3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Wed Mar 12 15:41:18 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 * Allow statistics last vacuum/analyze execution times to be displayed
51 without requiring stats_row_level to be enabled
52 * Allow log_min_messages to be specified on a per-module basis
54 This would allow administrators to see more detailed information from
55 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
56 Another idea is to allow separate configuration files for each module,
57 or allow arbitrary SET commands to be passed to them.
59 * Simplify ability to create partitioned tables
61 This would allow creation of partitioned tables without requiring
62 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
63 for rapid partition selection. Options could include range and hash
66 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
67 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
69 * Allow auto-selection of partitioned tables for min/max() operations
70 * Allow more complex user/database default GUC settings
72 Currently ALTER USER and ALTER DATABASE support per-user and
73 per-database defaults. Consider adding per-user-and-database
74 defaults so things like search_path can be defaulted for a
75 specific user connecting to a specific database.
77 * Allow custom variable classes that can restrict who can set the values
79 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
81 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
82 the privilege granted by the invoking role, and not those granted
85 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
87 * Allow SSL authentication/encryption over unix domain sockets
89 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
91 * Allow client certificate names to be checked against the client
94 This is already implemented in
95 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
100 o Allow pg_hba.conf to specify host names along with IP addresses
102 Host name lookup could occur when the postmaster reads the
103 pg_hba.conf file, or when the backend starts. Another
104 solution would be to reverse lookup the connection IP and
105 check that hostname against the host names in pg_hba.conf.
106 We could also then check that the host name maps to the IP
109 o %Allow postgresql.conf file values to be changed via an SQL
110 API, perhaps using SET GLOBAL
111 o Allow the server to be stopped/restarted via an SQL API
112 o Issue a warning if a change-on-restart-only postgresql.conf value
113 is modified and the server config files are reloaded
114 o Consider normalizing fractions in postgresql.conf, perhaps
117 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00550.php
122 o Allow a database in tablespace t1 with tables created in
123 tablespace t2 to be used as a template for a new database created
124 with default tablespace t2
126 Currently all objects in the default database tablespace must
127 have default tablespace specifications. This is because new
128 databases are created by copying directories. If you mix default
129 tablespace tables and tablespace-specified tables in the same
130 directory, creating a new database from such a mixed directory
131 would create a new database with tables that had incorrect
132 explicit tablespaces. To fix this would require modifying
133 pg_class in the newly copied database, which we don't currently
136 o Allow reporting of which objects are in which tablespaces
138 This item is difficult because a tablespace can contain objects
139 from multiple databases. There is a server-side function that
140 returns the databases which use a specific tablespace, so this
141 requires a tool that will call that function and connect to each
142 database to find the objects in each database for that tablespace.
144 o Allow WAL replay of CREATE TABLESPACE to work when the directory
145 structure on the recovery computer is different from the original
147 o Allow per-tablespace quotas
150 * Point-In-Time Recovery (PITR)
152 o Allow a warm standby system to also allow read-only statements
155 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
157 o %Create dump tool for write-ahead logs for use in determining
158 transaction id for point-in-time recovery
160 This is useful for checking PITR recovery.
162 o Allow recovery.conf to support the same syntax as
163 postgresql.conf, including quoting
165 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
171 * Change NUMERIC to enforce the maximum precision
172 * Reduce storage space for small NUMERICs
174 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
175 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
176 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
178 * Fix data types where equality comparison isn't intuitive, e.g. box
179 * Add support for public SYNONYMs
181 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
183 * Fix CREATE CAST on DOMAINs
185 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
186 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
188 * Allow domains to be cast
190 http://archives.postgresql.org/pgsql-hackers/2003-06/msg01206.php
191 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00289.php
193 * Add support for SQL-standard GENERATED/IDENTITY columns
195 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
196 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
197 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
198 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
199 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
201 * Improve XML support
203 http://developer.postgresql.org/index.php/XML_Support
205 * Consider placing all sequences in a single table, or create a system
208 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
210 * Allow the UUID type to accept non-standard formats
212 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
214 * Allow text search dictionary to filter out only stop words
216 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
218 * Consider a function-based API for '@@' full text searches
220 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
222 * Consider a special data type for regular expressions
224 http://archives.postgresql.org/pgsql-hackers/2007-08/msg01067.php
229 o Allow infinite dates and intervals just like infinite timestamps
230 o Merge hardwired timezone names with the TZ database; allow either
231 kind everywhere a TZ name is currently taken
232 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
233 information, either zone name or offset from UTC [timezone]
235 If the TIMESTAMP value is stored with a time zone name, interval
236 computations should adjust based on the time zone rules.
238 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
239 o Add a GUC variable to allow output of interval values in ISO8601
241 o Have timestamp subtraction not call justify_hours()?
243 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
245 o Improve timestamptz subtraction to be DST-aware
247 Currently subtracting one date from another that crosses a
248 daylight savings time adjustment can return '1 day 1 hour', but
249 adding that back to the first date returns a time one hour in
250 the future. This is caused by the adjustment of '25 hours' to
251 '1 day 1 hour', and '1 day' is the same time the next day, even
252 if daylight savings adjustments are involved.
