4 Current maintainer: Bruce Momjian (pgman@candle.pha.pa.us)
5 Last updated: Sat Aug 20 21:04:41 EDT 2005
7 The most recent version of this document can be viewed at
8 http://www.postgresql.org/docs/faqs.TODO.html.
10 #A hyphen, "-", marks changes that will appear in the upcoming 8.1 release.#
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
22 * Remove behavior of postmaster -o after making postmaster/postgres
24 * -Allow limits on per-db/role connections
25 * Allow pooled connections to list all prepared queries
27 This would allow an application inheriting a pooled connection to know
28 the queries prepared in the current session.
30 * Allow major upgrades without dump/reload, perhaps using pg_upgrade
32 * Check for unreferenced table files created by transactions that were
33 in-progress when the server terminated abruptly
34 * Allow administrators to safely terminate individual sessions either
35 via an SQL function or SIGTERM
37 Currently SIGTERM of a backend can lead to lock table corruption.
39 * -Prevent dropping user that still owns objects, or auto-drop the objects
40 * Set proper permissions on non-system schemas during db creation
42 Currently all schemas are owned by the super-user because they are
43 copied from the template1 database.
45 * -Add the client IP address and port to pg_stat_activity
46 * Support table partitioning that allows a single table to be stored
47 in subtables that are partitioned based on the primary key or a WHERE
51 * Improve replication solutions
55 You can use any of the master/slave replication servers to use a
56 standby server for data warehousing. To allow read/write queries to
57 multiple servers, you need multi-master replication like pgcluster.
59 o Allow replication over unreliable or non-persistent links
64 o Add "include file" functionality in postgresql.conf
65 o Allow postgresql.conf values to be set so they can not be changed
67 o Allow commenting of variables in postgresql.conf to restore them
69 o Allow pg_hba.conf settings to be controlled via SQL
71 This would add a function to load the SQL table from
72 pg_hba.conf, and one to writes its contents to the flat file.
73 The table should have a line number that is a float so rows
74 can be inserted between existing rows, e.g. row 2.5 goes
75 between row 2 and row 3.
77 o Allow postgresql.conf file values to be changed via an SQL
78 API, perhaps using SET GLOBAL
79 o Allow the server to be stopped/restarted via an SQL API
84 * Allow a database in tablespace t1 with tables created in
85 tablespace t2 to be used as a template for a new database created
86 with default tablespace t2
88 All objects in the default database tablespace must have default
89 tablespace specifications. This is because new databases are
90 created by copying directories. If you mix default tablespace
91 tables and tablespace-specified tables in the same directory,
92 creating a new database from such a mixed directory would create a
93 new database with tables that had incorrect explicit tablespaces.
94 To fix this would require modifying pg_class in the newly copied
95 database, which we don't currently do.
97 * Allow reporting of which objects are in which tablespaces
99 This item is difficult because a tablespace can contain objects
100 from multiple databases. There is a server-side function that
101 returns the databases which use a specific tablespace, so this
102 requires a tool that will call that function and connect to each
103 database to find the objects in each database for that tablespace.
105 o Add a GUC variable to control the tablespace for temporary objects
108 It could start with a random tablespace from a supplied list and
109 cycle through the list.
111 o Allow WAL replay of CREATE TABLESPACE to work when the directory
112 structure on the recovery computer is different from the original
114 o Allow per-tablespace quotas
117 * Point-In-Time Recovery (PITR)
119 o Allow point-in-time recovery to archive partially filled
120 write-ahead logs [pitr]
122 Currently only full WAL files are archived. This means that the
123 most recent transactions aren't available for recovery in case
124 of a disk failure. This could be triggered by a user command or
127 o Automatically force archiving of partially-filled WAL files when
128 pg_stop_backup() is called or the server is stopped
130 Doing this will allow administrators to know more easily when
131 the archive contins all the files needed for point-in-time
134 o Create dump tool for write-ahead logs for use in determining
135 transaction id for point-in-time recovery
136 o Allow a warm standby system to also allow read-only queries
139 This is useful for checking PITR recovery.
141 o Allow the PITR process to be debugged and data examined
147 * Allow server log information to be output as INSERT statements
149 This would allow server log information to be easily loaded into
150 a database for analysis.
152 * Add ability to monitor the use of temporary sort files
153 * -Add session start time and last statement time to pg_stat_activity
154 * -Add a function that returns the start time of the postmaster
155 * Allow server logs to be remotely read and removed using SQL commands
161 * Remove Money type, add money formatting for decimal type
162 * Change NUMERIC to enforce the maximum precision, and increase it
163 * Add NUMERIC division operator that doesn't round?
