3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Wed Apr 16 13:25:50 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
266 * Allow xml arrays to be cast to other data types
268 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00981.php
269 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00231.php
270 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00471.php
272 * Simplify integer cross-data-type operators
274 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00189.php
276 * Allow XML to accept more liberal DOCTYPE specifications
278 http://archives.postgresql.org/pgsql-general/2008-02/msg00347.php
283 o Allow infinite dates and intervals just like infinite timestamps
284 o Merge hardwired timezone names with the TZ database; allow either
285 kind everywhere a TZ name is currently taken
286 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
287 information, either zone name or offset from UTC [timezone]
289 If the TIMESTAMP value is stored with a time zone name, interval
290 computations should adjust based on the time zone rules.
292 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
293 o Add a GUC variable to allow output of interval values in ISO8601
295 o Have timestamp subtraction not call justify_hours()?
297 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
299 o Improve timestamptz subtraction to be DST-aware
301 Currently subtracting one date from another that crosses a
302 daylight savings time adjustment can return '1 day 1 hour', but
303 adding that back to the first date returns a time one hour in
304 the future. This is caused by the adjustment of '25 hours' to
305 '1 day 1 hour', and '1 day' is the same time the next day, even
306 if daylight savings adjustments are involved.
308 o Fix interval display to support values exceeding 2^31 hours
309 o Add overflow checking to timestamp and interval arithmetic
310 o Extend timezone code to allow 64-bit values so we can
311 represent years beyond 2038
313 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
315 o Use LC_TIME for localized weekday/month names, rather than
318 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
320 o Add ISO INTERVAL handling
322 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
323 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
325 o Support ISO INTERVAL syntax if units cannot be determined from
326 the string, and are supplied after the string
328 The SQL standard states that the units after the string
329 specify the units of the string, e.g. INTERVAL '2' MINUTE
330 should return '00:02:00'. The current behavior has the units
331 restrict the interval value to the specified unit or unit
332 range, INTERVAL '70' SECOND returns '00:00:10'.
334 For syntax that isn't uniquely ISO or PG syntax, like '1' or
335 '1:30', treat as ISO if there is a range specification clause,
336 and as PG if there no clause is present, e.g. interpret '1:30'
337 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
338 '1:30' as '1 hour, 30 minutes'.
340 This makes common cases like SELECT INTERVAL '1' MONTH
341 SQL-standard results. The SQL standard supports a limited
342 number of unit combinations and doesn't support unit names in
343 the string. The PostgreSQL syntax is more flexible in the
344 range of units supported, e.g. PostgreSQL supports '1 year 1
345 hour', while the SQL standard does not.
347 o Add support for year-month syntax, INTERVAL '50-6' YEAR
349 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
350 year' AS INTERVAL MONTH), and this should return '12 months'
351 o Round or truncate values to the requested precision, e.g.
352 INTERVAL '11 months' AS YEAR should return one or zero
353 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
358 o Delay resolution of array expression's data type so assignment
359 coercion can be performed on empty array expressions
360 o Add support for arrays of domains
362 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
364 o Allow single-byte header storage for array elements
369 o Improve vacuum of large objects, like contrib/vacuumlo?
370 o Add security checking for large objects
371 o Auto-delete large objects when referencing row is deleted
373 contrib/lo offers this functionality.
375 o Allow read/write into TOAST values like large objects
377 This requires the TOAST column to be stored EXTERNAL.
379 o Add API for 64-bit large object access
381 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
385 * Add locale-aware MONEY type, and support multiple currencies
387 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
388 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
390 * MONEY dumps in a locale-specific format making it difficult to
391 restore to a system with a different locale
392 * Allow MONEY to be easily cast to/from other numeric data types
398 * Allow INET subnet tests using non-constants to be indexed
399 * Allow to_date() and to_timestamp() accept localized month names
400 * Fix to_date()-related functions to consistently issue errors
402 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
404 * Add missing parameter handling in to_char()
406 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
408 * Allow substring/replace() to get/set bit values
409 * Allow to_char() on interval values to accumulate the highest unit
412 Some special format flag would be required to request such
413 accumulation. Such functionality could also be added to EXTRACT.
414 Prevent accumulation that crosses the month/day boundary because of
415 the uneven number of days in a month.
417 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
418 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
419 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
420 o to_char(INTERVAL '3 years 5 months','MM') => 41
422 * Implement inlining of set-returning functions defined in SQL
423 * Allow SQL-language functions to return results from RETURNING queries
425 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
427 * Allow SQL-language functions to reference parameters by parameter name
429 Currently SQL-language functions can only refer to dollar parameters,
432 * Add SPI_gettypmod() to return the typemod for a TupleDesc
433 * Enforce typmod for function inputs, function results and parameters for
434 spi_prepare'd statements called from PLs
436 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
438 * Allow holdable cursors in SPI
439 * Tighten function permission checks
441 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
443 * Fix IS OF so it matches the ISO specification, and add documentation
445 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
446 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
448 * Add missing operators for geometric data types
450 Some geometric types do not have the full suite of geometric operators,
453 * Implement Boyer-Moore searching in strpos()
455 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
457 * Prevent malicious functions from being executed with the permissions
458 of unsuspecting users
460 Index functions are safe, so VACUUM and ANALYZE are safe too.
461 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
462 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
464 * Reduce memory usage of aggregates in set returning functions
466 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
468 * Add temporal versions of generate_series()