254 o Fix interval display to support values exceeding 2^31 hours
255 o Add overflow checking to timestamp and interval arithmetic
256 o Extend timezone code to allow 64-bit values so we can
257 represent years beyond 2038
259 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
261 o Use LC_TIME for localized weekday/month names, rather than
264 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
266 o Add ISO INTERVAL handling
268 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
269 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
271 o Support ISO INTERVAL syntax if units cannot be determined from
272 the string, and are supplied after the string
274 The SQL standard states that the units after the string
275 specify the units of the string, e.g. INTERVAL '2' MINUTE
276 should return '00:02:00'. The current behavior has the units
277 restrict the interval value to the specified unit or unit
278 range, INTERVAL '70' SECOND returns '00:00:10'.
280 For syntax that isn't uniquely ISO or PG syntax, like '1' or
281 '1:30', treat as ISO if there is a range specification clause,
282 and as PG if there no clause is present, e.g. interpret '1:30'
283 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
284 '1:30' as '1 hour, 30 minutes'.
286 This makes common cases like SELECT INTERVAL '1' MONTH
287 SQL-standard results. The SQL standard supports a limited
288 number of unit combinations and doesn't support unit names in
289 the string. The PostgreSQL syntax is more flexible in the
290 range of units supported, e.g. PostgreSQL supports '1 year 1
291 hour', while the SQL standard does not.
293 o Add support for year-month syntax, INTERVAL '50-6' YEAR
295 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
296 year' AS INTERVAL MONTH), and this should return '12 months'
297 o Round or truncate values to the requested precision, e.g.
298 INTERVAL '11 months' AS YEAR should return one or zero
299 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
304 o Delay resolution of array expression's data type so assignment
305 coercion can be performed on empty array expressions
306 o Add support for arrays of domains
308 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
310 o Allow single-byte header storage for array elements
315 o Improve vacuum of large objects, like contrib/vacuumlo?
316 o Add security checking for large objects
317 o Auto-delete large objects when referencing row is deleted
319 contrib/lo offers this functionality.
321 o Allow read/write into TOAST values like large objects
323 This requires the TOAST column to be stored EXTERNAL.
325 o Add API for 64-bit large object access
327 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
331 * Add locale-aware MONEY type, and support multiple currencies
333 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
334 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
336 * MONEY dumps in a locale-specific format making it difficult to
337 restore to a system with a different locale
338 * Allow MONEY to be easily cast to/from other numeric data types
344 * Allow INET subnet tests using non-constants to be indexed
345 * Allow to_date() and to_timestamp() accept localized month names
346 * Fix to_date()-related functions to consistently issue errors
348 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
350 * Add missing parameter handling in to_char()
352 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
354 * Allow substring/replace() to get/set bit values
355 * Allow to_char() on interval values to accumulate the highest unit
358 Some special format flag would be required to request such
359 accumulation. Such functionality could also be added to EXTRACT.
360 Prevent accumulation that crosses the month/day boundary because of
361 the uneven number of days in a month.
363 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
364 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
365 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
366 o to_char(INTERVAL '3 years 5 months','MM') => 41
368 * Implement inlining of set-returning functions defined in SQL
369 * Allow SQL-language functions to return results from RETURNING queries
371 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
373 * Allow SQL-language functions to reference parameters by parameter name
375 Currently SQL-language functions can only refer to dollar parameters,
378 * Add SPI_gettypmod() to return the typemod for a TupleDesc
379 * Enforce typmod for function inputs, function results and parameters for
380 spi_prepare'd statements called from PLs
382 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
384 * Allow holdable cursors in SPI
385 * Tighten function permission checks
387 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
389 * Fix IS OF so it matches the ISO specification, and add documentation
391 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
392 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
394 * Add missing operators for geometric data types
396 Some geometric types do not have the full suite of geometric operators,
399 * Implement Boyer-Moore searching in strpos()
401 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
403 * Prevent malicious functions from being executed with the permissions
404 of unsuspecting users
406 Index functions are safe, so VACUUM and ANALYZE are safe too.
407 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
408 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
410 * Reduce memory usage of aggregates in set returning functions
412 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
414 * Add temporal versions of generate_series()
416 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
418 * Add array_accum() and array_to_set() functions for arrays
420 The standards specify array_agg() and UNNEST.
421 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00464.php
425 Multi-Language Support
426 ======================
428 * Add NCHAR (as distinguished from ordinary varchar),
429 * Allow locale to be set at database creation
431 Currently locale can only be set during initdb. No global tables have
432 locale-aware columns. However, the database template used during
433 database creation might have locale-aware indexes. The indexes would
434 need to be reindexed to match the new locale.
436 * Allow encoding on a per-column basis optionally using the ICU library:
438 Right now only one encoding is allowed per database. [locale]
440 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
441 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
442 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
443 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
445 * Add CREATE COLLATE? [locale]
446 * Support multiple simultaneous character sets, per SQL92
447 * Improve UTF8 combined character handling?
448 * Add octet_length_server() and octet_length_client()
449 * Make octet_length_client() the same as octet_length()?
450 * Fix problems with wrong runtime encoding conversion for NLS message files
451 * Add URL to more complete multi-byte regression tests
453 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
455 * Fix ILIKE and regular expressions to handle case insensitivity
456 properly in multibyte encodings
458 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
459 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
461 * Set client encoding based on the client operating system encoding
463 Currently client_encoding is set in postgresql.conf, which
464 defaults to the server encoding.
