3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Mon Mar 24 22:58:06 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 * Add more cross-data-type operators
274 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00189.php
280 o Allow infinite dates and intervals just like infinite timestamps
281 o Merge hardwired timezone names with the TZ database; allow either
282 kind everywhere a TZ name is currently taken
283 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
284 information, either zone name or offset from UTC [timezone]
286 If the TIMESTAMP value is stored with a time zone name, interval
287 computations should adjust based on the time zone rules.
289 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
290 o Add a GUC variable to allow output of interval values in ISO8601
292 o Have timestamp subtraction not call justify_hours()?
294 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
296 o Improve timestamptz subtraction to be DST-aware
298 Currently subtracting one date from another that crosses a
299 daylight savings time adjustment can return '1 day 1 hour', but
300 adding that back to the first date returns a time one hour in
301 the future. This is caused by the adjustment of '25 hours' to
302 '1 day 1 hour', and '1 day' is the same time the next day, even
303 if daylight savings adjustments are involved.
305 o Fix interval display to support values exceeding 2^31 hours
306 o Add overflow checking to timestamp and interval arithmetic
307 o Extend timezone code to allow 64-bit values so we can
308 represent years beyond 2038
310 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
312 o Use LC_TIME for localized weekday/month names, rather than
315 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
317 o Add ISO INTERVAL handling
319 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
320 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
322 o Support ISO INTERVAL syntax if units cannot be determined from
323 the string, and are supplied after the string
325 The SQL standard states that the units after the string
326 specify the units of the string, e.g. INTERVAL '2' MINUTE
327 should return '00:02:00'. The current behavior has the units
328 restrict the interval value to the specified unit or unit
329 range, INTERVAL '70' SECOND returns '00:00:10'.
331 For syntax that isn't uniquely ISO or PG syntax, like '1' or
332 '1:30', treat as ISO if there is a range specification clause,
333 and as PG if there no clause is present, e.g. interpret '1:30'
334 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
335 '1:30' as '1 hour, 30 minutes'.
337 This makes common cases like SELECT INTERVAL '1' MONTH
338 SQL-standard results. The SQL standard supports a limited
339 number of unit combinations and doesn't support unit names in
340 the string. The PostgreSQL syntax is more flexible in the
341 range of units supported, e.g. PostgreSQL supports '1 year 1
342 hour', while the SQL standard does not.
344 o Add support for year-month syntax, INTERVAL '50-6' YEAR
346 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
347 year' AS INTERVAL MONTH), and this should return '12 months'
348 o Round or truncate values to the requested precision, e.g.
349 INTERVAL '11 months' AS YEAR should return one or zero
350 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
355 o Delay resolution of array expression's data type so assignment
356 coercion can be performed on empty array expressions
357 o Add support for arrays of domains
359 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
361 o Allow single-byte header storage for array elements
366 o Improve vacuum of large objects, like contrib/vacuumlo?
367 o Add security checking for large objects
368 o Auto-delete large objects when referencing row is deleted
370 contrib/lo offers this functionality.
372 o Allow read/write into TOAST values like large objects
374 This requires the TOAST column to be stored EXTERNAL.
376 o Add API for 64-bit large object access
378 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
382 * Add locale-aware MONEY type, and support multiple currencies
384 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
385 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
387 * MONEY dumps in a locale-specific format making it difficult to
388 restore to a system with a different locale
389 * Allow MONEY to be easily cast to/from other numeric data types
395 * Allow INET subnet tests using non-constants to be indexed
396 * Allow to_date() and to_timestamp() accept localized month names
397 * Fix to_date()-related functions to consistently issue errors
399 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
401 * Add missing parameter handling in to_char()
403 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
405 * Allow substring/replace() to get/set bit values
406 * Allow to_char() on interval values to accumulate the highest unit
409 Some special format flag would be required to request such
410 accumulation. Such functionality could also be added to EXTRACT.
411 Prevent accumulation that crosses the month/day boundary because of
412 the uneven number of days in a month.
414 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
415 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
416 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
417 o to_char(INTERVAL '3 years 5 months','MM') => 41
419 * Implement inlining of set-returning functions defined in SQL
420 * Allow SQL-language functions to return results from RETURNING queries
422 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
424 * Allow SQL-language functions to reference parameters by parameter name
426 Currently SQL-language functions can only refer to dollar parameters,
429 * Add SPI_gettypmod() to return the typemod for a TupleDesc
430 * Enforce typmod for function inputs, function results and parameters for
431 spi_prepare'd statements called from PLs
433 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
435 * Allow holdable cursors in SPI
436 * Tighten function permission checks
438 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
440 * Fix IS OF so it matches the ISO specification, and add documentation
442 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
443 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
445 * Add missing operators for geometric data types
447 Some geometric types do not have the full suite of geometric operators,
450 * Implement Boyer-Moore searching in strpos()
452 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
454 * Prevent malicious functions from being executed with the permissions
455 of unsuspecting users
457 Index functions are safe, so VACUUM and ANALYZE are safe too.
458 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
459 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
461 * Reduce memory usage of aggregates in set returning functions
463 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
465 * Add temporal versions of generate_series()
467 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
469 * Add array_accum() and array_to_set() functions for arrays
471 The standards specify array_agg() and UNNEST.
472 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
474 * Fix /contrib/ltree operator
476 http://archives.postgresql.org/pgsql-bugs/2007-11/msg00044.php
478 * Fix inconsistent precedence of =, >, and < compared to <>, >=, and <=
480 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00145.php
482 * Fix regular expression bug when using complex back-references
484 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00000.php
486 * Have /contrib/dblink reuse unnamed connections
488 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00895.php
490 * Add SQL-standard array_agg() and unnest() array functions
492 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01017.php
497 Multi-Language Support
498 ======================
500 * Add NCHAR (as distinguished from ordinary varchar),
501 * Allow locale to be set at database creation
503 Currently locale can only be set during initdb. No global tables have
504 locale-aware columns. However, the database template used during
505 database creation might have locale-aware indexes. The indexes would
506 need to be reindexed to match the new locale.