470 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
472 * Add array_accum() and array_to_set() functions for arrays
474 The standards specify array_agg() and UNNEST.
475 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
477 * Fix /contrib/ltree operator
479 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
481 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
483 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
485 * Fix regular expression bug when using complex back-references
487 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00000.php
489 * Have /contrib/dblink reuse unnamed connections
491 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00895.php
493 * Add SQL-standard array_agg() and unnest() array functions
495 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01017.php
497 * Allow calling of a procedure outside a SELECT that can control the
500 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01375.php
504 Multi-Language Support
505 ======================
507 * Add NCHAR (as distinguished from ordinary varchar),
508 * Allow locale to be set at database creation
510 Currently locale can only be set during initdb. No global tables have
511 locale-aware columns. However, the database template used during
512 database creation might have locale-aware indexes. The indexes would
513 need to be reindexed to match the new locale.
515 * Allow encoding on a per-column basis optionally using the ICU library
517 Right now only one encoding is allowed per database. [locale]
519 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
520 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
521 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
522 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
523 http://wiki.postgresql.org/wiki/Todo:Collate
524 http://wiki.postgresql.org/wiki/Todo:ICU
526 * Add CREATE COLLATE? [locale]
527 * Support multiple simultaneous character sets, per SQL92
528 * Improve UTF8 combined character handling?
529 * Add octet_length_server() and octet_length_client()
530 * Make octet_length_client() the same as octet_length()?
531 * Fix problems with wrong runtime encoding conversion for NLS message files
532 * Add URL to more complete multi-byte regression tests
534 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
536 * Fix ILIKE and regular expressions to handle case insensitivity
537 properly in multibyte encodings
539 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
540 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
542 * Set client encoding based on the client operating system encoding
544 Currently client_encoding is set in postgresql.conf, which
545 defaults to the server encoding.
546 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
548 * Change memory allocation for multi-byte functions so memory is
549 allocated inside conversion functions
551 Currently we preallocate memory based on worst-case usage.
558 * Automatically create rules on views so they are updateable, per SQL99
560 We can only auto-create rules for simple views. For more complex
561 cases users will still have to write rules manually.
563 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
564 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
566 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
567 * Allow VIEW/RULE recompilation when the underlying tables change
569 Another issue is whether underlying table changes should be reflected
570 in the view, e.g. should SELECT * show additional columns if they
571 are added after the view is created.
573 * Make it possible to use RETURNING together with conditional DO INSTEAD
574 rules, such as for partitioning setups
576 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
578 * Add the ability to automatically create materialized views
580 Right now materialized views require the user to create triggers on the
581 main table to keep the summary table current. SQL syntax should be able
582 to manager the triggers and summary table automatically. A more
583 sophisticated implementation would automatically retrieve from the
584 summary table when the main table is referenced, if possible.
591 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
592 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
593 * %Allow SET CONSTRAINTS to be qualified by schema/table name
594 * %Add a separate TRUNCATE permission
596 Currently only the owner can TRUNCATE a table because triggers are not
597 called, and the table is locked in exclusive mode.
599 * Allow PREPARE of cursors
600 * Allow finer control over the caching of prepared query plans
602 Currently queries prepared via the libpq API are planned on first
603 execute using the supplied parameters --- allow SQL PREPARE to do the
604 same. Also, allow control over replanning prepared queries either
605 manually or automatically when statistics for execute parameters
606 differ dramatically from those used during planning.
608 * Improve logging of prepared transactions recovered during startup
610 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
612 * Add a GUC variable to warn about non-standard SQL usage in queries
613 * Add SQL-standard MERGE command, typically used to merge two tables
616 This is similar to UPDATE, then for unmatched rows, INSERT.
617 Whether concurrent access allows modifications which could cause
618 row loss is implementation independent.
620 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
623 To implement this cleanly requires that the table have a unique index
624 so duplicate checking can be easily performed. It is possible to
625 do it without a unique index if we require the user to LOCK the table
628 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
629 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
631 * Add NOVICE output level for helpful messages like automatic sequence/index
633 * Add GUC to issue notice about statements that use unjoined tables
634 * Allow EXPLAIN to identify tables that were skipped because of
636 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
637 * Enable standard_conforming_strings
638 * Make standard_conforming_strings the default in 8.5?