165 Currently NUMERIC _rounds_ the result to the specified precision.
166 This means division can return a result that multiplied by the
167 divisor is greater than the dividend, e.g. this returns a value > 10:
169 SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
171 The positive modulus result returned by NUMERICs might be considered
172 inaccurate, in one sense.
174 * Have sequence dependency track use of DEFAULT sequences,
176 * Disallow changing default expression of a SERIAL column?
177 * Fix data types where equality comparison isn't intuitive, e.g. box
178 * Prevent INET cast to CIDR if the unmasked bits are not zero, or
180 * Prevent INET cast to CIDR from droping netmask, SELECT '1.1.1.1'::inet::cidr
181 * Allow INET + INT4 to increment the host part of the address, or
182 throw an error on overflow
183 * Add 'tid != tid ' operator for use in corruption recovery
188 o Allow infinite dates just like infinite timestamps
189 o Add a GUC variable to allow output of interval values in ISO8601
191 o Merge hardwired timezone names with the TZ database; allow either
192 kind everywhere a TZ name is currently taken
193 o Allow customization of the known set of TZ names (generalize the
194 present australian_timezones hack)
195 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
196 information, either zone name or offset from UTC [timezone]
198 If the TIMESTAMP value is stored with a time zone name, interval
199 computations should adjust based on the time zone rules.
201 o Add ISO INTERVAL handling
202 o Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
204 o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
205 o For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
206 '1', treat as ISO if there is a range specification clause,
207 and as PG if there no clause is present, e.g. interpret
208 '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
209 interpret '1:30' as '1 hour, 30 minutes'
210 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
211 INTERVAL MONTH), and this should return '12 months'
212 o Round or truncate values to the requested precision, e.g.
213 INTERVAL '11 months' AS YEAR should return one or zero
214 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
219 o Allow NULLs in arrays
220 o Allow MIN()/MAX() on arrays
221 o Delay resolution of array expression's data type so assignment
222 coercion can be performed on empty array expressions
223 o Modify array literal representation to handle array index lower bound
229 o Improve vacuum of large objects, like /contrib/vacuumlo?
230 o Add security checking for large objects
232 Currently large objects entries do not have owners. Permissions can
233 only be set at the pg_largeobject table level.
235 o Auto-delete large objects when referencing row is deleted
236 o Allow read/write into TOAST values like large objects
238 This requires the TOAST column to be stored EXTERNAL.
244 * -Add function to return compressed length of TOAST data values
245 * Allow INET subnet tests using non-constants to be indexed
246 * Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
249 Current CURRENT_TIMESTAMP returns the start time of the current
250 transaction, and gettimeofday() returns the wallclock time. This will
251 make time reporting more consistent and will allow reporting of
252 the statement start time.
254 * Add pg_get_acldef(), pg_get_typedefault(), and pg_get_attrdef()
255 * Allow to_char() to print localized month names
256 * Allow functions to have a schema search path specified at creation time
257 * Allow substring/replace() to get/set bit values
258 * Allow to_char() on interval values to accumulate the highest unit
261 Some special format flag would be required to request such
262 accumulation. Such functionality could also be added to EXTRACT.
263 Prevent accumulation that crosses the month/day boundary because of
264 the uneven number of days in a month.
266 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
267 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
268 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
269 o to_char(INTERVAL '3 years 5 months','MM') => 41
271 * -Prevent to_char() on interval from returning meaningless values
273 For example, to_char('1 month', 'mon') is meaningless. Basically,
274 most date-related parameters to to_char() are meaningless for
275 intervals because interval is not anchored to a date.
278 Multi-Language Support
279 ======================
281 * Add NCHAR (as distinguished from ordinary varchar),
282 * Allow locale to be set at database creation
284 Currently locale can only be set during initdb. No global tables have
285 locale-aware columns. However, the database template used during
286 database creation might have locale-aware indexes. The indexes would
287 need to be reindexed to match the new locale.
289 * Allow encoding on a per-column basis
291 Right now only one encoding is allowed per database.
293 * Support multiple simultaneous character sets, per SQL92
294 * Improve UTF8 combined character handling?
295 * Add octet_length_server() and octet_length_client()
296 * Make octet_length_client() the same as octet_length()?
297 * Fix problems with wrong runtime encoding conversion for NLS message files
303 * Automatically create rules on views so they are updateable, per SQL99
305 We can only auto-create rules for simple views. For more complex
306 cases users will still have to write rules.