508 * Allow encoding on a per-column basis optionally using the ICU library:
510 Right now only one encoding is allowed per database. [locale]
512 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
513 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
514 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
515 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
517 * Add CREATE COLLATE? [locale]
518 * Support multiple simultaneous character sets, per SQL92
519 * Improve UTF8 combined character handling?
520 * Add octet_length_server() and octet_length_client()
521 * Make octet_length_client() the same as octet_length()?
522 * Fix problems with wrong runtime encoding conversion for NLS message files
523 * Add URL to more complete multi-byte regression tests
525 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
527 * Fix ILIKE and regular expressions to handle case insensitivity
528 properly in multibyte encodings
530 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
531 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
533 * Set client encoding based on the client operating system encoding
535 Currently client_encoding is set in postgresql.conf, which
536 defaults to the server encoding.
537 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
539 * Change memory allocation for multi-byte functions so memory is
540 allocated inside conversion functions
542 Currently we preallocate memory based on worst-case usage.
549 * Automatically create rules on views so they are updateable, per SQL99
551 We can only auto-create rules for simple views. For more complex
552 cases users will still have to write rules manually.
554 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
555 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
557 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
558 * Allow NOTIFY in rules involving conditionals
559 * Allow VIEW/RULE recompilation when the underlying tables change
561 Another issue is whether underlying table changes should be reflected
562 in the view, e.g. should SELECT * show additional columns if they
563 are added after the view is created.
565 * Make it possible to use RETURNING together with conditional DO INSTEAD
566 rules, such as for partitioning setups
568 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
570 * Add the ability to automatically create materialized views
572 Right now materialized views require the user to create triggers on the
573 main table to keep the summary table current. SQL syntax should be able
574 to manager the triggers and summary table automatically. A more
575 sophisticated implementation would automatically retrieve from the
576 summary table when the main table is referenced, if possible.
583 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
584 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
585 * %Allow SET CONSTRAINTS to be qualified by schema/table name
586 * %Add a separate TRUNCATE permission
588 Currently only the owner can TRUNCATE a table because triggers are not
589 called, and the table is locked in exclusive mode.
591 * Allow PREPARE of cursors
592 * Allow finer control over the caching of prepared query plans
594 Currently queries prepared via the libpq API are planned on first
595 execute using the supplied parameters --- allow SQL PREPARE to do the
596 same. Also, allow control over replanning prepared queries either
597 manually or automatically when statistics for execute parameters
598 differ dramatically from those used during planning.
600 * Improve logging of prepared transactions recovered during startup
602 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
604 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
606 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
607 such information in memory would improve performance.
609 * Add optional textual message to NOTIFY
611 This would allow an informational message to be added to the notify
612 message, perhaps indicating the row modified or other custom
615 * Allow multiple identical NOTIFY events to always be communicated to the
616 client, rather than sent as a single notification to the listener
618 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
620 * Add a GUC variable to warn about non-standard SQL usage in queries
621 * Add SQL-standard MERGE command, typically used to merge two tables
624 This is similar to UPDATE, then for unmatched rows, INSERT.
625 Whether concurrent access allows modifications which could cause
626 row loss is implementation independent.
628 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
631 To implement this cleanly requires that the table have a unique index
632 so duplicate checking can be easily performed. It is possible to
633 do it without a unique index if we require the user to LOCK the table
636 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
637 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
639 * Add NOVICE output level for helpful messages like automatic sequence/index
641 * Add GUC to issue notice about statements that use unjoined tables
642 * Allow EXPLAIN to identify tables that were skipped because of
644 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
645 * Enable standard_conforming_strings
646 * Make standard_conforming_strings the default in 8.5?