640 When this is done, backslash-quote should be prohibited in non-E''
641 strings because of possible confusion over how such strings treat
642 backslashes. Basically, '' is always safe for a literal single
643 quote, while \' might or might not be based on the backslash
646 * Simplify dropping roles that have objects in several databases
647 * Allow COMMENT ON to accept an expression rather than just a string
648 * Allow the count returned by SELECT, etc to be represented as an int64
649 to allow a higher range of values
650 * Add SQL99 WITH clause to SELECT
651 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
653 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
654 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
655 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
656 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
657 http://archives.postgresql.org/pgsql-patches/2008-01/msg00105.php
658 http://archives.postgresql.org/pgsql-patches/2008-03/msg00327.php
660 * Add DEFAULT .. AS OWNER so permission checks are done as the table
663 This would be useful for SERIAL nextval() calls and CHECK constraints.
665 * Allow DISTINCT to work in multiple-argument aggregate calls
666 * Add column to pg_stat_activity that shows the progress of long-running
667 commands like CREATE INDEX and VACUUM
668 * Implement SQL:2003 window functions
669 * Improve failure message when DROP DATABASE is used on a database that
670 has prepared transactions
671 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
673 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
674 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
676 * Increase locking when DROPing objects so dependent objects cannot
677 get dropped while the DROP operation is happening
679 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
681 * -Allow AS in "SELECT col AS label" to be optional in certain cases
683 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
684 RETURNING to supply values to the INSERT
685 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
687 * Add comments on system tables/columns using the information in
690 Ideally the information would be pulled from the SGML file
693 * Improve reporting of UNION type mismatches
695 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
696 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
701 o Allow CREATE TABLE AS to determine column lengths for complex
702 expressions like SELECT col1 || col2
704 o Have WITH CONSTRAINTS also create constraint indexes
706 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
708 o Have CONSTRAINT cname NOT NULL record the contraint name
710 Right now pg_attribute.attnotnull records the NOT NULL status
711 of the column, but does not record the contraint name
713 o Prevent concurrent CREATE TABLE table1 from sometimes returning
714 a cryptic error message
716 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00169.php
720 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
722 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
723 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
724 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
725 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
727 o Research self-referential UPDATEs that see inconsistent row versions
728 in read-committed mode
730 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
731 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
733 o Allow GLOBAL temporary tables to exist as empty by default in
736 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
741 o Have ALTER TABLE RENAME rename SERIAL sequence names
743 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
745 o Have ALTER SEQUENCE RENAME rename the sequence name stored
746 in the sequence table
748 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
749 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
750 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
752 o Add ALTER DOMAIN to modify the underlying data type
753 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
755 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
757 o %Allow ALTER TABLE to change constraint deferrability and actions
758 o Add missing object types for ALTER ... SET SCHEMA
759 o Allow ALTER TABLESPACE to move to different directories
760 o Allow databases to be moved to different tablespaces
761 o Allow moving system tables to other tablespaces, where possible
763 Currently non-global system tables must be in the default database
764 tablespace. Global system tables can never be moved.
766 o Prevent parent tables from altering or dropping constraints
767 like CHECK that are inherited by child tables unless CASCADE
769 o %Prevent child tables from altering or dropping constraints
770 like CHECK that were inherited from the parent table
771 o Have ALTER INDEX update the name of a constraint using that index
772 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
773 o Allow column display reordering by recording a display,
774 storage, and permanent id for every column?
776 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
782 o Automatically maintain clustering on a table
784 This might require some background daemon to maintain clustering
785 during periods of low usage. It might also require tables to be only
786 partially filled for easier reorganization. Another idea would
787 be to create a merged heap/index data file so an index lookup would
788 automatically access the heap data too. A third idea would be to
789 store heap rows in hashed groups, perhaps using a user-supplied
791 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
793 o %Add default clustering to system tables
795 To do this, determine the ideal cluster index for each system
796 table and set the cluster setting during initdb.
798 o %Add VERBOSE option to report tables as they are processed,
804 o Allow COPY to report error lines and continue
806 This requires the use of a savepoint before each COPY line is
807 processed, with ROLLBACK on COPY failure.
808 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
810 o Allow COPY on a newly-created table to skip WAL logging
812 On crash recovery, the table involved in the COPY would
813 be removed or have its heap and index files truncated. One
814 issue is that no other backend should be able to add to
815 the table at the same time, which is something that is
816 currently allowed. This currently is done if the table is
817 created inside the same transaction block as the COPY because
818 no other backends can see the table.
820 o Consider using a ring buffer for COPY FROM
822 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
823 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
825 o Allow COPY FROM to create index entries in bulk
827 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
829 o Allow COPY in CSV mode to control whether a quoted zero-length
830 string is treated as NULL
832 Currently this is always treated as a zero-length string,
833 which generates an error when loading into an integer column
834 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
840 o Allow column-level privileges
841 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
844 The proposed syntax is:
845 GRANT SELECT ON ALL TABLES IN public TO phpuser;
846 GRANT SELECT ON NEW TABLES IN public TO phpuser;
848 o Allow GRANT/REVOKE permissions to be inherited by objects based on
851 o Allow SERIAL sequences to inherit permissions from the base table?
856 o Prevent DROP TABLE from dropping a row referenced by its own open
862 o Allow INSERT/UPDATE of the system-generated oid value for a row
863 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
869 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
875 o Allow LISTEN/NOTIFY to store info in memory rather than tables?
877 Currently LISTEN/NOTIFY information is stored in pg_listener.
878 Storing such information in memory would improve performance.
880 o Add optional textual message to NOTIFY
882 This would allow an informational message to be added to the notify
883 message, perhaps indicating the row modified or other custom
886 o Allow multiple identical NOTIFY events to always be communicated
887 to the client, rather than sent as a single notification to the
890 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
892 o Allow NOTIFY in rules involving conditionals
893 o Improve LISTEN concurrency
895 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01106.php
899 Referential Integrity
900 =====================
902 * Add MATCH PARTIAL referential integrity
903 * Change foreign key constraint for array -> element to mean element
905 * Fix problem when cascading referential triggers make changes on
906 cascaded tables, seeing the tables in an intermediate state
908 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
909 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
911 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
913 This would allow UPDATE tab SET col = col + 1 to work if col has
914 a unique index. Currently, uniqueness checks are done while the
915 command is being executed, rather than at the end of the statement
917 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
918 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
920 * Optimize referential integrity checks
922 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
923 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
926 Server-Side Languages
927 =====================
930 o Fix RENAME to work on variables other than OLD/NEW
932 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
933 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
934 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
936 o Allow function parameters to be passed by name,
937 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
938 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
939 o Allow listing of record column names, and access to
940 record columns via variables, e.g. columns := r.(*),
943 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
944 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
945 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
947 o Add support for SCROLL cursors
948 o Add support for WITH HOLD cursors
949 o Allow row and record variables to be set to NULL constants,
950 and allow NULL tests on such variables
952 Because a row is not scalar, do not allow assignment
953 from NULL-valued scalars.