308 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
309 * Allow NOTIFY in rules involving conditionals
310 * Have views on temporary tables exist in the temporary namespace
311 * Allow temporary views on non-temporary tables
312 * Allow RULE recompilation
318 * -Add BETWEEN SYMMETRIC/ASYMMETRIC
319 * Change LIMIT/OFFSET and FETCH/MOVE to use int8
320 * -Add E'' escape string marker so eventually ordinary strings can treat
321 backslashes literally, for portability
323 * -Allow additional tables to be specified in DELETE for joins
325 UPDATE already allows this (UPDATE...FROM) but we need similar
326 functionality in DELETE. It's been agreed that the keyword should
327 be USING, to avoid anything as confusing as DELETE FROM a FROM b.
329 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
330 * -Allow REINDEX to rebuild all database indexes
331 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
332 * Allow SET CONSTRAINTS to be qualified by schema/table name
333 * Allow TRUNCATE ... CASCADE/RESTRICT
334 * Add a separate TRUNCATE permission
336 Currently only the owner can TRUNCATE a table because triggers are not
337 called, and the table is locked in exclusive mode.
339 * Allow PREPARE of cursors
340 * Allow PREPARE to automatically determine parameter types based on the SQL
342 * Allow finer control over the caching of prepared query plans
344 Currently, queries prepared via the libpq API are planned on first
345 execute using the supplied parameters --- allow SQL PREPARE to do the
346 same. Also, allow control over replanning prepared queries either
347 manually or automatically when statistics for execute parameters
348 differ dramatically from those used during planning.
350 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
352 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
353 such information in memory would improve performance.
355 * Add optional textual message to NOTIFY
357 This would allow an informational message to be added to the notify
358 message, perhaps indicating the row modified or other custom
361 * Add a GUC variable to warn about non-standard SQL usage in queries
362 * Add MERGE command that does UPDATE/DELETE, or on failure, INSERT (rules,
364 * Add NOVICE output level for helpful messages like automatic sequence/index
366 * Add COMMENT ON for all cluster global objects (roles, databases
368 * -Add an option to automatically use savepoints for each statement in a
369 multi-statement transaction.
371 When enabled, this would allow errors in multi-statement transactions
372 to be automatically ignored.
374 * Make row-wise comparisons work per SQL spec
375 * Add RESET CONNECTION command to reset all session state
377 This would include resetting of all variables (RESET ALL), dropping of
378 temporary tables, removing any NOTIFYs, cursors, open transactions,
379 prepared queries, currval()s, etc. This could be used for connection
380 pooling. We could also change RESET ALL to have this functionality.
381 The difficult of this features is allowing RESET ALL to not affect
382 changes made by the interface driver for its internal use. One idea
383 is for this to be a protocol-only feature. Another approach is to
384 notify the protocol when a RESET CONNECTION command is used.
386 * Add GUC to issue notice about queries that use unjoined tables
387 * Allow EXPLAIN to identify tables that were skipped because of
389 * Allow EXPLAIN output to be more easily processed by scripts
394 o Allow CREATE TABLE AS to determine column lengths for complex
395 expressions like SELECT col1 || col2
397 o Use more reliable method for CREATE DATABASE to get a consistent
400 o Currently the system uses the operating system COPY command to
401 create a new database. Add ON COMMIT capability to CREATE TABLE AS
406 o Allow UPDATE to handle complex aggregates [update]?
407 o Allow an alias to be provided for the target table in
410 This is not SQL-spec but many DBMSs allow it.
412 o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
414 o -Allow FOR UPDATE queries to do NOWAIT locks
419 o Have ALTER TABLE RENAME rename SERIAL sequence names
420 o Add ALTER DOMAIN TYPE
421 o Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
422 o Allow ALTER TABLE to change constraint deferrability and actions
423 o Disallow dropping of an inherited constraint
424 o -Allow objects to be moved to different schemas
425 o Allow ALTER TABLESPACE to move to different directories
426 o Allow databases to be moved to different tablespaces
427 o Allow moving system tables to other tablespaces, where possible
429 Currently non-global system tables must be in the default database
430 tablespace. Global system tables can never be moved.
432 o Prevent child tables from altering constraints like CHECK that were
433 inherited from the parent table
438 o Automatically maintain clustering on a table
440 This might require some background daemon to maintain clustering
441 during periods of low usage. It might also require tables to be only
442 paritally filled for easier reorganization. Another idea would
443 be to create a merged heap/index data file so an index lookup would
444 automatically access the heap data too. A third idea would be to
445 store heap rows in hashed groups, perhaps using a user-supplied
448 o Add default clustering to system tables
450 To do this, determine the ideal cluster index for each system
451 table and set the cluster setting during initdb.
456 o Allow COPY to report error lines and continue
458 This requires the use of a savepoint before each COPY line is
459 processed, with ROLLBACK on COPY failure.