648 When this is done, backslash-quote should be prohibited in non-E''
649 strings because of possible confusion over how such strings treat
650 backslashes. Basically, '' is always safe for a literal single
651 quote, while \' might or might not be based on the backslash
654 * Simplify dropping roles that have objects in several databases
655 * Allow COMMENT ON to accept an expression rather than just a string
656 * Allow the count returned by SELECT, etc to be represented as an int64
657 to allow a higher range of values
658 * Add SQL99 WITH clause to SELECT
659 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
661 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
662 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
663 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
664 http://archives.postgresql.org/pgsql-hackers/2007-11/msg01334.php
666 * Add DEFAULT .. AS OWNER so permission checks are done as the table
669 This would be useful for SERIAL nextval() calls and CHECK constraints.
671 * Allow DISTINCT to work in multiple-argument aggregate calls
672 * Add column to pg_stat_activity that shows the progress of long-running
673 commands like CREATE INDEX and VACUUM
674 * Implement SQL:2003 window functions
675 * Improve failure message when DROP DATABASE is used on a database that
676 has prepared transactions
677 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
679 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
680 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
682 * Increase locking when DROPing objects so dependent objects cannot
683 get dropped while the DROP operation is happening
685 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
687 * -Allow AS in "SELECT col AS label" to be optional in certain cases
689 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
690 RETURNING to supply values to the INSERT
691 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
693 * Add comments on system tables/columns using the information in
696 Ideally the information would be pulled from the SGML file
699 * Improve reporting of UNION type mismatches
701 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00944.php
702 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00597.php
707 o Allow CREATE TABLE AS to determine column lengths for complex
708 expressions like SELECT col1 || col2
710 o Have WITH CONSTRAINTS also create constraint indexes
712 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
714 o Have CONSTRAINT cname NOT NULL record the contraint name
716 Right now pg_attribute.attnotnull records the NOT NULL status
717 of the column, but does not record the contraint name
719 o Prevent concurrent CREATE TABLE table1 from sometimes returning
720 a cryptic error message
722 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00169.php
726 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
728 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
729 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
730 http://archives.postgresql.org/pgsql-patches/2007-04/msg00315.php
731 http://archives.postgresql.org/pgsql-patches/2008-03/msg00237.php
733 o Research self-referential UPDATEs that see inconsistent row versions
734 in read-committed mode
736 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
737 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
739 o Allow GLOBAL temporary tables to exist as empty by default in
742 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
747 o Have ALTER TABLE RENAME rename SERIAL sequence names
749 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
751 o Have ALTER SEQUENCE RENAME rename the sequence name stored
752 in the sequence table
754 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
755 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
756 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
758 o Add ALTER DOMAIN to modify the underlying data type
759 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
761 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
763 o %Allow ALTER TABLE to change constraint deferrability and actions
764 o Add missing object types for ALTER ... SET SCHEMA
765 o Allow ALTER TABLESPACE to move to different directories
766 o Allow databases to be moved to different tablespaces
767 o Allow moving system tables to other tablespaces, where possible
769 Currently non-global system tables must be in the default database
770 tablespace. Global system tables can never be moved.
772 o Prevent parent tables from altering or dropping constraints
773 like CHECK that are inherited by child tables unless CASCADE
775 o %Prevent child tables from altering or dropping constraints
776 like CHECK that were inherited from the parent table
777 o Have ALTER INDEX update the name of a constraint using that index
778 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
779 o Allow column display reordering by recording a display,
780 storage, and permanent id for every column?
782 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
788 o Automatically maintain clustering on a table
790 This might require some background daemon to maintain clustering
791 during periods of low usage. It might also require tables to be only
792 partially filled for easier reorganization. Another idea would
793 be to create a merged heap/index data file so an index lookup would
794 automatically access the heap data too. A third idea would be to
795 store heap rows in hashed groups, perhaps using a user-supplied
797 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
799 o %Add default clustering to system tables
801 To do this, determine the ideal cluster index for each system
802 table and set the cluster setting during initdb.
804 o %Add VERBOSE option to report tables as they are processed,
810 o Allow COPY to report error lines and continue
812 This requires the use of a savepoint before each COPY line is
813 processed, with ROLLBACK on COPY failure.
814 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00572.php
816 o Allow COPY on a newly-created table to skip WAL logging
818 On crash recovery, the table involved in the COPY would
819 be removed or have its heap and index files truncated. One
820 issue is that no other backend should be able to add to
821 the table at the same time, which is something that is
822 currently allowed. This currently is done if the table is
823 created inside the same transaction block as the COPY because
824 no other backends can see the table.
826 o Consider using a ring buffer for COPY FROM
828 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
829 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
831 o Allow COPY FROM to create index entries in bulk
833 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
835 o Allow COPY in CSV mode to control whether a quoted zero-length
836 string is treated as NULL
838 Currently this is always treated as a zero-length string,
839 which generates an error when loading into an integer column
840 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
846 o Allow column-level privileges
847 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
850 The proposed syntax is:
851 GRANT SELECT ON ALL TABLES IN public TO phpuser;
852 GRANT SELECT ON NEW TABLES IN public TO phpuser;
854 o Allow GRANT/REVOKE permissions to be inherited by objects based on
857 o Allow SERIAL sequences to inherit permissions from the base table?
862 o Prevent DROP TABLE from dropping a row referenced by its own open
868 o Allow INSERT/UPDATE of the system-generated oid value for a row
869 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
875 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
880 Referential Integrity
881 =====================
883 * Add MATCH PARTIAL referential integrity
884 * Change foreign key constraint for array -> element to mean element
886 * Fix problem when cascading referential triggers make changes on
887 cascaded tables, seeing the tables in an intermediate state
889 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
890 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
892 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
894 This would allow UPDATE tab SET col = col + 1 to work if col has
895 a unique index. Currently, uniqueness checks are done while the
896 command is being executed, rather than at the end of the statement
898 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
899 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
901 * Optimize referential integrity checks
903 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
904 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
907 Server-Side Languages
908 =====================
911 o Fix RENAME to work on variables other than OLD/NEW
913 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
914 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
915 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
917 o Allow function parameters to be passed by name,
918 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
919 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
920 o Allow listing of record column names, and access to
921 record columns via variables, e.g. columns := r.(*),
924 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
925 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
926 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
928 o Add support for SCROLL cursors
929 o Add support for WITH HOLD cursors
930 o Allow row and record variables to be set to NULL constants,
931 and allow NULL tests on such variables
933 Because a row is not scalar, do not allow assignment
934 from NULL-valued scalars.