955 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
957 o Review handling of MOVE and FETCH
959 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
961 o Improve logic of determining if an identifier is a a
962 variable or column name
964 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
966 o Consider keeping seperate cached copies when search_path changes
968 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01009.php
970 o Add CASE capability to language (already in SQL)
972 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00696.php
978 o Add table function support to pltcl, plpythonu
979 o Add support for polymorphic arguments and return types to
980 languages other than PL/PgSQL
981 o Add capability to create and call PROCEDURES
982 o Add support for OUT and INOUT parameters to languages other
984 o Add PL/PythonU tracebacks
986 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
988 o Allow data to be passed in native language formats, rather
991 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
993 o Add ability to obfuscate function bodies
995 http://archives.postgresql.org/pgsql-patches/2008-01/msg00125.php
1002 * Have pg_ctl look at PGHOST in case it is a socket directory?
1003 * Allow pg_ctl to work properly with configuration files located outside
1004 the PGDATA directory
1006 pg_ctl can not read the pid file because it isn't located in the
1007 config directory but in the PGDATA directory. The solution is to
1008 allow pg_ctl to read and understand postgresql.conf to find the
1009 data_directory value.
1011 * Add a function like pg_get_indexdef() that report more detailed index
1014 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
1017 o Have psql show current values for a sequence
1018 o Move psql backslash database information into the backend, use
1019 mnemonic commands? [psql]
1021 This would allow non-psql clients to pull the same information out
1022 of the database as psql.
1024 o Fix psql's \d commands more consistent
1026 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1027 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1029 o Consistently display privilege information for all objects in psql
1030 o Add auto-expanded mode so expanded output is used if the row
1031 length is wider than the screen width.
1033 Consider using auto-expanded mode for backslash commands like \df+.
1035 o Prevent tab completion of SET TRANSACTION from querying the
1036 database and therefore preventing the transaction isolation
1037 level from being set.
1039 Currently SET <tab> causes a database lookup to check all
1040 supported session variables. This query causes problems
1041 because setting the transaction isolation level must be the
1042 first statement of a transaction.
1044 o Add a \set variable to control whether \s displays line numbers
1046 Another option is to add \# which lists line numbers, and
1047 allows command execution.
1049 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
1051 o Prevent escape string warnings when object names have
1054 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1056 o Have \d show foreign keys that reference a table's primary key
1058 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1060 o Have \d show child tables that inherit from the specified parent
1061 o Have \l+ show database size, if permissions allow
1063 Ideally it will not generate an error for invalid permissions
1065 o Include the symbolic SQLSTATE name in verbose error reports
1067 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1069 o Improve display if enums
1071 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00826.php
1076 * pg_dump / pg_restore
1077 o %Add dumping of comments on index columns and composite type columns
1078 o %Add full object name to the tag field. eg. for operators we need
1079 '=(integer, integer)', instead of just '='.
1080 o Add pg_dumpall custom format dumps?
1081 o Allow selection of individual object(s) of all types, not just
1083 o In a selective dump, allow dumping of an object and all its
1085 o Add options like pg_restore -l and -L to pg_dump
1086 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1087 o Allow pg_dump --clean to drop roles that own objects or have
1089 o Change pg_dump so that a comment on the dumped database is
1090 applied to the loaded database, even if the database has a
1091 different name. This will require new backend syntax, perhaps
1092 COMMENT ON CURRENT DATABASE.
1093 o Remove unnecessary function pointer abstractions in pg_dump source
1095 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1096 multiple objects simultaneously
1098 The difficulty with this is getting multiple dump processes to
1099 produce a single dump output file. It also would require
1100 several sessions to share the same snapshot.
1101 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1103 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1104 restoring multiple objects simultaneously
1106 This might require a pg_restore flag to indicate how many
1107 simultaneous operations should be performed. Only pg_dump's
1108 -Fc format has the necessary dependency information.
1109 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1111 o To better utilize resources, allow pg_restore to check foreign
1112 keys simultaneously, where possible
1113 o Allow pg_restore to create all indexes of a table
1114 concurrently, via a single heap scan
1116 This requires a pg_dump -Fc file because that format contains
1117 the required dependency information.
1118 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1120 o Allow pg_restore to load different parts of the COPY data
1122 o Prevent pg_dump/pg_restore from being affected by
1125 Using psql to restore a pg_dump dump is also affected.
1127 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1128 o Allow pre/data/post files when schema and data are dumped
1129 separately, for performance reasons
1131 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1138 Document differences between ecpg and the SQL standard and
1139 information about the Informix-compatibility module.
1141 o Solve cardinality > 1 for input descriptors / variables?
1142 o Add a semantic check level, e.g. check if a table really exists
1143 o fix handling of DB attributes that are arrays
1144 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1146 o Fix nested C comments
1147 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1148 o Make SET CONNECTION thread-aware, non-standard?