461 o -Allow COPY to understand \x as a hex byte
462 o Have COPY return the number of rows loaded/unloaded?
463 o -Allow COPY to optionally include column headings in the first line
464 o -Allow COPY FROM ... CSV to interpret newlines and carriage
470 o Allow column-level privileges
471 o Allow GRANT/REVOKE permissions to be applied to all schema objects
474 The proposed syntax is:
475 GRANT SELECT ON ALL TABLES IN public TO phpuser;
476 GRANT SELECT ON NEW TABLES IN public TO phpuser;
478 * Allow GRANT/REVOKE permissions to be inherited by objects based on
484 o Allow UPDATE/DELETE WHERE CURRENT OF cursor
486 This requires using the row ctid to map cursor rows back to the
487 original heap row. This become more complicated if WITH HOLD cursors
488 are to be supported because WITH HOLD cursors have a copy of the row
489 and no FOR UPDATE lock.
491 o Prevent DROP TABLE from dropping a row referenced by its own open
494 o Allow pooled connections to list all open WITH HOLD cursors
496 Because WITH HOLD cursors exist outside transactions, this allows
497 them to be listed so they can be closed.
502 o Allow INSERT/UPDATE of the system-generated oid value for a row
503 o Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
504 o Allow INSERT/UPDATE ... RETURNING new.col or old.col
506 This is useful for returning the auto-generated key for an INSERT.
507 One complication is how to handle rules that run as part of
513 o -Have SHOW ALL show descriptions for server-side variables
514 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
516 o Add SET PATH for schemas?
518 This is basically the same as SET search_path.
521 * Server-Side Languages
523 o -Allow PL/PgSQL's RAISE function to take expressions
525 Currently only constants are supported.
527 o -Change PL/PgSQL to use palloc() instead of malloc()
528 o Handle references to temporary tables that are created, destroyed,
529 then recreated during a session, and EXECUTE is not used
531 This requires the cached PL/PgSQL byte code to be invalidated when
532 an object referenced in the function is changed.
534 o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
535 o Allow function parameters to be passed by name,
536 get_employee_salary(emp_id => 12345, tax_year => 2001)
537 o Add Oracle-style packages
538 o Add table function support to pltcl, plperl, plpython?
539 o Allow PL/pgSQL to name columns by ordinal position, e.g. rec.(3)
540 o -Allow PL/pgSQL EXECUTE query_var INTO record_var;
541 o Add capability to create and call PROCEDURES
542 o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
543 o Add MOVE to PL/pgSQL
544 o Pass arrays natively instead of as text between plperl and postgres
545 o Add support for polymorphic arguments and return types to plperl
551 * Add a libpq function to support Parse/DescribeStatement capability
552 * Prevent libpq's PQfnumber() from lowercasing the column name?
553 * Allow libpq to access SQLSTATE so pg_ctl can test for connection failure
555 This would be used for checking if the server is up.
557 * Add PQescapeIdentifier() to libpq
558 * Have initdb set DateStyle based on locale?
559 * Have pg_ctl look at PGHOST in case it is a socket directory?
560 * Add a schema option to createlang
561 * Allow pg_ctl to work properly with configuration files located outside
564 pg_ctl can not read the pid file because it isn't located in the
565 config directory but in the PGDATA directory. The solution is to
566 allow pg_ctl to read and understand postgresql.conf to find the
567 data_directory value.
572 o Have psql show current values for a sequence
573 o Move psql backslash database information into the backend, use
574 mnemonic commands? [psql]
576 This would allow non-psql clients to pull the same information out
577 of the database as psql.
579 o Fix psql's display of schema information (Neil)
580 o Allow psql \pset boolean variables to set to fixed values, rather
582 o Consistently display privilege information for all objects in psql
583 o Improve psql's handling of multi-line queries
588 o Have pg_dump use multi-statement transactions for INSERT dumps
589 o Allow pg_dump to use multiple -t and -n switches [pg_dump]
590 o Add dumping of comments on composite type columns
591 o Add dumping of comments on index columns
592 o Replace crude DELETE FROM method of pg_dumpall --clean for
593 cleaning of roles with separate DROP commands
594 o -Add dumping and restoring of LOB comments
595 o Stop dumping CASCADE on DROP TYPE commands in clean mode
596 o Add full object name to the tag field. eg. for operators we need
597 '=(integer, integer)', instead of just '='.
598 o Add pg_dumpall custom format dumps.
600 This is probably best done by combining pg_dump and pg_dumpall
601 into a single binary.