936 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
938 o Review handling of MOVE and FETCH
940 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
942 o Improve logic of determining if an identifier is a a
943 variable or column name
945 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
947 o Consider invalidating the cache or keeping seperate cached
948 copies when search_path changes
950 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01009.php
954 o Add table function support to pltcl, plpythonu
955 o Add support for polymorphic arguments and return types to
956 languages other than PL/PgSQL
957 o Add capability to create and call PROCEDURES
958 o Add support for OUT and INOUT parameters to languages other
960 o Add PL/PythonU tracebacks
962 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
964 o Allow data to be passed in native language formats, rather
967 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
974 * Have pg_ctl look at PGHOST in case it is a socket directory?
975 * Allow pg_ctl to work properly with configuration files located outside
978 pg_ctl can not read the pid file because it isn't located in the
979 config directory but in the PGDATA directory. The solution is to
980 allow pg_ctl to read and understand postgresql.conf to find the
981 data_directory value.
983 * Add a function like pg_get_indexdef() that report more detailed index
986 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
988 * Prefix command-line utilities like createuser with 'pg_'
990 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00025.php
994 o Have psql show current values for a sequence
995 o Move psql backslash database information into the backend, use
996 mnemonic commands? [psql]
998 This would allow non-psql clients to pull the same information out
999 of the database as psql.
1001 o Fix psql's \d commands more consistent
1003 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1004 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
1006 o Consistently display privilege information for all objects in psql
1007 o Add auto-expanded mode so expanded output is used if the row
1008 length is wider than the screen width.
1010 Consider using auto-expanded mode for backslash commands like \df+.
1012 o Prevent tab completion of SET TRANSACTION from querying the
1013 database and therefore preventing the transaction isolation
1014 level from being set.
1016 Currently SET <tab> causes a database lookup to check all
1017 supported session variables. This query causes problems
1018 because setting the transaction isolation level must be the
1019 first statement of a transaction.
1021 o Add a \set variable to control whether \s displays line numbers
1023 Another option is to add \# which lists line numbers, and
1024 allows command execution.
1026 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
1028 o Prevent escape string warnings when object names have
1031 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
1033 o Have \d show foreign keys that reference a table's primary key
1035 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
1037 o Have \d show child tables that inherit from the specified parent
1038 o Have \l+ show database size, if permissions allow
1040 Ideally it will not generate an error for invalid permissions
1042 o Include the symbolic SQLSTATE name in verbose error reports
1044 http://archives.postgresql.org/pgsql-general/2007-09/msg00438.php
1046 o Improve display if enums
1048 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00826.php
1053 * pg_dump / pg_restore
1054 o %Add dumping of comments on index columns and composite type columns
1055 o %Add full object name to the tag field. eg. for operators we need
1056 '=(integer, integer)', instead of just '='.
1057 o Add pg_dumpall custom format dumps?
1058 o Allow selection of individual object(s) of all types, not just
1060 o In a selective dump, allow dumping of an object and all its
1062 o Add options like pg_restore -l and -L to pg_dump
1063 o Stop dumping CASCADE on DROP TYPE commands in clean mode
1064 o Allow pg_dump --clean to drop roles that own objects or have
1066 o Change pg_dump so that a comment on the dumped database is
1067 applied to the loaded database, even if the database has a
1068 different name. This will require new backend syntax, perhaps
1069 COMMENT ON CURRENT DATABASE.
1070 o Remove unnecessary function pointer abstractions in pg_dump source
1072 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
1073 multiple objects simultaneously
1075 The difficulty with this is getting multiple dump processes to
1076 produce a single dump output file. It also would require
1077 several sessions to share the same snapshot.
1078 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1080 o Allow pg_restore to utilize multiple CPUs and I/O channels by
1081 restoring multiple objects simultaneously
1083 This might require a pg_restore flag to indicate how many
1084 simultaneous operations should be performed. Only pg_dump's
1085 -Fc format has the necessary dependency information.
1086 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
1088 o To better utilize resources, allow pg_restore to check foreign
1089 keys simultaneously, where possible
1090 o Allow pg_restore to create all indexes of a table
1091 concurrently, via a single heap scan
1093 This requires a pg_dump -Fc file because that format contains
1094 the required dependency information.
1095 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
1097 o Allow pg_restore to load different parts of the COPY data
1099 o Prevent pg_dump/pg_restore from being affected by
1102 Using psql to restore a pg_dump dump is also affected.
1104 o Remove pre-7.3 pg_dump code that assumes pg_depend does not exit
1105 o Allow pre/data/post files when schema and data are dumped
1106 separately, for performance reasons
1108 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
1115 Document differences between ecpg and the SQL standard and
1116 information about the Informix-compatibility module.
1118 o Solve cardinality > 1 for input descriptors / variables?
1119 o Add a semantic check level, e.g. check if a table really exists
1120 o fix handling of DB attributes that are arrays
1121 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1123 o Fix nested C comments
1124 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1125 o Make SET CONNECTION thread-aware, non-standard?
1126 o Allow multidimensional arrays
1127 o Add internationalized message strings
1128 o Implement COPY FROM STDIN
1132 o Add PQescapeIdentifierConn()
1133 o Prevent PQfnumber() from lowercasing unquoted the column name
1135 PQfnumber() should never have been doing lowercasing, but
1136 historically it has so we need a way to prevent it
1138 o Allow statement results to be automatically batched to the client
1140 Currently all statement results are transferred to the libpq
1141 client before libpq makes the results available to the
1142 application. This feature would allow the application to make
1143 use of the first result rows while the rest are transferred, or
1144 held on the server waiting for them to be requested by libpq.
1145 One complexity is that a statement like SELECT 1/col could error
1146 out mid-way through the result set.
1148 o Consider disallowing multiple queries in PQexec() as an
1149 additional barrier to SQL injection attacks
1151 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1153 o Add PQexecf() that allows complex parameter substitution
1155 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1157 o Add SQLSTATE severity to PGconn return status
1159 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1165 * Add deferred trigger queue file
1167 Right now all deferred trigger information is stored in backend
1168 memory. This could exhaust memory for very large trigger queues.
1169 This item involves dumping large queues into files.
1171 * Allow triggers to be disabled in only the current session.
1173 This is currently possible by starting a multi-statement transaction,
1174 modifying the system tables, performing the desired SQL, restoring the
1175 system tables, and committing the transaction. ALTER TABLE ...
1176 TRIGGER requires a table lock so it is not ideal for this usage.
1178 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1180 If the dump is known to be valid, allow foreign keys to be added
1181 without revalidating the data.
1183 * Allow statement-level triggers to access modified rows
1184 * Support triggers on columns
1186 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1188 * Allow AFTER triggers on system tables
1190 System tables are modified in many places in the backend without going
1191 through the executor and therefore not causing triggers to fire. To
1192 complete this item, the functions that modify system tables will have
1195 * Tighten trigger permission checks
1197 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1199 * Allow BEFORE INSERT triggers on views
1201 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1203 * Add ability to trigger on TRUNCATE
1205 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
1212 * Add UNIQUE capability to non-btree indexes
1213 * Prevent index uniqueness checks when UPDATE does not modify the column
1215 Uniqueness (index) checks are done when updating a column even if the
1216 column is not modified by the UPDATE.
1218 * Allow the creation of on-disk bitmap indexes which can be quickly
1219 combined with other bitmap indexes
1221 Such indexes could be more compact if there are only a few distinct values.
1222 Such indexes can also be compressed. Keeping such indexes updated can be
1225 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1226 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1228 * Allow accurate statistics to be collected on indexes with more than
1229 one column or expression indexes, perhaps using per-index statistics
1231 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1232 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1234 * Consider increasing the number of default statistics target, and
1235 reduce statistics target overhead
1237 Also consider having a larger statistics target for indexed columns
1238 and expression indexes
1239 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1240 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1241 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01066.php
1243 * Consider compressing indexes by storing key values duplicated in
1244 several rows as a single index entry
1246 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1247 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1248 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1249 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1251 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1253 This is difficult because you must upgrade to an exclusive table lock
1254 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1255 have this complication. This would allow index compaction without
1257 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1259 * Allow multiple indexes to be created concurrently, ideally via a
1260 single heap scan, and have pg_restore use it
1262 * Consider sorting entries before inserting into btree index
1264 http://archives.postgresql.org/pgsql-general/2008-01/msg01010.php
1270 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1271 and primary/foreign keys
1272 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1273 on inherited table, e.g. INSERT INTO inherit_table
1274 (unique_index_col) VALUES (dup) should fail
1276 The main difficulty with this item is the problem of
1277 creating an index that can span multiple tables.
1279 o Allow SELECT ... FOR UPDATE on inherited tables
1280 o Require all CHECK constraints to be inherited
1282 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1284 o Add checks to prevent a CREATE RULE views on inherited tables
1286 http://archives.postgresql.org/pgsql-general/2008-02/msg01420.php
1287 http://archives.postgresql.org/pgsql-general/2008-03/msg00077.php
1292 o Add more GIST index support for geometric data types
1293 o Allow GIST indexes to create certain complex index types, like
1294 digital trees (see Aoki)
1299 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1301 o Pack hash index buckets onto disk pages more efficiently
1303 Currently only one hash bucket can be stored on a page. Ideally
1304 several hash buckets could be stored on a single page and greater
1305 granularity used for the hash algorithm.
1307 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1309 o Consider sorting hash buckets so entries can be found using a
1310 binary search, rather than a linear scan
1311 o In hash indexes, consider storing the hash value with or instead
1313 o Add WAL logging for crash recovery
1314 o Allow multi-column hash indexes
1315 o -During index creation, pre-sort the tuples to improve build speed
1322 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1324 Ideally this requires a separate test program that can be run
1325 at initdb time or optionally later. Consider O_SYNC when
1328 * Add program to test if fsync has a delay compared to non-fsync
1329 * Consider sorting writes during checkpoint
1331 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1340 We could use a fixed row count and a +/- count to follow MVCC
1341 visibility rules, or a single cached value could be used and
1342 invalidated if anyone modifies the table. Another idea is to
1343 get a count directly from a unique index, but for this to be
1344 faster than a sequential scan it must avoid access to the heap
1345 to obtain tuple visibility information.
1347 * Provide a way to calculate an "estimated COUNT(*)"
1349 Perhaps by using the optimizer's cardinality estimates or random
1352 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1354 * Allow data to be pulled directly from indexes
1356 Currently indexes do not have enough tuple visibility information
1357 to allow data to be pulled from the index without also accessing
1358 the heap. One way to allow this is to set a bit on index tuples
1359 to indicate if a tuple is currently visible to all transactions
1360 when the first valid heap lookup happens. This bit would have to
1361 be cleared when a heap tuple is expired.
1363 Another idea is to maintain a bitmap of heap pages where all rows
1364 are visible to all backends, and allow index lookups to reference
1365 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1366 add someday to determine which heap pages need vacuuming. Frequently
1367 accessed bitmaps would have to be stored in shared memory. One 8k
1368 page of bitmaps could track 512MB of heap pages.
1370 A third idea would be for a heap scan to check if all rows are visible
1371 and if so set a per-table flag which can be checked by index scans.
1372 Any change to the table would have to clear the flag. To detect
1373 changes during the heap scan a counter could be set at the start and
1374 checked at the end --- if it is the same, the table has not been
1375 modified --- any table change would increment the counter.
1377 * Consider automatic caching of statements at various levels:
1380 o Query execute plan
1383 * Consider increasing internal areas when shared buffers is increased
1385 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1387 * Consider decreasing the amount of memory used by PrivateRefCount
1389 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1390 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1392 * Consider allowing higher priority queries to have referenced buffer
1393 cache pages stay in memory longer
1395 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1402 * Improve speed with indexes
1404 For large table adjustments during VACUUM FULL, it is faster to cluster
1405 or reindex rather than update the index. Also, index updates can bloat
1408 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1409 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1410 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1412 * Auto-fill the free space map by scanning the buffer cache or by
1413 checking pages written by the background writer
1415 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1416 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1418 * Create a bitmap of pages that need vacuuming
1420 Instead of sequentially scanning the entire table, have the background
1421 writer or some other process record pages that have expired rows, then
1422 VACUUM can look at just those pages rather than the entire table. In
1423 the event of a system crash, the bitmap would probably be invalidated.
1424 One complexity is that index entries still have to be vacuumed, and
1425 doing this without an index scan (by using the heap values to find the
1426 index entry) might be slow and unreliable, especially for user-defined
1429 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1430 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1432 * Allow FSM to return free space toward the beginning of the heap file,
1433 in hopes that empty pages at the end can be truncated by VACUUM
1434 * Allow FSM page return free space based on table clustering, to assist
1435 in maintaining clustering?
1436 * Improve dead row detection during multi-statement transactions usage
1438 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1440 * Consider a more compact data representation for dead tuples
1442 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1447 o %Issue log message to suggest VACUUM FULL if a table is nearly
1449 o Improve control of auto-vacuum
1451 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1453 o Prevent long-lived temporary tables from causing frozen-xid
1454 advancement starvation
1456 The problem is that autovacuum cannot vacuum them to set frozen xids;
1457 only the session that created them can do that.
1458 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1460 o Store per-table autovacuum settings in pg_class.reloptions.
1462 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1463 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1465 o Prevent autovacuum from running if an old transaction is still
1466 running from the last vacuum
1468 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1475 * Fix priority ordering of read and write light-weight locks (Neil)
1477 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1478 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1480 * Fix problem when multiple subtransactions of the same outer transaction
1481 hold different types of locks, and one subtransaction aborts
1483 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1484 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1485 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1486 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1488 * Allow UPDATEs on only non-referential integrity columns not to conflict
1489 with referential integrity locks
1491 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1493 * Add idle_in_transaction_timeout GUC so locks are not held for long
1496 * Improve deadlock detection when a page cleaning lock conflicts
1497 with a shared buffer that is pinned
1499 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1500 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1501 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1503 * Detect deadlocks involving LockBufferForCleanup()
1505 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1509 Startup Time Improvements
1510 =========================
1512 * Experiment with multi-threaded backend for backend creation [thread]
1514 This would prevent the overhead associated with process creation. Most
1515 operating systems have trivial process creation time compared to
1516 database startup overhead, but a few operating systems (Win32,
1517 Solaris) might benefit from threading. Also explore the idea of
1518 a single session using multiple threads to execute a statement faster.
1525 * Eliminate need to write full pages to WAL before page modification [wal]
1527 Currently, to protect against partial disk page writes, we write
1528 full page images to WAL before they are modified so we can correct any
1529 partial page writes during recovery. These pages can also be
1530 eliminated from point-in-time archive files.
1532 o When off, write CRC to WAL and check file system blocks
1535 If CRC check fails during recovery, remember the page in case
1536 a later CRC for that page properly matches.
1538 o Write full pages during file system write and not when
1539 the page is modified in the buffer cache
1541 This allows most full page writes to happen in the background
1542 writer. It might cause problems for applying WAL on recovery
1543 into a partially-written page, but later the full page will be
1546 * Allow WAL traffic to be streamed to another server for stand-by
1548 * Reduce WAL traffic so only modified values are written rather than
1551 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1553 * Allow WAL information to recover corrupted pg_controldata
1555 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1557 * Find a way to reduce rotational delay when repeatedly writing
1560 Currently fsync of WAL requires the disk platter to perform a full
1561 rotation to fsync again. One idea is to write the WAL to different
1562 offsets that might reduce the rotational delay.
1564 * Allow WAL logging to be turned off for a table, but the table
1565 might be dropped or truncated during crash recovery [walcontrol]
1567 Allow tables to bypass WAL writes and just fsync() dirty pages on
1568 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1569 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1570 non-default logging should not use referential integrity with
1571 default-logging tables. A table without dirty buffers during a
1572 crash could perhaps avoid the drop/truncate.
1574 * Allow WAL logging to be turned off for a table, but the table would
1575 avoid being truncated/dropped [walcontrol]
1577 To do this, only a single writer can modify the table, and writes
1578 must happen only on new pages so the new pages can be removed during
1579 crash recovery. Readers can continue accessing the table. Such
1580 tables probably cannot have indexes. One complexity is the handling
1581 of indexes on TOAST tables.
1583 * Speed WAL recovery by allowing more than one page to be prefetched
1585 This should be done utilizing the same infrastructure used for
1586 prefetching in general to avoid introducing complex error-prone code
1588 http://archives.postgresql.org/pgsql-general/2007-12/msg00683.php
1589 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00497.php
1590 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1592 * Improve WAL concurrency by increasing lock granularity
1594 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00556.php
1596 * Be more aggressive about creating WAL files
1598 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01325.php
1599 http://archives.postgresql.org/pgsql-hackers/2004-07/msg01075.php
1600 http://archives.postgresql.org/pgsql-hackers/2005-04/msg00556.php
1602 * Have resource managers report the duration of their status changes
1604 http://archives.postgresql.org/pgsql-hackers/2007-10/msg01468.php
1606 * Move pgfoundry's xlogdump to /contrib and have it rely more closely
1607 on the WAL backend code
1609 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00035.php
1614 Optimizer / Executor
1615 ====================
1617 * Improve selectivity functions for geometric operators
1618 * Precompile SQL functions to avoid overhead
1619 * Create utility to compute accurate random_page_cost value
1620 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1621 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1622 actual row counts differ by a specified percentage
1623 * Consider using hash buckets to do DISTINCT, rather than sorting
1625 This would be beneficial when there are few distinct values. This is
1626 already used by GROUP BY.
1628 * Log statements where the optimizer row estimates were dramatically
1629 different from the number of rows actually found?
1630 * Consider compressed annealing to search for query plans
1632 This might replace GEQO, http://sixdemonbag.org/Djinni.
1634 * Improve merge join performance by allowing mark/restore of
1637 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1639 * Consider using a hash for joining to a large IN (VALUES ...) list
1641 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1648 * Consider having the background writer update the transaction status
1649 hint bits before writing out the page
1651 Implementing this requires the background writer to have access to system
1652 catalogs and the transaction status log.
1654 * Consider adding buffers the background writer finds reusable to the
1657 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1659 * Automatically tune bgwriter_delay based on activity rather then using a
1662 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1664 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1666 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1668 * Test to see if calling PreallocXlogFiles() from the background writer
1669 will help with WAL segment creation latency
1671 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1675 Miscellaneous Performance
1676 =========================
1678 * Do async I/O for faster random read-ahead of data
1680 Async I/O allows multiple I/O requests to be sent to the disk with
1681 results coming back asynchronously.
1683 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1684 http://archives.postgresql.org/pgsql-performance/2007-09/msg00255.php
1685 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1686 http://archives.postgresql.org/pgsql-patches/2008-01/msg00170.php
1688 * Use mmap() rather than SYSV shared memory or to write WAL files?
1690 This would remove the requirement for SYSV SHM but would introduce
1691 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1692 to prevent I/O overhead.
1694 * Consider mmap()'ing files into a backend?
1696 Doing I/O to large tables would consume a lot of address space or
1697 require frequent mapping/unmapping. Extending the file also causes
1698 mapping problems that might require mapping only individual pages,
1699 leading to thousands of mappings. Another problem is that there is no
1700 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1701 could hit disk before WAL is written.
1703 * Add a script to ask system configuration questions and tune postgresql.conf
1704 * Consider ways of storing rows more compactly on disk
1706 o Reduce the row header size?
1707 o Consider reducing on-disk varlena length from four bytes to
1708 two because a heap row cannot be more than 64k in length
1710 * Consider increasing NUM_CLOG_BUFFERS
1712 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1713 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1715 * Allow user configuration of TOAST thresholds
1717 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1718 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1720 * Allow configuration of backend priorities via the operating system
1722 Though backend priorities make priority inversion during lock
1723 waits possible, research shows that this is not a huge problem.
1725 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1727 * Experiment with multi-threaded backend better I/O utilization
1729 This would allow a single query to make use of multiple I/O channels
1730 simultaneously. One idea is to create a background reader that can
1731 pre-fetch sequential and index scan pages needed by other backends.
1732 This could be expanded to allow concurrent reads from multiple devices
1733 in a partitioned table.
1735 * Experiment with multi-threaded backend better CPU utilization
1737 This would allow several CPUs to be used for a single query, such as
1738 for sorting or query execution.
1740 * Consider increasing the minimum allowed number of shared buffers
1742 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1744 * Expire published xmin for read-only and idle transactions
1746 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00343.php
1748 * Consider if CommandCounterIncrement() can avoid its
1749 AcceptInvalidationMessages() call
1751 http://archives.postgresql.org/pgsql-committers/2007-11/msg00585.php
1753 * Improve performance of shared invalidation queue for multiple CPUs
1755 http://archives.postgresql.org/pgsql-performance/2008-01/msg00023.php
1757 * Consider Cartesian joins when both relations are needed to form an
1758 indexscan qualification for a third relation
1760 http://archives.postgresql.org/pgsql-performance/2007-12/msg00090.php
1762 * Consider not storing a NULL bitmap on disk if all the NULLs are
1765 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00624.php
1766 http://archives.postgresql.org/pgsql-patches/2007-12/msg00109.php
1768 * Sort large UPDATE/DELETEs so it is done in heap order
1770 http://archives.postgresql.org/pgsql-hackers/2008-01/msg01119.php
1772 * Avoid tuple some tuple copying in sort routines
1774 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01206.php
1775 http://archives.postgresql.org/pgsql-patches/2008-02/msg00176.php
1777 * SMP scalability improvements
1779 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00439.php
1780 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00206.php
1781 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00361.php
1783 * Research reducing deTOASTing in more places
1785 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00895.php
1792 * Add use of 'const' for variables in source tree
1793 * Move some things from contrib into main tree
1794 * %Remove warnings created by -Wcast-align
1795 * Move platform-specific ps status display info from ps_status.c to ports
1796 * Add optional CRC checksum to heap and index pages
1797 * Improve documentation to build only interfaces (Marc)
1798 * Remove or relicense modules that are not under the BSD license, if possible
1799 * Acquire lock on a relation before building a relcache entry for it
1800 * Allow cross-compiling by generating the zic database on the target system
1801 * Improve NLS maintenance of libpgport messages linked onto applications
1802 * Clean up casting in contrib/isn
1804 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1806 * Use UTF8 encoding for NLS messages so all server encodings can
1808 * Update Bonjour to work with newer cross-platform SDK
1810 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1811 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1813 * Consider detoasting keys before sorting
1814 * Consider GnuTLS if OpenSSL license becomes a problem
1816 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1817 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1819 * Consider changing documentation format from SGML to XML
1821 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1823 * Consider making NAMEDATALEN more configurable in future releases
1824 * Update our code to handle 64-bit timezone files to match the zic
1825 source code, which now uses them
1826 * Have configure choose integer datetimes by default
1828 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1830 * Support scoped IPv6 addresses
1832 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1834 * Consider allowing 64-bit integers and floats to be passed by value on
1837 Also change 32-bit floats (float4) to be passed by value at the same
1840 * Research use of signals and sleep wake ups
1842 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1844 * Add automated check for invalid C++ source code constructs
1846 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1848 * Consider simplifying how memory context resets handle child contexts
1850 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1852 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1854 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1856 * Convert single quotes to apostrophes in the PDF documentation
1858 http://archives.postgresql.org/pgsql-docs/2007-12/msg00059.php
1860 * Create three versions of libpgport to simplify client code
1862 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00154.php
1864 * Remove old-style routines for manipulating tuples
1866 http://archives.postgresql.org/pgsql-hackers/2007-10/msg00851.php
1868 * Improve detection of shared memory segments being used by other
1871 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00656.php
1877 o Remove configure.in check for link failure when cause is found
1878 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1880 o Remove psql newline patch when we find out why mingw outputs an
1882 o Allow psql to use readline once non-US code pages work with
1884 o Fix problem with shared memory on the Win32 Terminal Server
1885 o Diagnose problem where shared memory can sometimes not be
1886 attached by postmaster children
1888 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
1890 o Improve signal handling
1892 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1894 o Convert MSVC build system to remove most batch files
1896 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
1898 o Prevent SSL from sending network packets to avoid interference
1899 with Win32 signal emulation
1901 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00455.php
1903 o Support pgxs when using MSVC
1905 o Fix MSVC NLS support, like for to_char()
1907 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00485.php
1908 http://archives.postgresql.org/pgsql-patches/2008-02/msg00038.php
1910 o Fix port/rint.c to be spec-compliant
1912 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00808.php
1916 * Wire Protocol Changes
1918 o Allow dynamic character set handling
1919 o Add decoded type, length, precision
1921 o Update clients to use data types, typmod, schema.table.column names
1922 of result sets using new statement protocol
1928 * Add pre-parsing phase that converts non-ISO syntax to supported
1931 This could allow SQL written for other databases to run without
1934 * Allow plug-in modules to emulate features from other databases
1935 * Add features of Oracle-style packages (Pavel)
1937 A package would be a schema with session-local variables,
1938 public/private functions, and initialization functions. It
1939 is also possible to implement these capabilities
1940 in any schema and not use a separate "packages"
1943 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1945 * Consider allowing control of upper/lower case folding of unquoted
1948 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1949 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1951 * Add autonomous transactions
1953 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1957 Features We Do _Not_ Want
1958 =========================
1960 * All backends running as threads in a single process (not wanted)
1962 This eliminates the process protection we get from the current setup.
1963 Thread creation is usually the same overhead as process creation on
1964 modern systems, so it seems unwise to use a pure threaded model.
1966 * Optimizer hints (not wanted)
1968 Optimizer hints are used to work around problems in the optimizer. We
1969 would rather have the problems reported and fixed.
1971 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1972 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1973 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1976 Because we support postfix operators, it isn't possible to make AS
1977 optional and continue to use bison.
1978 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1980 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1982 * Embedded server (not wanted)
1984 While PostgreSQL clients runs fine in limited-resource environments, the
1985 server requires multiple processes and a stable pool of resources to
1986 run reliabily and efficiently. Stripping down the PostgreSQL server
1987 to run in the same process address space as the client application
1988 would add too much complexity and failure cases.