1149 o Allow multidimensional arrays
1150 o Add internationalized message strings
1151 o Implement COPY FROM STDIN
1155 o Add PQescapeIdentifierConn()
1156 o Prevent PQfnumber() from lowercasing unquoted the column name
1158 PQfnumber() should never have been doing lowercasing, but
1159 historically it has so we need a way to prevent it
1161 o Allow statement results to be automatically batched to the client
1163 Currently all statement results are transferred to the libpq
1164 client before libpq makes the results available to the
1165 application. This feature would allow the application to make
1166 use of the first result rows while the rest are transferred, or
1167 held on the server waiting for them to be requested by libpq.
1168 One complexity is that a statement like SELECT 1/col could error
1169 out mid-way through the result set.
1171 o Consider disallowing multiple queries in PQexec() as an
1172 additional barrier to SQL injection attacks
1174 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1176 o Add PQexecf() that allows complex parameter substitution
1178 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1180 o Add SQLSTATE severity to PGconn return status
1182 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1188 * Add deferred trigger queue file
1190 Right now all deferred trigger information is stored in backend
1191 memory. This could exhaust memory for very large trigger queues.
1192 This item involves dumping large queues into files.
1194 * Allow triggers to be disabled in only the current session.
1196 This is currently possible by starting a multi-statement transaction,
1197 modifying the system tables, performing the desired SQL, restoring the
1198 system tables, and committing the transaction. ALTER TABLE ...
1199 TRIGGER requires a table lock so it is not ideal for this usage.
1201 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1203 If the dump is known to be valid, allow foreign keys to be added
1204 without revalidating the data.
1206 * Allow statement-level triggers to access modified rows
1207 * Support triggers on columns
1209 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1211 * Allow AFTER triggers on system tables
1213 System tables are modified in many places in the backend without going
1214 through the executor and therefore not causing triggers to fire. To
1215 complete this item, the functions that modify system tables will have
1218 * Tighten trigger permission checks
1220 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1222 * Allow BEFORE INSERT triggers on views
1224 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1226 * -Add ability to trigger on TRUNCATE
1233 * Add UNIQUE capability to non-btree indexes
1234 * Prevent index uniqueness checks when UPDATE does not modify the column
1236 Uniqueness (index) checks are done when updating a column even if the
1237 column is not modified by the UPDATE.
1239 * Allow the creation of on-disk bitmap indexes which can be quickly
1240 combined with other bitmap indexes
1242 Such indexes could be more compact if there are only a few distinct values.
1243 Such indexes can also be compressed. Keeping such indexes updated can be
1246 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1247 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1248 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00265.php
1249 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01214.php
1250 http://archives.postgresql.org/pgsql-patches/2007-05/msg00013.php
1251 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00741.php
1253 * Allow accurate statistics to be collected on indexes with more than
1254 one column or expression indexes, perhaps using per-index statistics
1256 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1257 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1259 * Consider increasing the number of default statistics target, and
1260 reduce statistics target overhead
1262 Also consider having a larger statistics target for indexed columns
1263 and expression indexes
1264 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1265 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1266 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1268 * Consider compressing indexes by storing key values duplicated in
1269 several rows as a single index entry
1271 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1272 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1273 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1274 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1275 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00014.php
1276 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00487.php
1278 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1280 This is difficult because you must upgrade to an exclusive table lock
1281 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1282 have this complication. This would allow index compaction without
1284 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1286 * Allow multiple indexes to be created concurrently, ideally via a
1287 single heap scan, and have pg_restore use it
1289 * Consider sorting entries before inserting into btree index
1291 http://archives.postgresql.org/pgsql-general/2008-01/msg01010.php
1293 * Allow index scans to return matching index keys
1295 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01079.php
1300 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1301 and primary/foreign keys
1302 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1303 on inherited table, e.g. INSERT INTO inherit_table
1304 (unique_index_col) VALUES (dup) should fail
1306 The main difficulty with this item is the problem of
1307 creating an index that can span multiple tables.
1309 o Allow SELECT ... FOR UPDATE on inherited tables
1310 o Require all CHECK constraints to be inherited
1312 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1314 o Add checks to prevent a CREATE RULE views on inherited tables
1316 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1317 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1322 o Add more GIST index support for geometric data types
1323 o Allow GIST indexes to create certain complex index types, like
1324 digital trees (see Aoki)
1329 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1331 o Pack hash index buckets onto disk pages more efficiently
1333 Currently only one hash bucket can be stored on a page. Ideally
1334 several hash buckets could be stored on a single page and greater
1335 granularity used for the hash algorithm.
1337 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1339 o Consider sorting hash buckets so entries can be found using a
1340 binary search, rather than a linear scan
1341 o In hash indexes, consider storing the hash value with or instead
1343 o Add WAL logging for crash recovery
1344 o Allow multi-column hash indexes
1345 o -During index creation, pre-sort the tuples to improve build speed
1352 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1354 Ideally this requires a separate test program that can be run
1355 at initdb time or optionally later. Consider O_SYNC when
1358 * Add program to test if fsync has a delay compared to non-fsync
1359 * Consider sorting writes during checkpoint
1361 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1370 We could use a fixed row count and a +/- count to follow MVCC
1371 visibility rules, or a single cached value could be used and
1372 invalidated if anyone modifies the table. Another idea is to
1373 get a count directly from a unique index, but for this to be
1374 faster than a sequential scan it must avoid access to the heap
1375 to obtain tuple visibility information.