603 o Add CSV output format
604 o Update pg_dump and psql to use the new COPY libpq API (Christopher)
605 o Remove unnecessary abstractions in pg_dump source code
612 Document differences between ecpg and the SQL standard and
613 information about the Informix-compatibility module.
615 o Solve cardinality > 1 for input descriptors / variables?
616 o Add a semantic check level, e.g. check if a table really exists
617 o fix handling of DB attributes that are arrays
618 o Use backend PREPARE/EXECUTE facility for ecpg where possible
620 o Fix nested C comments
621 o sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
622 o Make SET CONNECTION thread-aware, non-standard?
623 o Allow multidimensional arrays
624 o Add internationalized message strings
627 Referential Integrity
628 =====================
630 * Add MATCH PARTIAL referential integrity
631 * Add deferred trigger queue file
633 Right now all deferred trigger information is stored in backend
634 memory. This could exhaust memory for very large trigger queues.
635 This item involves dumping large queues into files.
637 * -Implement shared row locks and use them in RI triggers
638 * Change foreign key constraint for array -> element to mean element
640 * Allow DEFERRABLE UNIQUE constraints?
641 * Allow triggers to be disabled [trigger]
643 Currently the only way to disable triggers is to modify the system
646 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
648 If the dump is known to be valid, allow foreign keys to be added
649 without revalidating the data.
651 * Allow statement-level triggers to access modified rows
652 * Support triggers on columns (Greg Sabino Mullane)
653 * Remove CREATE CONSTRAINT TRIGGER
655 This was used in older releases to dump referential integrity
658 * Enforce referential integrity for system tables
659 * Allow AFTER triggers on system tables
661 System tables are modified in many places in the backend without going
662 through the executor and therefore not causing triggers to fire. To
663 complete this item, the functions that modify system tables will have
670 * Flush cached query plans when the dependent objects change
671 * Track dependencies in function bodies and recompile/invalidate
677 * Add SQL99 WITH clause to SELECT
678 * Add SQL99 WITH RECURSIVE to SELECT
679 * Add pre-parsing phase that converts non-ISO syntax to supported
682 This could allow SQL written for other databases to run without
685 * Allow plug-in modules to emulate features from other databases
686 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
688 * Allow queries across databases or servers with transaction
691 This can be done using dblink and two-phase commit.
693 * -Add two-phase commit
696 * Add the features of packages
698 o Make private objects accessable only to objects in the same schema
699 o Allow current_schema.objname to access current schema objects
700 o Add session variables
701 o Allow nested schemas
707 * Allow inherited tables to inherit index, UNIQUE constraint, and primary
709 * UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
710 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
713 The main difficulty with this item is the problem of creating an index
714 that can span more than one table.
716 * Allow SELECT ... FOR UPDATE on inherited tables
717 * -Prevent inherited tables from expanding temporary subtables of other
719 * Add UNIQUE capability to non-btree indexes
720 * -Use indexes for MIN() and MAX()
722 MIN/MAX queries can already be rewritten as SELECT col FROM tab ORDER
723 BY col {DESC} LIMIT 1. Completing this item involves doing this
724 transformation automatically.
726 * -Use index to restrict rows returned by multi-key index when used with
727 non-consecutive keys to reduce heap accesses
729 For an index on col1,col2,col3, and a WHERE clause of col1 = 5 and
730 col3 = 9, spin though the index checking for col1 and col3 matches,
731 rather than just col1; also called skip-scanning.
733 * Prevent index uniqueness checks when UPDATE does not modify the column
735 Uniqueness (index) checks are done when updating a column even if the
736 column is not modified by the UPDATE.
738 * Fetch heap pages matching index entries in sequential order
740 Rather than randomly accessing heap pages based on index entries, mark
741 heap pages needing access in a bitmap and do the lookups in sequential
742 order. Another method would be to sort heap ctids matching the index
743 before accessing the heap rows.
745 * -Allow non-bitmap indexes to be combined by creating bitmaps in memory
747 This feature allows separate indexes to be ANDed or ORed together. This
748 is particularly useful for data warehousing applications that need to
749 query the database in an many permutations. This feature scans an index
750 and creates an in-memory bitmap, and allows that bitmap to be combined
751 with other bitmap created in a similar way. The bitmap can either index
752 all TIDs, or be lossy, meaning it records just page numbers and each
753 page tuple has to be checked for validity in a separate pass.