465 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
467 * Change memory allocation for multi-byte functions so memory is
468 allocated inside conversion functions
470 Currently we preallocate memory based on worst-case usage.
477 * Automatically create rules on views so they are updateable, per SQL99
479 We can only auto-create rules for simple views. For more complex
480 cases users will still have to write rules manually.
482 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
483 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
485 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
486 * Allow NOTIFY in rules involving conditionals
487 * Allow VIEW/RULE recompilation when the underlying tables change
489 Another issue is whether underlying table changes should be reflected
490 in the view, e.g. should SELECT * show additional columns if they
491 are added after the view is created.
493 * Make it possible to use RETURNING together with conditional DO INSTEAD
494 rules, such as for partitioning setups
496 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
498 * Add the ability to automatically create materialized views
500 Right now materialized views require the user to create triggers on the
501 main table to keep the summary table current. SQL syntax should be able
502 to manager the triggers and summary table automatically. A more
503 sophisticated implementation would automatically retrieve from the
504 summary table when the main table is referenced, if possible.
511 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
512 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
513 * %Allow SET CONSTRAINTS to be qualified by schema/table name
514 * %Add a separate TRUNCATE permission
516 Currently only the owner can TRUNCATE a table because triggers are not
517 called, and the table is locked in exclusive mode.
519 * Allow PREPARE of cursors
520 * Allow finer control over the caching of prepared query plans
522 Currently queries prepared via the libpq API are planned on first
523 execute using the supplied parameters --- allow SQL PREPARE to do the
524 same. Also, allow control over replanning prepared queries either
525 manually or automatically when statistics for execute parameters
526 differ dramatically from those used during planning.
528 * Improve logging of prepared transactions recovered during startup
530 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
532 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
534 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
535 such information in memory would improve performance.
537 * Add optional textual message to NOTIFY
539 This would allow an informational message to be added to the notify
540 message, perhaps indicating the row modified or other custom
543 * Allow multiple identical NOTIFY events to always be communicated to the
544 client, rather than sent as a single notification to the listener
546 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
548 * Add a GUC variable to warn about non-standard SQL usage in queries
549 * Add SQL-standard MERGE command, typically used to merge two tables
552 This is similar to UPDATE, then for unmatched rows, INSERT.
553 Whether concurrent access allows modifications which could cause
554 row loss is implementation independent.
556 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
559 To implement this cleanly requires that the table have a unique index
560 so duplicate checking can be easily performed. It is possible to
561 do it without a unique index if we require the user to LOCK the table
564 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
565 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
567 * Add NOVICE output level for helpful messages like automatic sequence/index
569 * Add GUC to issue notice about statements that use unjoined tables
570 * Allow EXPLAIN to identify tables that were skipped because of
572 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
573 * Enable standard_conforming_strings
574 * Make standard_conforming_strings the default in 8.5?
576 When this is done, backslash-quote should be prohibited in non-E''
577 strings because of possible confusion over how such strings treat
578 backslashes. Basically, '' is always safe for a literal single
579 quote, while \' might or might not be based on the backslash
582 * Simplify dropping roles that have objects in several databases
583 * Allow COMMENT ON to accept an expression rather than just a string
584 * Allow the count returned by SELECT, etc to be represented as an int64
585 to allow a higher range of values
586 * Add SQL99 WITH clause to SELECT
587 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
589 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
590 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
591 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
593 * Add DEFAULT .. AS OWNER so permission checks are done as the table
596 This would be useful for SERIAL nextval() calls and CHECK constraints.
598 * Allow DISTINCT to work in multiple-argument aggregate calls
599 * Add column to pg_stat_activity that shows the progress of long-running
600 commands like CREATE INDEX and VACUUM
601 * Implement SQL:2003 window functions
602 * Improve failure message when DROP DATABASE is used on a database that
603 has prepared transactions
604 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
606 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
607 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
609 * Increase locking when DROPing objects so dependent objects cannot
610 get dropped while the DROP operation is happening
612 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
614 * -Allow AS in "SELECT col AS label" to be optional in certain cases
616 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
617 RETURNING to supply values to the INSERT
618 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
620 * Add comments on system tables/columns using the information in
623 Ideally the information would be pulled from the SGML file
629 o Allow CREATE TABLE AS to determine column lengths for complex
630 expressions like SELECT col1 || col2
632 o Have WITH CONSTRAINTS also create constraint indexes
634 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
636 o Have CONSTRAINT cname NOT NULL record the contraint name
638 Right now pg_attribute.attnotnull records the NOT NULL status
639 of the column, but does not record the contraint name
644 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
646 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
647 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
649 o Research self-referential UPDATEs that see inconsistent row versions
650 in read-committed mode
652 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
653 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
655 o Allow GLOBAL temporary tables to exist as empty by default in
658 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
663 o Have ALTER TABLE RENAME rename SERIAL sequence names
665 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
667 o Have ALTER SEQUENCE RENAME rename the sequence name stored
668 in the sequence table
670 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
671 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
672 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
674 o Add ALTER DOMAIN to modify the underlying data type
675 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
677 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
679 o %Allow ALTER TABLE to change constraint deferrability and actions
680 o Add missing object types for ALTER ... SET SCHEMA
681 o Allow ALTER TABLESPACE to move to different directories
682 o Allow databases to be moved to different tablespaces
683 o Allow moving system tables to other tablespaces, where possible
685 Currently non-global system tables must be in the default database
686 tablespace. Global system tables can never be moved.
