O channels
1668 simultaneously. One idea is to create a background reader that can
1669 pre-fetch sequential and index scan pages needed by other backends.
1670 This could be expanded to allow concurrent reads from multiple devices
1671 in a partitioned table.
1673 * Experiment with multi-threaded backend better CPU utilization
1675 This would allow several CPUs to be used for a single query, such as
1676 for sorting or query execution.
1678 * Consider increasing the minimum allowed number of shared buffers
1680 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1682 * Expire published xmin for read-only and idle transactions
1684 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1686 * Consider if CommandCounterIncrement() can avoid its
1687 AcceptInvalidationMessages() call
1689 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1691 * Improve performance of shared invalidation queue for multiple CPUs
1693 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1695 * Consider Cartesian joins when both relations are needed to form an
1696 indexscan qualification for a third relation
1698 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1700 * Consider not storing a NULL bitmap on disk if all the NULLs are
1703 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1704 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1706 * Sort large UPDATE/DELETEs so it is done in heap order
1708 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1710 * Avoid tuple some tuple copying in sort routines
1712 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01206.php
1713 http://archives.postgresql.org/pgsql-patches/2008-02/msg00176.php
1715 * SMP scalability improvements
1717 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1718 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1719 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1721 * Research reducing deTOASTing in more places
1723 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1730 * Add use of 'const' for variables in source tree
1731 * Move some things from contrib into main tree
1732 * %Remove warnings created by -Wcast-align
1733 * Move platform-specific ps status display info from ps_status.c to ports
1734 * Add optional CRC checksum to heap and index pages
1735 * Improve documentation to build only interfaces (Marc)
1736 * Remove or relicense modules that are not under the BSD license, if possible
1737 * Acquire lock on a relation before building a relcache entry for it
1738 * Allow cross-compiling by generating the zic database on the target system
1739 * Improve NLS maintenance of libpgport messages linked onto applications
1740 * Clean up casting in contrib/isn
1742 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1744 * Use UTF8 encoding for NLS messages so all server encodings can
1746 * Update Bonjour to work with newer cross-platform SDK
1748 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1749 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1751 * Consider detoasting keys before sorting
1752 * Consider GnuTLS if OpenSSL license becomes a problem
1754 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1755 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1757 * Consider changing documentation format from SGML to XML
1759 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1761 * Consider making NAMEDATALEN more configurable in future releases
1762 * Update our code to handle 64-bit timezone files to match the zic
1763 source code, which now uses them
1764 * Have configure choose integer datetimes by default
1766 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1768 * Support scoped IPv6 addresses
1770 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1772 * Consider allowing 64-bit integers and floats to be passed by value on
1775 Also change 32-bit floats (float4) to be passed by value at the same
1778 * Research use of signals and sleep wake ups
1780 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1782 * Add automated check for invalid C++ source code constructs
1784 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1786 * Consider simplifying how memory context resets handle child contexts
1788 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1790 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1792 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1794 * Convert single quotes to apostrophes in the PDF documentation
1796 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
1802 o Remove configure.in check for link failure when cause is found
1803 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1805 o Remove psql newline patch when we find out why mingw outputs an
1807 o Allow psql to use readline once non-US code pages work with
1809 o Fix problem with shared memory on the Win32 Terminal Server
1810 o Diagnose problem where shared memory can sometimes not be
1811 attached by postmaster children
1813 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
1815 o Improve signal handling
1817 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1819 o Convert MSVC build system to remove most batch files
1821 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
1823 o Prevent SSL from sending network packets to avoid interference
1824 with Win32 signal emulation
1826 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
1828 o Support pgxs when using MSVC
1830 o Fix MSVC NLS support, like for to_char()
1832 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
1833 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
1837 * Wire Protocol Changes
1839 o Allow dynamic character set handling
1840 o Add decoded type, length, precision
1842 o Update clients to use data types, typmod, schema.table.column names
1843 of result sets using new statement protocol
1849 * Add pre-parsing phase that converts non-ISO syntax to supported
1852 This could allow SQL written for other databases to run without
1855 * Allow plug-in modules to emulate features from other databases
1856 * Add features of Oracle-style packages (Pavel)
1858 A package would be a schema with session-local variables,
1859 public/private functions, and initialization functions. It
1860 is also possible to implement these capabilities
1861 in any schema and not use a separate "packages"
1864 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1866 * Consider allowing control of upper/lower case folding of unquoted
1869 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1870 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1872 * Add autonomous transactions
1874 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1878 Features We Do _Not_ Want
1879 =========================
1881 * All backends running as threads in a single process (not wanted)
1883 This eliminates the process protection we get from the current setup.
1884 Thread creation is usually the same overhead as process creation on
1885 modern systems, so it seems unwise to use a pure threaded model.
1887 * Optimizer hints (not wanted)
1889 Optimizer hints are used to work around problems in the optimizer. We
1890 would rather have the problems reported and fixed.
1892 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1893 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1894 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1897 Because we support postfix operators, it isn't possible to make AS
1898 optional and continue to use bison.
1899 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1901 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1903 * Embedded server (not wanted)
1905 While PostgreSQL clients runs fine in limited-resource environments, the
1906 server requires multiple processes and a stable pool of resources to
1907 run reliabily and efficiently. Stripping down the PostgreSQL server
1908 to run in the same process address space as the client application
1909 would add too much complexity and failure cases.