755 * Allow the creation of on-disk bitmap indexes which can be quickly
756 combined with other bitmap indexes
758 Such indexes could be more compact if there are only a few distinct values.
759 Such indexes can also be compressed. Keeping such indexes updated can be
762 * Allow use of indexes to search for NULLs
764 One solution is to create a partial index on an IS NULL expression.
766 * Allow accurate statistics to be collected on indexes with more than
767 one column or expression indexes, perhaps using per-index statistics
768 * Add fillfactor to control reserved free space during index creation
769 * Allow the creation of indexes with mixed ascending/descending specifiers
770 * -Fix incorrect rtree results due to wrong assumptions about "over"
772 * Allow constraint_exclusion to work for UNIONs like it does for
773 inheritance, and allow it to work for UPDATE and DELETE queries
778 o Add more GIST index support for geometric data types
779 o -Add concurrency to GIST
780 o Allow GIST indexes to create certain complex index types, like
781 digital trees (see Aoki)
785 o Pack hash index buckets onto disk pages more efficiently
787 Currently no only one hash bucket can be stored on a page. Ideally
788 several hash buckets could be stored on a single page and greater
789 granularity used for the hash algorithm.
791 o Consider sorting hash buckets so entries can be found using a
792 binary search, rather than a linear scan
794 o In hash indexes, consider storing the hash value with or instead
801 * Improve commit_delay handling to reduce fsync()
802 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
803 * -Allow multiple blocks to be written to WAL with one write()
804 * Add an option to sync() before fsync()'ing checkpoint files
805 * Add program to test if fsync has a delay compared to non-fsync
811 * Allow free-behind capability for large sequential scans, perhaps using
814 Posix_fadvise() can control both sequential/random file caching and
815 free-behind behavior, but it is unclear how the setting affects other
816 backends that also have the file open, and the feature is not supported
817 on all operating systems.
819 * -Consider use of open/fcntl(O_DIRECT) to minimize OS caching,
822 O_DIRECT doesn't have the same media write guarantees as fsync, so it
823 is in addition to the fsync method, not in place of it.
825 * -Cache last known per-tuple offsets to speed long tuple access
828 We could use a fixed row count and a +/- count to follow MVCC
829 visibility rules, or a single cached value could be used and
830 invalidated if anyone modifies the table. Another idea is to
831 get a count directly from a unique index, but for this to be
832 faster than a sequential scan it must avoid access to the heap
833 to obtain tuple visibility information.
835 * Allow data to be pulled directly from indexes
837 Currently indexes do not have enough tuple visibility information
838 to allow data to be pulled from the index without also accessing
839 the heap. One way to allow this is to set a bit to index tuples
840 to indicate if a tuple is currently visible to all transactions
841 when the first valid heap lookup happens. This bit would have to
842 be cleared when a heap tuple is expired.
845 * Consider automatic caching of queries at various levels:
851 * -Allow the size of the buffer cache used by temporary objects to be
852 specified as a GUC variable
854 Larger local buffer cache sizes requires more efficient handling of
857 * Improve the background writer
859 Allow the background writer to more efficiently write dirty buffers
860 from the end of the LRU cache and use a clock sweep algorithm to
861 write other dirty buffers to reduced checkpoint I/O
863 * Allow sequential scans to take advantage of other concurrent
864 sequentiqal scans, also called "Synchronised Scanning"
866 One possible implementation is to start sequential scans from the lowest
867 numbered buffer in the shared cache, and when reaching the end wrap
868 around to the beginning, rather than always starting sequential scans
869 at the start of the table.
875 * Improve speed with indexes
877 For large table adjustements during vacuum, it is faster to reindex
878 rather than update the index.
880 * Reduce lock time by moving tuples with read lock, then write
881 lock and truncate table
883 Moved tuples are invisible to other backends so they don't require a
884 write lock. However, the read lock promotion to write lock could lead
885 to deadlock situations.
887 * -Add a warning when the free space map is too small
888 * Maintain a map of recently-expired rows
890 This allows vacuum to target specific pages for possible free space
891 without requiring a sequential scan.
893 * Auto-fill the free space map by scanning the buffer cache or by
894 checking pages written by the background writer
895 * Create a bitmap of pages that need vacuuming
897 Instead of sequentially scanning the entire table, have the background
898 writer or some other process record pages that have expired rows, then
899 VACUUM can look at just those pages rather than the entire table. In
900 the event of a system crash, the bitmap would probably be invalidated.
902 * Add system view to show free space map contents
907 o -Move into the backend code
908 o Use free-space map information to guide refilling
909 o Do VACUUM FULL if table is nearly empty?
910 o Improve xid wraparound detection by recording per-table rather
917 * -Make locking of shared data structures more fine-grained
919 This requires that more locks be acquired but this would reduce lock
920 contention, improving concurrency.
922 * Add code to detect an SMP machine and handle spinlocks accordingly
923 from distributted.net, http://www1.distributed.net/source,
924 in client/common/cpucheck.cpp
926 On SMP machines, it is possible that locks might be released shortly,
927 while on non-SMP machines, the backend should sleep so the process
928 holding the lock can complete and release it.