1377 * Provide a way to calculate an "estimated COUNT(*)"
1379 Perhaps by using the optimizer's cardinality estimates or random
1382 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1384 * Allow data to be pulled directly from indexes
1386 Currently indexes do not have enough tuple visibility information
1387 to allow data to be pulled from the index without also accessing
1388 the heap. One way to allow this is to set a bit on index tuples
1389 to indicate if a tuple is currently visible to all transactions
1390 when the first valid heap lookup happens. This bit would have to
1391 be cleared when a heap tuple is expired.
1393 Another idea is to maintain a bitmap of heap pages where all rows
1394 are visible to all backends, and allow index lookups to reference
1395 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1396 add someday to determine which heap pages need vacuuming. Frequently
1397 accessed bitmaps would have to be stored in shared memory. One 8k
1398 page of bitmaps could track 512MB of heap pages.
1400 A third idea would be for a heap scan to check if all rows are visible
1401 and if so set a per-table flag which can be checked by index scans.
1402 Any change to the table would have to clear the flag. To detect
1403 changes during the heap scan a counter could be set at the start and
1404 checked at the end --- if it is the same, the table has not been
1405 modified --- any table change would increment the counter.
1407 http://archives.postgresql.org/pgsql-patches/2007-10/msg00166.php
1408 http://archives.postgresql.org/pgsql-patches/2008-01/msg00049.php
1410 * Consider automatic caching of statements at various levels:
1413 o Query execute plan
1415 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00823.php
1417 * Consider increasing internal areas when shared buffers is increased
1419 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1421 * Consider decreasing the amount of memory used by PrivateRefCount
1423 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1424 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1426 * Consider allowing higher priority queries to have referenced buffer
1427 cache pages stay in memory longer
1429 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1436 * Improve speed with indexes
1438 For large table adjustments during VACUUM FULL, it is faster to cluster
1439 or reindex rather than update the index. Also, index updates can bloat
1442 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1443 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1444 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1446 * Auto-fill the free space map by scanning the buffer cache or by
1447 checking pages written by the background writer
1449 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1450 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1452 * Create a bitmap of pages that need vacuuming
1454 Instead of sequentially scanning the entire table, have the background
1455 writer or some other process record pages that have expired rows, then
1456 VACUUM can look at just those pages rather than the entire table. In
1457 the event of a system crash, the bitmap would probably be invalidated.
1458 One complexity is that index entries still have to be vacuumed, and
1459 doing this without an index scan (by using the heap values to find the
1460 index entry) might be slow and unreliable, especially for user-defined
1463 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1464 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1465 http://archives.postgresql.org/pgsql-patches/2007-03/msg00508.php
1466 http://archives.postgresql.org/pgsql-patches/2007-04/msg00347.php
1467 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00156.php
1468 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00546.php
1469 http://archives.postgresql.org/pgsql-hackers/2008-04/msg00416.php
1471 * Allow FSM to return free space toward the beginning of the heap file,
1472 in hopes that empty pages at the end can be truncated by VACUUM
1473 * Allow FSM page return free space based on table clustering, to assist
1474 in maintaining clustering?
1475 * Improve dead row detection during multi-statement transactions usage
1477 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1479 * Consider a more compact data representation for dead tuples
1481 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1486 o %Issue log message to suggest VACUUM FULL if a table is nearly
1488 o Improve control of auto-vacuum
1490 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1492 o Prevent long-lived temporary tables from causing frozen-xid
1493 advancement starvation
1495 The problem is that autovacuum cannot vacuum them to set frozen xids;
1496 only the session that created them can do that.
1497 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1499 o Store per-table autovacuum settings in pg_class.reloptions.
1501 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1502 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1504 o Prevent autovacuum from running if an old transaction is still
1505 running from the last vacuum
1507 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1514 * Fix priority ordering of read and write light-weight locks (Neil)
1516 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1517 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1519 * Fix problem when multiple subtransactions of the same outer transaction
1520 hold different types of locks, and one subtransaction aborts
1522 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1523 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1524 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1525 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1527 * Allow UPDATEs on only non-referential integrity columns not to conflict
1528 with referential integrity locks
1530 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1532 * Add idle_in_transaction_timeout GUC so locks are not held for long
1535 * Improve deadlock detection when a page cleaning lock conflicts
1536 with a shared buffer that is pinned
1538 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1539 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1540 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1542 * Detect deadlocks involving LockBufferForCleanup()
1544 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1548 Startup Time Improvements
1549 =========================
1551 * Experiment with multi-threaded backend for backend creation [thread]
1553 This would prevent the overhead associated with process creation. Most
1554 operating systems have trivial process creation time compared to
1555 database startup overhead, but a few operating systems (Win32,
1556 Solaris) might benefit from threading. Also explore the idea of
1557 a single session using multiple threads to execute a statement faster.
1564 * Eliminate need to write full pages to WAL before page modification [wal]
1566 Currently, to protect against partial disk page writes, we write
1567 full page images to WAL before they are modified so we can correct any
1568 partial page writes during recovery. These pages can also be
1569 eliminated from point-in-time archive files.
1571 o When off, write CRC to WAL and check file system blocks
1574 If CRC check fails during recovery, remember the page in case
1575 a later CRC for that page properly matches.
1577 o Write full pages during file system write and not when
1578 the page is modified in the buffer cache
1580 This allows most full page writes to happen in the background
1581 writer. It might cause problems for applying WAL on recovery
1582 into a partially-written page, but later the full page will be
1585 * Allow WAL traffic to be streamed to another server for stand-by
1587 * Reduce WAL traffic so only modified values are written rather than
1590 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1592 * Allow WAL information to recover corrupted pg_controldata
1594 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1596 * Find a way to reduce rotational delay when repeatedly writing
1599 Currently fsync of WAL requires the disk platter to perform a full
1600 rotation to fsync again. One idea is to write the WAL to different
1601 offsets that might reduce the rotational delay.