688 o Prevent parent tables from altering or dropping constraints
689 like CHECK that are inherited by child tables unless CASCADE
691 o %Prevent child tables from altering or dropping constraints
692 like CHECK that were inherited from the parent table
693 o Have ALTER INDEX update the name of a constraint using that index
694 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
695 o Allow column display reordering by recording a display,
696 storage, and permanent id for every column?
698 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
704 o Automatically maintain clustering on a table
706 This might require some background daemon to maintain clustering
707 during periods of low usage. It might also require tables to be only
708 partially filled for easier reorganization. Another idea would
709 be to create a merged heap/index data file so an index lookup would
710 automatically access the heap data too. A third idea would be to
711 store heap rows in hashed groups, perhaps using a user-supplied
713 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
715 o %Add default clustering to system tables
717 To do this, determine the ideal cluster index for each system
718 table and set the cluster setting during initdb.
720 o %Add VERBOSE option to report tables as they are processed,
726 o Allow COPY to report error lines and continue
728 This requires the use of a savepoint before each COPY line is
729 processed, with ROLLBACK on COPY failure.
731 o Allow COPY on a newly-created table to skip WAL logging
733 On crash recovery, the table involved in the COPY would
734 be removed or have its heap and index files truncated. One
735 issue is that no other backend should be able to add to
736 the table at the same time, which is something that is
737 currently allowed. This currently is done if the table is
738 created inside the same transaction block as the COPY because
739 no other backends can see the table.
741 o Consider using a ring buffer for COPY FROM
743 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
744 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
746 o Allow COPY FROM to create index entries in bulk
748 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
750 o Allow COPY in CSV mode to control whether a quoted zero-length
751 string is treated as NULL
753 Currently this is always treated as a zero-length string,
754 which generates an error when loading into an integer column
755 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00905.php
761 o Allow column-level privileges
762 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
765 The proposed syntax is:
766 GRANT SELECT ON ALL TABLES IN public TO phpuser;
767 GRANT SELECT ON NEW TABLES IN public TO phpuser;
769 o Allow GRANT/REVOKE permissions to be inherited by objects based on
772 o Allow SERIAL sequences to inherit permissions from the base table?
777 o Prevent DROP TABLE from dropping a row referenced by its own open
783 o Allow INSERT/UPDATE of the system-generated oid value for a row
784 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
790 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
795 Referential Integrity
796 =====================
798 * Add MATCH PARTIAL referential integrity
799 * Change foreign key constraint for array -> element to mean element
801 * Fix problem when cascading referential triggers make changes on
802 cascaded tables, seeing the tables in an intermediate state
804 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
805 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
807 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
809 This would allow UPDATE tab SET col = col + 1 to work if col has
810 a unique index. Currently, uniqueness checks are done while the
811 command is being executed, rather than at the end of the statement
813 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
814 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
816 * Optimize referential integrity checks
818 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
819 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
822 Server-Side Languages
823 =====================
826 o Fix RENAME to work on variables other than OLD/NEW
828 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
829 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
830 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
832 o Allow function parameters to be passed by name,
833 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
834 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
835 o Allow listing of record column names, and access to
836 record columns via variables, e.g. columns := r.(*),
839 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
840 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
841 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
843 o Add support for SCROLL cursors
844 o Add support for WITH HOLD cursors
845 o Allow row and record variables to be set to NULL constants,
846 and allow NULL tests on such variables
848 Because a row is not scalar, do not allow assignment
849 from NULL-valued scalars.
851 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
853 o Review handling of MOVE and FETCH
855 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
857 o Improve logic of determining if an identifier is a a
858 variable or column name
860 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00436.php
864 o Add table function support to pltcl, plpythonu
865 o Add support for polymorphic arguments and return types to
866 languages other than PL/PgSQL
867 o Add capability to create and call PROCEDURES
868 o Add support for OUT and INOUT parameters to languages other
870 o Add PL/PythonU tracebacks
872 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
874 o Allow data to be passed in native language formats, rather
877 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
884 * Have pg_ctl look at PGHOST in case it is a socket directory?
885 * Allow pg_ctl to work properly with configuration files located outside
888 pg_ctl can not read the pid file because it isn't located in the
889 config directory but in the PGDATA directory. The solution is to
890 allow pg_ctl to read and understand postgresql.conf to find the
891 data_directory value.
893 * Add a function like pg_get_indexdef() that report more detailed index
896 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
898 * Prefix command-line utilities like createuser with 'pg_'
900 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00025.php
904 o Have psql show current values for a sequence
905 o Move psql backslash database information into the backend, use
906 mnemonic commands? [psql]
908 This would allow non-psql clients to pull the same information out
909 of the database as psql.
911 o Fix psql's \d commands more consistent
913 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
914 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
916 o Consistently display privilege information for all objects in psql
917 o Add auto-expanded mode so expanded output is used if the row
918 length is wider than the screen width.
920 Consider using auto-expanded mode for backslash commands like \df+.
922 o Prevent tab completion of SET TRANSACTION from querying the
923 database and therefore preventing the transaction isolation
924 level from being set.
926 Currently SET <tab> causes a database lookup to check all
927 supported session variables. This query causes problems
928 because setting the transaction isolation level must be the
929 first statement of a transaction.
931 o Add a \set variable to control whether \s displays line numbers
933 Another option is to add \# which lists line numbers, and
934 allows command execution.
936 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
938 o Prevent escape string warnings when object names have
941 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
943 o Have \d show foreign keys that reference a table's primary key
945 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
947 o Have \d show child tables that inherit from the specified parent
948 o Have \l+ show database size, if permissions allow
950 Ideally it will not generate an error for invalid permissions
954 * pg_dump / pg_restore
955 o %Add dumping of comments on index columns and composite type columns
956 o %Add full object name to the tag field. eg. for operators we need
957 '=(integer, integer)', instead of just '='.
958 o Add pg_dumpall custom format dumps?
959 o Allow selection of individual object(s) of all types, not just
961 o In a selective dump, allow dumping of an object and all its
963 o Add options like pg_restore -l and -L to pg_dump
964 o Stop dumping CASCADE on DROP TYPE commands in clean mode
965 o Allow pg_dump --clean to drop roles that own objects or have
967 o Change pg_dump so that a comment on the dumped database is
968 applied to the loaded database, even if the database has a
969 different name. This will require new backend syntax, perhaps
970 COMMENT ON CURRENT DATABASE.
971 o Remove unnecessary function pointer abstractions in pg_dump source
973 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
974 multiple objects simultaneously
976 The difficulty with this is getting multiple dump processes to
977 produce a single dump output file. It also would require
978 several sessions to share the same snapshot.
979 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
981 o Allow pg_restore to utilize multiple CPUs and I/O channels by
982 restoring multiple objects simultaneously
984 This might require a pg_restore flag to indicate how many
985 simultaneous operations should be performed. Only pg_dump's
986 -Fc format has the necessary dependency information.
987 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
989 o To better utilize resources, allow pg_restore to check foreign
990 keys simultaneously, where possible
991 o Allow pg_restore to create all indexes of a table
992 concurrently, via a single heap scan
994 This requires a pg_dump -Fc file because that format contains
995 the required dependency information.
996 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
998 o Allow pg_restore to load different parts of the COPY data
1000 o Prevent pg_dump/pg_restore from being affected by
1003 Using psql to restore a pg_dump dump is also affected.
1009 Document differences between ecpg and the SQL standard and
1010 information about the Informix-compatibility module.
1012 o Solve cardinality > 1 for input descriptors / variables?
1013 o Add a semantic check level, e.g. check if a table really exists
1014 o fix handling of DB attributes that are arrays
1015 o Use backend PREPARE/EXECUTE facility for ecpg where possible
1017 o Fix nested C comments
1018 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
1019 o Make SET CONNECTION thread-aware, non-standard?
1020 o Allow multidimensional arrays
1021 o Add internationalized message strings
1022 o Implement COPY FROM STDIN
1026 o Add PQescapeIdentifierConn()
1027 o Prevent PQfnumber() from lowercasing unquoted the column name
1029 PQfnumber() should never have been doing lowercasing, but
1030 historically it has so we need a way to prevent it
1032 o Allow statement results to be automatically batched to the client
1034 Currently all statement results are transferred to the libpq
1035 client before libpq makes the results available to the
1036 application. This feature would allow the application to make
1037 use of the first result rows while the rest are transferred, or
1038 held on the server waiting for them to be requested by libpq.
1039 One complexity is that a statement like SELECT 1/col could error
1040 out mid-way through the result set.
1042 o Consider disallowing multiple queries in PQexec() as an
1043 additional barrier to SQL injection attacks
1045 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1047 o Add PQexecf() that allows complex parameter substitution
1049 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1051 o Add SQLSTATE severity to PGconn return status
1053 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1059 * Add deferred trigger queue file
1061 Right now all deferred trigger information is stored in backend
1062 memory. This could exhaust memory for very large trigger queues.
1063 This item involves dumping large queues into files.
1065 * Allow triggers to be disabled in only the current session.
1067 This is currently possible by starting a multi-statement transaction,
1068 modifying the system tables, performing the desired SQL, restoring the
1069 system tables, and committing the transaction. ALTER TABLE ...
1070 TRIGGER requires a table lock so it is not ideal for this usage.
1072 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1074 If the dump is known to be valid, allow foreign keys to be added
1075 without revalidating the data.
1077 * Allow statement-level triggers to access modified rows
1078 * Support triggers on columns
1080 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1082 * Allow AFTER triggers on system tables
1084 System tables are modified in many places in the backend without going
1085 through the executor and therefore not causing triggers to fire. To
1086 complete this item, the functions that modify system tables will have
1089 * Tighten trigger permission checks
1091 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1093 * Allow BEFORE INSERT triggers on views
1095 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1097 * Add ability to trigger on TRUNCATE
1099 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
1106 * Add UNIQUE capability to non-btree indexes
1107 * Prevent index uniqueness checks when UPDATE does not modify the column
1109 Uniqueness (index) checks are done when updating a column even if the
1110 column is not modified by the UPDATE.
1112 * Allow the creation of on-disk bitmap indexes which can be quickly
1113 combined with other bitmap indexes
1115 Such indexes could be more compact if there are only a few distinct values.
1116 Such indexes can also be compressed. Keeping such indexes updated can be
1119 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1120 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1122 * Allow accurate statistics to be collected on indexes with more than
1123 one column or expression indexes, perhaps using per-index statistics
1125 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1126 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1128 * Consider increasing the number of default statistics target, and
1129 reduce statistics target overhead
1131 Also consider having a larger statistics target for indexed columns
1132 and expression indexes
1133 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1134 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1136 * Consider compressing indexes by storing key values duplicated in
1137 several rows as a single index entry
1139 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1140 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1141 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1142 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1144 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1146 This is difficult because you must upgrade to an exclusive table lock
1147 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1148 have this complication. This would allow index compaction without
1150 http://archives.postgresql.org/pgsql-performance/2007-08/msg00289.php
1152 * Allow multiple indexes to be created concurrently, ideally via a
1153 single heap scan, and have pg_restore use it
1159 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1160 and primary/foreign keys
1161 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1162 on inherited table, e.g. INSERT INTO inherit_table
1163 (unique_index_col) VALUES (dup) should fail
1165 The main difficulty with this item is the problem of
1166 creating an index that can span multiple tables.
1168 o Allow SELECT ... FOR UPDATE on inherited tables
1169 o Require all CHECK constraints to be inherited
1171 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1176 o Add more GIST index support for geometric data types
1177 o Allow GIST indexes to create certain complex index types, like
1178 digital trees (see Aoki)
1183 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1185 o Pack hash index buckets onto disk pages more efficiently
1187 Currently only one hash bucket can be stored on a page. Ideally
1188 several hash buckets could be stored on a single page and greater
1189 granularity used for the hash algorithm.
1191 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1193 o Consider sorting hash buckets so entries can be found using a
1194 binary search, rather than a linear scan
1195 o In hash indexes, consider storing the hash value with or instead
1197 o Add WAL logging for crash recovery
1198 o Allow multi-column hash indexes
1199 o During index creation, pre-sort the tuples to improve build speed
1201 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
1208 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1210 Ideally this requires a separate test program that can be run
1211 at initdb time or optionally later. Consider O_SYNC when
1214 * Add program to test if fsync has a delay compared to non-fsync
1215 * Consider sorting writes during checkpoint
1217 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1226 We could use a fixed row count and a +/- count to follow MVCC
1227 visibility rules, or a single cached value could be used and
1228 invalidated if anyone modifies the table. Another idea is to
1229 get a count directly from a unique index, but for this to be
1230 faster than a sequential scan it must avoid access to the heap
1231 to obtain tuple visibility information.
1233 * Provide a way to calculate an "estimated COUNT(*)"
1235 Perhaps by using the optimizer's cardinality estimates or random
1238 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1240 * Allow data to be pulled directly from indexes
1242 Currently indexes do not have enough tuple visibility information
1243 to allow data to be pulled from the index without also accessing
1244 the heap. One way to allow this is to set a bit on index tuples
1245 to indicate if a tuple is currently visible to all transactions
1246 when the first valid heap lookup happens. This bit would have to
1247 be cleared when a heap tuple is expired.
1249 Another idea is to maintain a bitmap of heap pages where all rows
1250 are visible to all backends, and allow index lookups to reference
1251 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1252 add someday to determine which heap pages need vacuuming. Frequently
1253 accessed bitmaps would have to be stored in shared memory. One 8k
1254 page of bitmaps could track 512MB of heap pages.
1256 A third idea would be for a heap scan to check if all rows are visible
1257 and if so set a per-table flag which can be checked by index scans.
1258 Any change to the table would have to clear the flag. To detect
1259 changes during the heap scan a counter could be set at the start and
1260 checked at the end --- if it is the same, the table has not been
1261 modified --- any table change would increment the counter.
1263 * Consider automatic caching of statements at various levels:
1266 o Query execute plan
1269 * Consider increasing internal areas when shared buffers is increased
1271 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1273 * Consider decreasing the amount of memory used by PrivateRefCount
1275 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1276 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1278 * Consider allowing higher priority queries to have referenced buffer
1279 cache pages stay in memory longer
1281 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1288 * Improve speed with indexes
1290 For large table adjustments during VACUUM FULL, it is faster to cluster
1291 or reindex rather than update the index. Also, index updates can bloat
1294 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1295 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1296 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00307.php
1298 * Auto-fill the free space map by scanning the buffer cache or by
1299 checking pages written by the background writer
1301 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1302 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1304 * Create a bitmap of pages that need vacuuming
1306 Instead of sequentially scanning the entire table, have the background
1307 writer or some other process record pages that have expired rows, then
1308 VACUUM can look at just those pages rather than the entire table. In
1309 the event of a system crash, the bitmap would probably be invalidated.
1310 One complexity is that index entries still have to be vacuumed, and
1311 doing this without an index scan (by using the heap values to find the
1312 index entry) might be slow and unreliable, especially for user-defined
1315 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1316 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1318 * Allow FSM to return free space toward the beginning of the heap file,
1319 in hopes that empty pages at the end can be truncated by VACUUM
1320 * Allow FSM page return free space based on table clustering, to assist
1321 in maintaining clustering?
1322 * Improve dead row detection during multi-statement transactions usage
1324 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1326 * Consider a more compact data representation for dead tuples
1328 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1333 o %Issue log message to suggest VACUUM FULL if a table is nearly
1335 o Improve control of auto-vacuum
1337 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1339 o Prevent long-lived temporary tables from causing frozen-xid
1340 advancement starvation
1342 The problem is that autovacuum cannot vacuum them to set frozen xids;
1343 only the session that created them can do that.
1344 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1346 o Store per-table autovacuum settings in pg_class.reloptions.
1348 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1349 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1351 o Prevent autovacuum from running if an old transaction is still
1352 running from the last vacuum
1354 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1361 * Fix priority ordering of read and write light-weight locks (Neil)
1363 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1364 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1366 * Fix problem when multiple subtransactions of the same outer transaction
1367 hold different types of locks, and one subtransaction aborts
1369 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1370 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1371 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1372 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1374 * Allow UPDATEs on only non-referential integrity columns not to conflict
1375 with referential integrity locks
1377 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1379 * Add idle_in_transaction_timeout GUC so locks are not held for long
1382 * Improve deadlock detection when a page cleaning lock conflicts
1383 with a shared buffer that is pinned
1385 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1386 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1387 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1392 Startup Time Improvements
1393 =========================
1395 * Experiment with multi-threaded backend for backend creation [thread]
1397 This would prevent the overhead associated with process creation. Most
1398 operating systems have trivial process creation time compared to
1399 database startup overhead, but a few operating systems (Win32,
1400 Solaris) might benefit from threading. Also explore the idea of
1401 a single session using multiple threads to execute a statement faster.
1408 * Eliminate need to write full pages to WAL before page modification [wal]
1410 Currently, to protect against partial disk page writes, we write
1411 full page images to WAL before they are modified so we can correct any
1412 partial page writes during recovery. These pages can also be
1413 eliminated from point-in-time archive files.
1415 o When off, write CRC to WAL and check file system blocks
1418 If CRC check fails during recovery, remember the page in case
1419 a later CRC for that page properly matches.
1421 o Write full pages during file system write and not when
1422 the page is modified in the buffer cache
1424 This allows most full page writes to happen in the background
1425 writer. It might cause problems for applying WAL on recovery
1426 into a partially-written page, but later the full page will be
1429 * Allow WAL traffic to be streamed to another server for stand-by
1431 * Reduce WAL traffic so only modified values are written rather than
1434 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1436 * Allow WAL information to recover corrupted pg_controldata
1438 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1440 * Find a way to reduce rotational delay when repeatedly writing
1443 Currently fsync of WAL requires the disk platter to perform a full
1444 rotation to fsync again. One idea is to write the WAL to different
1445 offsets that might reduce the rotational delay.
1447 * Allow WAL logging to be turned off for a table, but the table
1448 might be dropped or truncated during crash recovery [walcontrol]
1450 Allow tables to bypass WAL writes and just fsync() dirty pages on
1451 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1452 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1453 non-default logging should not use referential integrity with
1454 default-logging tables. A table without dirty buffers during a
1455 crash could perhaps avoid the drop/truncate.
1457 * Allow WAL logging to be turned off for a table, but the table would
1458 avoid being truncated/dropped [walcontrol]
1460 To do this, only a single writer can modify the table, and writes
1461 must happen only on new pages so the new pages can be removed during
1462 crash recovery. Readers can continue accessing the table. Such
1463 tables probably cannot have indexes. One complexity is the handling
1464 of indexes on TOAST tables.
1466 * Speed WAL recovery by allowing more than one page to be prefetched
1468 This involves having a separate process that can be told which pages
1469 the recovery process will need in the near future.
1470 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1474 Optimizer / Executor
1475 ====================
1477 * Improve selectivity functions for geometric operators
1478 * Precompile SQL functions to avoid overhead
1479 * Create utility to compute accurate random_page_cost value
1480 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1481 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1482 actual row counts differ by a specified percentage
1483 * Consider using hash buckets to do DISTINCT, rather than sorting
1485 This would be beneficial when there are few distinct values. This is
1486 already used by GROUP BY.
1488 * Log statements where the optimizer row estimates were dramatically
1489 different from the number of rows actually found?
1490 * Consider compressed annealing to search for query plans
1492 This might replace GEQO, http://sixdemonbag.org/Djinni.
1494 * Improve merge join performance by allowing mark/restore of
1497 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1499 * Consider using a hash for joining to a large IN (VALUES ...) list
1501 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1508 * Consider having the background writer update the transaction status
1509 hint bits before writing out the page
1511 Implementing this requires the background writer to have access to system
1512 catalogs and the transaction status log.
1514 * Consider adding buffers the background writer finds reusable to the
1517 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1519 * Automatically tune bgwriter_delay based on activity rather then using a
1522 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1524 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1526 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1528 * Test to see if calling PreallocXlogFiles() from the background writer
1529 will help with WAL segment creation latency
1531 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1535 Miscellaneous Performance
1536 =========================
1538 * Do async I/O for faster random read-ahead of data
1540 Async I/O allows multiple I/O requests to be sent to the disk with
1541 results coming back asynchronously.
1543 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1544 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1546 * Use mmap() rather than SYSV shared memory or to write WAL files?
1548 This would remove the requirement for SYSV SHM but would introduce
1549 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1550 to prevent I/O overhead.
1552 * Consider mmap()'ing files into a backend?
1554 Doing I/O to large tables would consume a lot of address space or
1555 require frequent mapping/unmapping. Extending the file also causes
1556 mapping problems that might require mapping only individual pages,
1557 leading to thousands of mappings. Another problem is that there is no
1558 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1559 could hit disk before WAL is written.
1561 * Add a script to ask system configuration questions and tune postgresql.conf
1562 * Consider ways of storing rows more compactly on disk
1564 o Reduce the row header size?
1565 o Consider reducing on-disk varlena length from four bytes to
1566 two because a heap row cannot be more than 64k in length
1568 * Consider increasing NUM_CLOG_BUFFERS
1570 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00030.php
1571 http://archives.postgresql.org/pgsql-performance/2007-08/msg00024.php
1573 * Allow user configuration of TOAST thresholds
1575 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1576 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00082.php
1578 * Allow configuration of backend priorities via the operating system
1580 Though backend priorities make priority inversion during lock
1581 waits possible, research shows that this is not a huge problem.
1583 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1585 * Experiment with multi-threaded backend better resource utilization
1587 This would allow a single query to make use of multiple CPU's or
1588 multiple I/O channels simultaneously. One idea is to create a
1589 background reader that can pre-fetch sequential and index scan
1590 pages needed by other backends. This could be expanded to allow
1591 concurrent reads from multiple devices in a partitioned table.
1593 * Consider increasing the minimum allowed number of shared buffers
1595 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1603 * Add use of 'const' for variables in source tree
1604 * Move some things from contrib into main tree
1605 * %Remove warnings created by -Wcast-align
1606 * Move platform-specific ps status display info from ps_status.c to ports
1607 * Add optional CRC checksum to heap and index pages
1608 * Improve documentation to build only interfaces (Marc)
1609 * Remove or relicense modules that are not under the BSD license, if possible
1610 * Acquire lock on a relation before building a relcache entry for it
1611 * Allow cross-compiling by generating the zic database on the target system
1612 * Improve NLS maintenance of libpgport messages linked onto applications
1613 * Clean up casting in contrib/isn
1615 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1617 * Use UTF8 encoding for NLS messages so all server encodings can
1619 * Update Bonjour to work with newer cross-platform SDK
1621 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1622 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1624 * Consider detoasting keys before sorting
1625 * Consider GnuTLS if OpenSSL license becomes a problem
1627 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1628 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1630 * Consider changing documentation format from SGML to XML
1632 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1634 * Consider making NAMEDATALEN more configurable in future releases
1635 * Update our code to handle 64-bit timezone files to match the zic
1636 source code, which now uses them
1637 * Have configure choose integer datetimes by default
1639 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1641 * Support scoped IPv6 addresses
1643 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1645 * Consider allowing 64-bit integers and floats to be passed by value on
1648 Also change 32-bit floats (float4) to be passed by value at the same
1651 * Research use of signals and sleep wake ups
1653 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1655 * Add automated check for invalid C++ source code constructs
1657 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1659 * Consider simplifying how memory context resets handle child contexts
1661 http://archives.postgresql.org/pgsql-patches/2007-08/msg00067.php
1663 * Remove use of MAKE_PTR and MAKE_OFFSET macros
1665 http://archives.postgresql.org/pgsql-general/2007-08/msg01510.php
1670 o Remove configure.in check for link failure when cause is found
1671 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1673 o Remove psql newline patch when we find out why mingw outputs an
1675 o Allow psql to use readline once non-US code pages work with
1677 o Fix problem with shared memory on the Win32 Terminal Server
1678 o Diagnose problem where shared memory can sometimes not be
1679 attached by postmaster children
1681 http://archives.postgresql.org/pgsql-general/2007-08/msg01377.php
1683 o Improve signal handling
1685 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1687 o Convert MSVC build system to remove most batch files
1689 http://archives.postgresql.org/pgsql-hackers/2007-08/msg00961.php
1693 * Wire Protocol Changes
1695 o Allow dynamic character set handling
1696 o Add decoded type, length, precision
1698 o Update clients to use data types, typmod, schema.table.column names
1699 of result sets using new statement protocol
1705 * Add pre-parsing phase that converts non-ISO syntax to supported
1708 This could allow SQL written for other databases to run without
1711 * Allow plug-in modules to emulate features from other databases
1712 * Add features of Oracle-style packages (Pavel)
1714 A package would be a schema with session-local variables,
1715 public/private functions, and initialization functions. It
1716 is also possible to implement these capabilities
1717 in any schema and not use a separate "packages"
1720 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1722 * Consider allowing control of upper/lower case folding of unquoted
1725 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1726 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1728 * Add autonomous transactions
1730 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1734 Features We Do _Not_ Want
1735 =========================
1737 * All backends running as threads in a single process (not wanted)
1739 This eliminates the process protection we get from the current setup.
1740 Thread creation is usually the same overhead as process creation on
1741 modern systems, so it seems unwise to use a pure threaded model.
1743 * Optimizer hints (not wanted)
1745 Optimizer hints are used to work around problems in the optimizer. We
1746 would rather have the problems reported and fixed.
1748 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1749 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1750 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1753 Because we support postfix operators, it isn't possible to make AS
1754 optional and continue to use bison.
1755 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1757 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1759 * Embedded server (not wanted)
1761 While PostgreSQL clients runs fine in limited-resource environments, the
1762 server requires multiple processes and a stable pool of resources to
1763 run reliabily and efficiently. Stripping down the PostgreSQL server
1764 to run in the same process address space as the client application
1765 would add too much complexity and failure cases.