930 * -Improve SMP performance on i386 machines
932 i386-based SMP machines can generate excessive context switching
933 caused by lock failure in high concurrency situations. This may be
934 caused by CPU cache line invalidation inefficiencies.
936 * Research use of sched_yield() for spinlock acquisition failure
937 * Fix priority ordering of read and write light-weight locks (Neil)
940 Startup Time Improvements
941 =========================
943 * Experiment with multi-threaded backend [thread]
945 This would prevent the overhead associated with process creation. Most
946 operating systems have trivial process creation time compared to
947 database startup overhead, but a few operating systems (WIn32,
948 Solaris) might benefit from threading. Also explore the idea of
949 a single session using multiple threads to execute a query faster.
951 * Add connection pooling
953 It is unclear if this should be done inside the backend code or done
954 by something external like pgpool. The passing of file descriptors to
955 existing backends is one of the difficulties with a backend approach.
961 * Eliminate need to write full pages to WAL before page modification [wal]
963 Currently, to protect against partial disk page writes, we write
964 full page images to WAL before they are modified so we can correct any
965 partial page writes during recovery. These pages can also be
966 eliminated from point-in-time archive files.
968 o -Add ability to turn off full page writes
969 o When off, write CRC to WAL and check file system blocks
972 If CRC check fails during recovery, remember the page in case
973 a later CRC for that page properly matches.
975 o Write full pages during file system write and not when
976 the page is modified in the buffer cache
978 This allows most full page writes to happen in the background
979 writer. It might cause problems for applying WAL on recovery
980 into a partially-written page, but later the full page will be
983 * Reduce WAL traffic so only modified values are written rather than
985 * Add WAL index reliability improvement to non-btree indexes
986 * Allow the pg_xlog directory location to be specified during initdb
987 with a symlink back to the /data location
988 * Allow WAL information to recover corrupted pg_controldata
989 * Find a way to reduce rotational delay when repeatedly writing
992 Currently fsync of WAL requires the disk platter to perform a full
993 rotation to fsync again. One idea is to write the WAL to different
994 offsets that might reduce the rotational delay.
996 * Allow buffered WAL writes and fsync
998 Instead of guaranteeing recovery of all committed transactions, this
999 would provide improved performance by delaying WAL writes and fsync
1000 so an abrupt operating system restart might lose a few seconds of
1001 committed transactions but still be consistent. We could perhaps
1002 remove the 'fsync' parameter (which results in an an inconsistent
1003 database) in favor of this capability.
1005 * -Eliminate WAL logging for CREATE TABLE AS when not doing WAL archiving
1006 * -Change WAL to use 32-bit CRC, for performance reasons
1009 Optimizer / Executor
1010 ====================
1012 * Add missing optimizer selectivities for date, r-tree, etc
1013 * Allow ORDER BY ... LIMIT # to select high/low value without sort or
1014 index using a sequential scan for highest/lowest values
1016 Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
1017 all values to return the high/low value. Instead The idea is to do a
1018 sequential scan to find the high/low value, thus avoiding the sort.
1019 MIN/MAX already does this, but not for LIMIT > 1.
1021 * Precompile SQL functions to avoid overhead
1022 * Create utility to compute accurate random_page_cost value
1023 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1024 * Have EXPLAIN ANALYZE highlight poor optimizer estimates
1025 * -Use CHECK constraints to influence optimizer decisions
1027 CHECK constraints contain information about the distribution of values
1028 within the table. This is also useful for implementing subtables where
1029 a tables content is distributed across several subtables.
1031 * Consider using hash buckets to do DISTINCT, rather than sorting
1033 This would be beneficial when there are few distinct values.
1035 * ANALYZE should record a pg_statistic entry for an all-NULL column
1036 * Log queries where the optimizer row estimates were dramatically
1037 different from the number of rows actually found?
1040 Miscellaneous Performance
1041 =========================
1043 * Do async I/O for faster random read-ahead of data
1045 Async I/O allows multiple I/O requests to be sent to the disk with
1046 results coming back asynchronously.
1048 * Use mmap() rather than SYSV shared memory or to write WAL files?
1050 This would remove the requirement for SYSV SHM but would introduce
1051 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1052 to prevent I/O overhead.
1054 * Consider mmap()'ing files into a backend?
1056 Doing I/O to large tables would consume a lot of address space or
1057 require frequent mapping/unmapping. Extending the file also causes
1058 mapping problems that might require mapping only individual pages,
1059 leading to thousands of mappings. Another problem is that there is no
1060 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1061 could hit disk before WAL is written.