1603 * Allow WAL logging to be turned off for a table, but the table
1604 might be dropped or truncated during crash recovery [walcontrol]
1606 Allow tables to bypass WAL writes and just fsync() dirty pages on
1607 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1608 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1609 non-default logging should not use referential integrity with
1610 default-logging tables. A table without dirty buffers during a
1611 crash could perhaps avoid the drop/truncate.
1613 * Allow WAL logging to be turned off for a table, but the table would
1614 avoid being truncated/dropped [walcontrol]
1616 To do this, only a single writer can modify the table, and writes
1617 must happen only on new pages so the new pages can be removed during
1618 crash recovery. Readers can continue accessing the table. Such
1619 tables probably cannot have indexes. One complexity is the handling
1620 of indexes on TOAST tables.
1622 * Speed WAL recovery by allowing more than one page to be prefetched
1624 This should be done utilizing the same infrastructure used for
1625 prefetching in general to avoid introducing complex error-prone code
1627 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1628 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1629 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1631 * Improve WAL concurrency by increasing lock granularity
1633 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1635 * Be more aggressive about creating WAL files
1637 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01325.php
1638 http://archives.postgresql.org/pgsql-hackers/2004-07/msg01075.php
1639 http://archives.postgresql.org/pgsql-hackers/2005-04/msg00556.php
1641 * Have resource managers report the duration of their status changes
1643 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01468.php
1645 * Move pgfoundry's xlogdump to /contrib and have it rely more closely
1646 on the WAL backend code
1648 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00035.php
1653 Optimizer / Executor
1654 ====================
1656 * Improve selectivity functions for geometric operators
1657 * Precompile SQL functions to avoid overhead
1658 * Create utility to compute accurate random_page_cost value
1659 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1660 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1661 actual row counts differ by a specified percentage
1662 * Improve how ANALYZE computes in-doubt tuples
1664 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00771.php
1666 * Consider using hash buckets to do DISTINCT, rather than sorting
1668 This would be beneficial when there are few distinct values. This is
1669 already used by GROUP BY.
1671 * Log statements where the optimizer row estimates were dramatically
1672 different from the number of rows actually found?
1673 * Consider compressed annealing to search for query plans
1675 This might replace GEQO, http://sixdemonbag.org/Djinni.
1677 * Improve merge join performance by allowing mark/restore of
1680 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1682 * Consider using a hash for joining to a large IN (VALUES ...) list
1684 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1691 * Consider having the background writer update the transaction status
1692 hint bits before writing out the page
1694 Implementing this requires the background writer to have access to system
1695 catalogs and the transaction status log.
1697 * Consider adding buffers the background writer finds reusable to the
1700 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1702 * Automatically tune bgwriter_delay based on activity rather then using a
1705 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1707 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1709 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1711 * Test to see if calling PreallocXlogFiles() from the background writer
1712 will help with WAL segment creation latency
1714 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1718 Miscellaneous Performance
1719 =========================
1721 * Do async I/O for faster random read-ahead of data
1723 Async I/O allows multiple I/O requests to be sent to the disk with
1724 results coming back asynchronously.
1726 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1727 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1728 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1729 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1731 * Use mmap() rather than SYSV shared memory or to write WAL files?
1733 This would remove the requirement for SYSV SHM but would introduce
1734 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1735 to prevent I/O overhead.
1737 * Consider mmap()'ing files into a backend?
1739 Doing I/O to large tables would consume a lot of address space or
1740 require frequent mapping/unmapping. Extending the file also causes
1741 mapping problems that might require mapping only individual pages,
1742 leading to thousands of mappings. Another problem is that there is no
1743 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1744 could hit disk before WAL is written.
1746 * Add a script to ask system configuration questions and tune postgresql.conf
1747 * Consider ways of storing rows more compactly on disk
1749 o Reduce the row header size?
1750 o Consider reducing on-disk varlena length from four bytes to
1751 two because a heap row cannot be more than 64k in length
1753 * Consider increasing NUM_CLOG_BUFFERS
1755 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1756 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1758 * Consider transaction start/end performance improvements
1760 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00948.php
1761 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1763 * Allow user configuration of TOAST thresholds
1765 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1766 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1768 * Allow configuration of backend priorities via the operating system
1770 Though backend priorities make priority inversion during lock
1771 waits possible, research shows that this is not a huge problem.
1773 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1775 * Experiment with multi-threaded backend better I/O utilization
1777 This would allow a single query to make use of multiple I/O channels
1778 simultaneously. One idea is to create a background reader that can
1779 pre-fetch sequential and index scan pages needed by other backends.
1780 This could be expanded to allow concurrent reads from multiple devices
1781 in a partitioned table.
1783 * Experiment with multi-threaded backend better CPU utilization
1785 This would allow several CPUs to be used for a single query, such as
1786 for sorting or query execution.
1788 * Consider increasing the minimum allowed number of shared buffers
1790 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1792 * Expire published xmin for read-only and idle transactions
1794 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1796 * Consider if CommandCounterIncrement() can avoid its
1797 AcceptInvalidationMessages() call
1799 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1801 * Improve performance of shared invalidation queue for multiple CPUs
1803 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1805 * Consider Cartesian joins when both relations are needed to form an
1806 indexscan qualification for a third relation
1808 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1810 * Consider not storing a NULL bitmap on disk if all the NULLs are
1813 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1814 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1816 * Sort large UPDATE/DELETEs so it is done in heap order
1818 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1820 * -Avoid tuple some tuple copying in sort routines
1821 * SMP scalability improvements
1823 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1824 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1825 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1827 * Research reducing deTOASTing in more places
1829 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1831 * Consider being smarter about memory and external files used during
1834 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01101.php
1835 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00045.php
1837 * Allow one transaction to see tuples using the snapshot of another
1840 This would assist multiple backends in working together.
1841 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00400.php
1847 * Add use of 'const' for variables in source tree
1848 * Move some things from contrib into main tree
1849 * %Remove warnings created by -Wcast-align
1850 * Move platform-specific ps status display info from ps_status.c to ports
1851 * Add optional CRC checksum to heap and index pages
1852 * Improve documentation to build only interfaces (Marc)
1853 * Remove or relicense modules that are not under the BSD license, if possible
1854 * Acquire lock on a relation before building a relcache entry for it
1855 * Allow cross-compiling by generating the zic database on the target system
1856 * Improve NLS maintenance of libpgport messages linked onto applications
1857 * Clean up casting in contrib/isn
1859 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1861 * Use UTF8 encoding for NLS messages so all server encodings can
1863 * Update Bonjour to work with newer cross-platform SDK
1865 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1866 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1868 * Consider detoasting keys before sorting
1869 * Consider GnuTLS if OpenSSL license becomes a problem
1871 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1872 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1874 * Consider changing documentation format from SGML to XML
1876 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1878 * Consider making NAMEDATALEN more configurable in future releases
1879 * Update our code to handle 64-bit timezone files to match the zic
1880 source code, which now uses them
1881 * Have configure choose integer datetimes by default
1883 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1885 * Support scoped IPv6 addresses
1887 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1889 * Consider allowing 64-bit integers and floats to be passed by value on
1892 Also change 32-bit floats (float4) to be passed by value at the same
1895 * Research use of signals and sleep wake ups
1897 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1899 * Add automated check for invalid C++ source code constructs
1901 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1903 * Consider simplifying how memory context resets handle child contexts
1905 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1907 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1909 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1911 * Convert single quotes to apostrophes in the PDF documentation
1913 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
1915 * Create three versions of libpgport to simplify client code
1917 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00154.php
1919 * Remove old-style routines for manipulating tuples
1921 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00851.php
1923 * Improve detection of shared memory segments being used by other
1926 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00656.php
1928 * Implement the non-threaded Avahi service discovery protocol
1930 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00939.php
1931 http://archives.postgresql.org/pgsql-patches/2008-02/msg00097.php
1932 http://archives.postgresql.org/pgsql-hackers/2008-03/msg01211.php
1933 http://archives.postgresql.org/pgsql-patches/2008-04/msg00001.php
1937 o Remove configure.in check for link failure when cause is found
1938 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1940 o Remove psql newline patch when we find out why mingw outputs an
1942 o Allow psql to use readline once non-US code pages work with
1944 o Fix problem with shared memory on the Win32 Terminal Server
1945 o Diagnose problem where shared memory can sometimes not be
1946 attached by postmaster children
1948 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
1950 o Improve signal handling
1952 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1954 o Convert MSVC build system to remove most batch files
1956 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
1958 o Prevent SSL from sending network packets to avoid interference
1959 with Win32 signal emulation
1961 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
1963 o Support pgxs when using MSVC
1965 o Fix MSVC NLS support, like for to_char()
1967 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
1968 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
1970 o Find a correct rint() substitute on Windows
1972 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00808.php
1976 * Wire Protocol Changes
1978 o Allow dynamic character set handling
1979 o Add decoded type, length, precision
1981 o Update clients to use data types, typmod, schema.table.column names
1982 of result sets using new statement protocol
1988 * Add pre-parsing phase that converts non-ISO syntax to supported
1991 This could allow SQL written for other databases to run without
1994 * Allow plug-in modules to emulate features from other databases
1995 * Add features of Oracle-style packages (Pavel)
1997 A package would be a schema with session-local variables,
1998 public/private functions, and initialization functions. It
1999 is also possible to implement these capabilities
2000 in any schema and not use a separate "packages"
2003 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
2005 * Consider allowing control of upper/lower case folding of unquoted
2008 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
2009 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
2010 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00849.php
2012 * Add autonomous transactions
2014 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
2018 Features We Do _Not_ Want
2019 =========================
2021 * All backends running as threads in a single process (not wanted)
2023 This eliminates the process protection we get from the current setup.
2024 Thread creation is usually the same overhead as process creation on
2025 modern systems, so it seems unwise to use a pure threaded model.
2027 * Optimizer hints (not wanted)
2029 Optimizer hints are used to work around problems in the optimizer. We
2030 would rather have the problems reported and fixed.
2032 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
2033 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
2034 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
2037 Because we support postfix operators, it isn't possible to make AS
2038 optional and continue to use bison.
2039 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
2041 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
2043 * Embedded server (not wanted)
2045 While PostgreSQL clients runs fine in limited-resource environments, the
2046 server requires multiple processes and a stable pool of resources to
2047 run reliabily and efficiently. Stripping down the PostgreSQL server
2048 to run in the same process address space as the client application
2049 would add too much complexity and failure cases.