1063 * Add a script to ask system configuration questions and tune postgresql.conf
1064 * Use a phantom command counter for nested subtransactions to reduce
1066 * Research storing disk pages with no alignment/padding
1071 * Add use of 'const' for variables in source tree
1072 * Rename some /contrib modules from pg* to pg_*
1073 * Move some things from /contrib into main tree
1074 * Move some /contrib modules out to their own project sites
1075 * Remove warnings created by -Wcast-align
1076 * Move platform-specific ps status display info from ps_status.c to ports
1077 * Add optional CRC checksum to heap and index pages
1078 * Improve documentation to build only interfaces (Marc)
1079 * Remove or relicense modules that are not under the BSD license, if possible
1080 * Remove memory/file descriptor freeing before ereport(ERROR)
1081 * Acquire lock on a relation before building a relcache entry for it
1082 * Promote debug_query_string into a server-side function current_query()
1083 * Allow the identifier length to be increased via a configure option
1084 * Remove Win32 rename/unlink looping if unnecessary
1085 * -Remove kerberos4 from source tree
1086 * Allow cross-compiling by generating the zic database on the target system
1087 * Improve NLS maintenace of libpgport messages linked onto applications
1088 * Allow ecpg to work with MSVC and BCC
1089 * -Make src/port/snprintf.c thread-safe
1090 * Add xpath_array() to /contrib/xml2 to return results as an array
1091 * Allow building in directories containing spaces
1093 This is probably not possible because 'gmake' and other compiler tools
1094 do not fully support quoting of paths with spaces.
1096 * Allow installing to directories containing spaces
1098 This is possible if proper quoting is added to the makefiles for the
1099 install targets. Because PostgreSQL supports relocatable installs, it
1100 is already possible to install into a directory that doesn't contain
1101 spaces and then copy the install to a directory with spaces.
1103 * Fix cross-compiling of time zone database via 'zic'
1104 * Fix sgmltools so PDFs can be generated with bookmarks
1105 * -Add C code on Unix to copy directories for use in creating new databases
1110 o Remove configure.in check for link failure when cause is found
1111 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1113 o Remove psql newline patch when we find out why mingw outputs an
1115 o Allow psql to use readline once non-US code pages work with
1117 o Re-enable timezone output on log_line_prefix '%t' when a
1118 shorter timezone string is available
1119 o Improve dlerror() reporting string
1120 o Fix problem with shared memory on the Win32 Terminal Server
1121 o Add support for Unicode
1123 To fix this, the data needs to be converted to/from UTF16/UTF8
1124 so the Win32 wcscoll() can be used, and perhaps other functions
1125 like towupper(). However, UTF8 already works with normal
1126 locales but provides no ordering or character set classes.
1129 * Wire Protocol Changes
1131 o Allow dynamic character set handling
1132 o Add decoded type, length, precision
1134 o Update clients to use data types, typmod, schema.table.column names
1135 of result sets using new query protocol
1138 ---------------------------------------------------------------------------
1141 Developers who have claimed items are:
1142 --------------------------------------
1143 * Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
1144 * Andrew is Andrew Dunstan <andrew@dunslane.net>
1145 * Bruce is Bruce Momjian <pgman@candle.pha.pa.us> of Software Research Assoc.
1146 * Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
1147 Family Health Network
1148 * Claudio is Claudio Natoli <claudio.natoli@memetrics.com>
1149 * D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
1150 * Fabien is Fabien Coelho <coelho@cri.ensmp.fr>
1151 * Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
1152 * Greg is Greg Sabino Mullane <greg@turnstep.com>
1153 * Hiroshi is Hiroshi Inoue <Inoue@tpf.co.jp>
1154 * Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
1155 * Joe is Joe Conway <mail@joeconway.com>
1156 * Karel is Karel Zak <zakkr@zf.jcu.cz>
1157 * Magnus is Magnus Hagander <mha@sollentuna.net>
1158 * Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
1159 * Matthew T. O'Connor <matthew@zeut.net>
1160 * Michael is Michael Meskes <meskes@postgresql.org> of Credativ
1161 * Neil is Neil Conway <neilc@samurai.com>
1162 * Oleg is Oleg Bartunov <oleg@sai.msu.su>
1163 * Peter is Peter Eisentraut <peter_e@gmx.net>
1164 * Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
1165 * Rod is Rod Taylor <pg@rbt.ca>
1166 * Simon is Simon Riggs <simon@2ndquadrant.com>
1167 * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
1168 * Tatsuo is Tatsuo Ishii <t-ishii@sra.co.jp> of Software Research Assoc.
1169 * Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat