4 Current maintainer: Bruce Momjian (pgman@candle.pha.pa.us)
5 Last updated: Thu Dec 1 23:28:03 EST 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 8.2 release.#
11 #A percent sign, "%", marks items that are easier to implement.#
13 Bracketed items, "[]", have more detail.
15 This list contains all known PostgreSQL bugs and feature requests. If
16 you would like to work on an item, please read the Developer's FAQ
23 * %Remove behavior of postmaster -o after making postmaster/postgres
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 Lock table corruption following SIGTERM of an individual backend
38 has been reported in 8.0. A possible cause was fixed in 8.1, but
39 it is unknown whether other problems exist. This item mostly
40 requires additional testing rather than of writing any new code.
42 * %Set proper permissions on non-system schemas during db creation
44 Currently all schemas are owned by the super-user because they are
45 copied from the template1 database.
47 * Support table partitioning that allows a single table to be stored
48 in subtables that are partitioned based on the primary key or a WHERE
50 * Add function to report the time of the most recent server reload
53 * Improve replication solutions
57 You can use any of the master/slave replication servers to use a
58 standby server for data warehousing. To allow read/write queries to
59 multiple servers, you need multi-master replication like pgcluster.
61 o Allow replication over unreliable or non-persistent links
66 o %Add "include file" functionality in postgresql.conf
67 o %Allow commenting of variables in postgresql.conf to restore them
70 Currently, if a variable is commented out, it keeps the
71 previous uncommented value until a server restarted.
73 o %Allow pg_hba.conf settings to be controlled via SQL
75 This would add a function to load the SQL table from
76 pg_hba.conf, and one to writes its contents to the flat file.
77 The table should have a line number that is a float so rows
78 can be inserted between existing rows, e.g. row 2.5 goes
79 between row 2 and row 3.
81 o %Allow postgresql.conf file values to be changed via an SQL
82 API, perhaps using SET GLOBAL
83 o Allow the server to be stopped/restarted via an SQL API
84 o Issue a warning if a change-on-restart-only postgresql.conf value
85 is modified and the server config files are reloaded
86 o Mark change-on-restart-only values in postgresql.conf
91 * Allow a database in tablespace t1 with tables created in
92 tablespace t2 to be used as a template for a new database created
93 with default tablespace t2
95 All objects in the default database tablespace must have default
96 tablespace specifications. This is because new databases are
97 created by copying directories. If you mix default tablespace
98 tables and tablespace-specified tables in the same directory,
99 creating a new database from such a mixed directory would create a
100 new database with tables that had incorrect explicit tablespaces.
101 To fix this would require modifying pg_class in the newly copied
102 database, which we don't currently do.
104 * Allow reporting of which objects are in which tablespaces
106 This item is difficult because a tablespace can contain objects
107 from multiple databases. There is a server-side function that
108 returns the databases which use a specific tablespace, so this
109 requires a tool that will call that function and connect to each
110 database to find the objects in each database for that tablespace.
112 o %Add a GUC variable to control the tablespace for temporary objects
115 It could start with a random tablespace from a supplied list and
116 cycle through the list.
118 o Allow WAL replay of CREATE TABLESPACE to work when the directory
119 structure on the recovery computer is different from the original
121 o Allow per-tablespace quotas
124 * Point-In-Time Recovery (PITR)
126 o Allow point-in-time recovery to archive partially filled
127 write-ahead logs [pitr]
129 Currently only full WAL files are archived. This means that the
130 most recent transactions aren't available for recovery in case
131 of a disk failure. This could be triggered by a user command or
134 o Automatically force archiving of partially-filled WAL files when
135 pg_stop_backup() is called or the server is stopped
137 Doing this will allow administrators to know more easily when
138 the archive contins all the files needed for point-in-time
141 o %Create dump tool for write-ahead logs for use in determining
142 transaction id for point-in-time recovery
143 o Allow a warm standby system to also allow read-only queries
146 This is useful for checking PITR recovery.
148 o Allow the PITR process to be debugged and data examined
154 * Allow server log information to be output as INSERT statements
156 This would allow server log information to be easily loaded into
157 a database for analysis.
159 * %Add ability to monitor the use of temporary sort files
160 * Allow server logs to be remotely read and removed using SQL commands
161 * Allow protocol-level BIND parameter values to be logged
167 * Improve the MONEY data type
169 Change the MONEY data type to use DECIMAL internally, with special
170 locale-aware output formatting.
172 * Change NUMERIC to enforce the maximum precision, and increase it
173 * Add NUMERIC division operator that doesn't round?
175 Currently NUMERIC _rounds_ the result to the specified precision.
176 This means division can return a result that multiplied by the
177 divisor is greater than the dividend, e.g. this returns a value > 10:
179 SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
181 The positive modulus result returned by NUMERICs might be considered
182 inaccurate, in one sense.
184 * %Disallow changing default expression of a SERIAL column?
185 * Fix data types where equality comparison isn't intuitive, e.g. box
186 * %Prevent INET cast to CIDR if the unmasked bits are not zero, or
188 * %Prevent INET cast to CIDR from droping netmask, SELECT '1.1.1.1'::inet::cidr
189 * Allow INET + INT4 to increment the host part of the address, or
190 throw an error on overflow
191 * %Add 'tid != tid ' operator for use in corruption recovery
192 * Allow user-defined types to specify a type modifier at table creation
198 o Allow infinite dates just like infinite timestamps
199 o Merge hardwired timezone names with the TZ database; allow either
200 kind everywhere a TZ name is currently taken
201 o Allow customization of the known set of TZ names (generalize the
202 present australian_timezones hack)
203 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
204 information, either zone name or offset from UTC [timezone]
206 If the TIMESTAMP value is stored with a time zone name, interval
207 computations should adjust based on the time zone rules.
209 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
210 o Fix SELECT INTERVAL '1' MONTH
211 o Add a GUC variable to allow output of interval values in ISO8601
213 o Improve timestamptz subtraction to be DST-aware
215 Currently, subtracting one date from another that crosses a
216 daylight savings time adjustment can return '1 day 1 hour', but
217 adding that back to the first date returns a time one hour in
218 the future. This is caused by the adjustment of '25 hours' to
219 '1 day 1 hour', and '1 day' is the same time the next day, even
220 if daylight savings adjustments are involved.
222 o Fix interval display to support values exceeding 2^31 hours
223 o Add overflow checking to timestamp and interval arithmetic
224 o Add ISO INTERVAL handling
225 o Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
227 o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
228 o For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
229 '1', treat as ISO if there is a range specification clause,
230 and as PG if there no clause is present, e.g. interpret
231 '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
232 interpret '1:30' as '1 hour, 30 minutes'
233 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
234 INTERVAL MONTH), and this should return '12 months'
235 o Round or truncate values to the requested precision, e.g.
236 INTERVAL '11 months' AS YEAR should return one or zero
237 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
242 o -Allow NULLs in arrays
243 o Delay resolution of array expression's data type so assignment
244 coercion can be performed on empty array expressions
249 o Improve vacuum of large objects, like /contrib/vacuumlo?
250 o Add security checking for large objects
251 o Auto-delete large objects when referencing row is deleted
253 /contrib/lo offers this functionality.
255 o Allow read/write into TOAST values like large objects
257 This requires the TOAST column to be stored EXTERNAL.
263 * Allow INET subnet tests using non-constants to be indexed
264 * Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
267 Current CURRENT_TIMESTAMP returns the start time of the current
268 transaction, and gettimeofday() returns the wallclock time. This will
269 make time reporting more consistent and will allow reporting of
270 the statement start time.
272 * %Add pg_get_acldef(), pg_get_typedefault(), and pg_get_attrdef()
273 * Allow to_char() to print localized month names
274 * Allow functions to have a schema search path specified at creation time
275 * Allow substring/replace() to get/set bit values
276 * Allow to_char() on interval values to accumulate the highest unit
279 Some special format flag would be required to request such
280 accumulation. Such functionality could also be added to EXTRACT.
281 Prevent accumulation that crosses the month/day boundary because of
282 the uneven number of days in a month.
284 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
285 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
286 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
287 o to_char(INTERVAL '3 years 5 months','MM') => 41
289 * Add sleep() function, remove from regress.c
290 * Allow user-defined functions retuning a domain value to enforce domain
294 Multi-Language Support
295 ======================
297 * Add NCHAR (as distinguished from ordinary varchar),
298 * Allow locale to be set at database creation
300 Currently locale can only be set during initdb. No global tables have
301 locale-aware columns. However, the database template used during
302 database creation might have locale-aware indexes. The indexes would
303 need to be reindexed to match the new locale.
305 * Allow encoding on a per-column basis
307 Right now only one encoding is allowed per database.
309 * Support multiple simultaneous character sets, per SQL92
310 * Improve UTF8 combined character handling?
311 * Add octet_length_server() and octet_length_client()
312 * Make octet_length_client() the same as octet_length()?
313 * Fix problems with wrong runtime encoding conversion for NLS message files
319 * %Automatically create rules on views so they are updateable, per SQL99
321 We can only auto-create rules for simple views. For more complex
322 cases users will still have to write rules.
324 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
325 * Allow NOTIFY in rules involving conditionals
326 * Allow VIEW/RULE recompilation when the underlying tables change
328 Another issue is whether underlying table changes should be reflected
329 in the view, e.g. should SELECT * show additional columns if they
330 are added after the view is created.
336 * Change LIMIT/OFFSET and FETCH/MOVE to use int8
337 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
338 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
339 * %Allow SET CONSTRAINTS to be qualified by schema/table name
340 * %Allow TRUNCATE ... CASCADE/RESTRICT
342 This is like DELETE CASCADE, but truncates.
344 * %Add a separate TRUNCATE permission
346 Currently only the owner can TRUNCATE a table because triggers are not
347 called, and the table is locked in exclusive mode.
349 * Allow PREPARE of cursors
350 * Allow PREPARE to automatically determine parameter types based on the SQL
352 * Allow finer control over the caching of prepared query plans
354 Currently, queries prepared via the libpq API are planned on first
355 execute using the supplied parameters --- allow SQL PREPARE to do the
356 same. Also, allow control over replanning prepared queries either
357 manually or automatically when statistics for execute parameters
358 differ dramatically from those used during planning.
360 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
362 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
363 such information in memory would improve performance.
365 * Add optional textual message to NOTIFY
367 This would allow an informational message to be added to the notify
368 message, perhaps indicating the row modified or other custom
371 * Add a GUC variable to warn about non-standard SQL usage in queries
372 * Add SQL-standard MERGE command, typically used to merge two tables
375 This is similar to UPDATE, then for unmatched rows, INSERT.
376 Whether concurrent access allows modifications which could cause
377 row loss is implementation independent.
379 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
382 To implement this cleanly requires that the table have a unique index
383 so duplicate checking can be easily performed. It is possible to
384 do it without a unique index if we require the user to LOCK the table
387 * Add NOVICE output level for helpful messages like automatic sequence/index
389 * %Add COMMENT ON for all cluster global objects (roles, databases
391 * %Make row-wise comparisons work per SQL spec
392 * Add RESET CONNECTION command to reset all session state
394 This would include resetting of all variables (RESET ALL), dropping of
395 temporary tables, removing any NOTIFYs, cursors, open transactions,
396 prepared queries, currval()s, etc. This could be used for connection
397 pooling. We could also change RESET ALL to have this functionality.
398 The difficult of this features is allowing RESET ALL to not affect
399 changes made by the interface driver for its internal use. One idea
400 is for this to be a protocol-only feature. Another approach is to
401 notify the protocol when a RESET CONNECTION command is used.
403 * Add GUC to issue notice about queries that use unjoined tables
404 * Allow EXPLAIN to identify tables that were skipped because of
406 * Allow EXPLAIN output to be more easily processed by scripts
407 * Eventually enable escape_string_warning and standard_conforming_strings
408 * Simplify dropping roles that have objects in several databases
409 * Allow COMMENT ON to accept an expression rather than just a string
414 o Allow CREATE TABLE AS to determine column lengths for complex
415 expressions like SELECT col1 || col2
416 o Use more reliable method for CREATE DATABASE to get a consistent
418 o Add ON COMMIT capability to CREATE TABLE AS ... SELECT
422 o Allow UPDATE to handle complex aggregates [update]?
423 o Allow an alias to be provided for the target table in
426 This is not SQL-spec but many DBMSs allow it.
428 o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
434 o %Have ALTER TABLE RENAME rename SERIAL sequence names
435 o Add ALTER DOMAIN to modify the underlying data type
436 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
437 o %Allow ALTER TABLE to change constraint deferrability and actions
438 o Add missing object types for ALTER ... SET SCHEMA
439 o Allow ALTER TABLESPACE to move to different directories
440 o Allow databases to be moved to different tablespaces
441 o Allow moving system tables to other tablespaces, where possible
443 Currently non-global system tables must be in the default database
444 tablespace. Global system tables can never be moved.
446 o %Disallow dropping of an inherited constraint
447 o %Prevent child tables from altering or dropping constraints
448 like CHECK that were inherited from the parent table
449 o Have ALTER INDEX update the name of a constraint using that index
450 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
455 o Automatically maintain clustering on a table
457 This might require some background daemon to maintain clustering
458 during periods of low usage. It might also require tables to be only
459 paritally filled for easier reorganization. Another idea would
460 be to create a merged heap/index data file so an index lookup would
461 automatically access the heap data too. A third idea would be to
462 store heap rows in hashed groups, perhaps using a user-supplied
465 o %Add default clustering to system tables
467 To do this, determine the ideal cluster index for each system
468 table and set the cluster setting during initdb.
473 o Allow COPY to report error lines and continue
475 This requires the use of a savepoint before each COPY line is
476 processed, with ROLLBACK on COPY failure.
478 o %Have COPY return the number of rows loaded/unloaded?
479 o Allow COPY on a newly-created table to skip WAL logging
481 On crash recovery, the table involved in the COPY would
482 be removed or have its heap and index files truncated. One
483 issue is that no other backend should be able to add to
484 the table at the same time, which is something that is
487 o Allow COPY to output from views
489 Another idea would be to allow actual SELECT queries in a COPY.
494 o Allow column-level privileges
495 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
498 The proposed syntax is:
499 GRANT SELECT ON ALL TABLES IN public TO phpuser;
500 GRANT SELECT ON NEW TABLES IN public TO phpuser;
502 * Allow GRANT/REVOKE permissions to be inherited by objects based on
508 o Allow UPDATE/DELETE WHERE CURRENT OF cursor
510 This requires using the row ctid to map cursor rows back to the
511 original heap row. This become more complicated if WITH HOLD cursors
512 are to be supported because WITH HOLD cursors have a copy of the row
513 and no FOR UPDATE lock.
515 o Prevent DROP TABLE from dropping a row referenced by its own open
518 o %Allow pooled connections to list all open WITH HOLD cursors
520 Because WITH HOLD cursors exist outside transactions, this allows
521 them to be listed so they can be closed.
526 o Allow INSERT/UPDATE of the system-generated oid value for a row
527 o Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
528 o Allow INSERT/UPDATE ... RETURNING new.col or old.col
530 This is useful for returning the auto-generated key for an INSERT.
531 One complication is how to handle rules that run as part of
537 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
539 o Add SET PATH for schemas?
541 This is basically the same as SET search_path.
544 * Server-Side Languages
546 o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
547 o Allow function parameters to be passed by name,
548 get_employee_salary(emp_id => 12345, tax_year => 2001)
549 o Add Oracle-style packages
550 o Add table function support to pltcl, plpython
551 o Add capability to create and call PROCEDURES
552 o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
553 o Allow function argument names to be queries from PL/PgSQL
554 o Add MOVE to PL/pgSQL
555 o Add support for polymorphic arguments and return types to
556 languages other than PL/PgSQL
557 o Add support for OUT and INOUT parameters to languages other
559 o Add single-step debugging of PL/PgSQL functions
560 o Allow PL/PgSQL to support WITH HOLD cursors
566 * Have initdb set the input DateStyle (MDY or DMY) based on locale?
567 * Have pg_ctl look at PGHOST in case it is a socket directory?
568 * Allow pg_ctl to work properly with configuration files located outside
571 pg_ctl can not read the pid file because it isn't located in the
572 config directory but in the PGDATA directory. The solution is to
573 allow pg_ctl to read and understand postgresql.conf to find the
574 data_directory value.
579 o Have psql show current values for a sequence
580 o Move psql backslash database information into the backend, use
581 mnemonic commands? [psql]
583 This would allow non-psql clients to pull the same information out
584 of the database as psql.
586 o Fix psql's display of schema information (Neil)
587 o Allow psql \pset boolean variables to set to fixed values, rather
589 o Consistently display privilege information for all objects in psql
590 o Improve psql's handling of multi-line queries
592 Currently, while \e saves a single query as one entry, interactive
593 queries are saved one line at a time. Ideally all queries
594 whould be saved like \e does.
596 o Allow multi-line column values to align in the proper columns
598 If the second output column value is 'a\nb', the 'b' should appear
599 in the second display column, rather than the first column as it
602 o Display IN, INOUT, and OUT parameters in \df+
604 It probably requires psql to output newlines in the proper
605 column, which is already on the TODO list.
607 o Add auto-expanded mode so expanded output is used if the row
608 length is wider than the screen width.
610 Consider using auto-expanded mode for backslash commands like \df+.
615 o %Have pg_dump use multi-statement transactions for INSERT dumps
616 o %Allow pg_dump to use multiple -t and -n switches [pg_dump]
617 o %Add dumping of comments on index columns and composite type columns
618 o %Add full object name to the tag field. eg. for operators we need
619 '=(integer, integer)', instead of just '='.
620 o Add pg_dumpall custom format dumps?
621 o %Add CSV output format
622 o Update pg_dump and psql to use the new COPY libpq API (Christopher)
623 o Remove unnecessary function pointer abstractions in pg_dump source
625 o Allow selection of individual object(s) of all types, not just
627 o In a selective dump, allow dumping of an object and all its
629 o Add options like pg_restore -l and -L to pg_dump
630 o Stop dumping CASCADE on DROP TYPE commands in clean mode
631 o Allow pg_dump --clean to drop roles that own objects or have
633 o Add -f to pg_dumpall
640 Document differences between ecpg and the SQL standard and
641 information about the Informix-compatibility module.
643 o Solve cardinality > 1 for input descriptors / variables?
644 o Add a semantic check level, e.g. check if a table really exists
645 o fix handling of DB attributes that are arrays
646 o Use backend PREPARE/EXECUTE facility for ecpg where possible
648 o Fix nested C comments
649 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
650 o Make SET CONNECTION thread-aware, non-standard?
651 o Allow multidimensional arrays
652 o Add internationalized message strings
657 o Add a function to support Parse/DescribeStatement capability
658 o Add PQescapeIdentifier()
659 o Prevent PQfnumber() from lowercasing unquoted the column name
661 PQfnumber() should never have been doing lowercasing, but
662 historically it has so we need a way to prevent it
664 o Allow query results to be automatically batched to the client
666 Currently, all query results are transfered to the libpq
667 client before libpq makes the results available to the
668 application. This feature would allow the application to make
669 use of the first result rows while the rest are transfered, or
670 held on the server waiting for them to be requested by libpq.
671 One complexity is that a query like SELECT 1/col could error
672 out mid-way through the result set.
675 Referential Integrity
676 =====================
678 * Add MATCH PARTIAL referential integrity
679 * Add deferred trigger queue file
681 Right now all deferred trigger information is stored in backend
682 memory. This could exhaust memory for very large trigger queues.
683 This item involves dumping large queues into files.
685 * Change foreign key constraint for array -> element to mean element
687 * Allow DEFERRABLE UNIQUE constraints?
688 * Allow triggers to be disabled in only the current session.
690 This is currently possible by starting a multi-statement transaction,
691 modifying the system tables, performing the desired SQL, restoring the
692 system tables, and committing the transaction. ALTER TABLE ...
693 TRIGGER requires a table lock so it is not ideal for this usage.
695 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
697 If the dump is known to be valid, allow foreign keys to be added
698 without revalidating the data.
700 * Allow statement-level triggers to access modified rows
701 * Support triggers on columns (Greg Sabino Mullane)
702 * Enforce referential integrity for system tables
703 * Allow AFTER triggers on system tables
705 System tables are modified in many places in the backend without going
706 through the executor and therefore not causing triggers to fire. To
707 complete this item, the functions that modify system tables will have
714 * Flush cached query plans when the dependent objects change
715 * Track dependencies in function bodies and recompile/invalidate
717 This is particularly important for references to temporary tables
718 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
719 in PL/PgSQL is to use EXECUTE. One complexity is that a function
720 might itself drop and recreate dependent tables, causing it to
721 invalidate its own query plan.
727 * Add SQL99 WITH clause to SELECT
728 * Add SQL99 WITH RECURSIVE to SELECT
729 * Add pre-parsing phase that converts non-ISO syntax to supported
732 This could allow SQL written for other databases to run without
735 * Allow plug-in modules to emulate features from other databases
736 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
738 * Allow queries across databases or servers with transaction
741 This can be done using dblink and two-phase commit.
743 * Add the features of packages
745 o Make private objects accessable only to objects in the same schema
746 o Allow current_schema.objname to access current schema objects
747 o Add session variables
748 o Allow nested schemas
754 * Allow inherited tables to inherit index, UNIQUE constraint, and primary
756 * UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
757 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
760 The main difficulty with this item is the problem of creating an index
761 that can span more than one table.
763 * Allow SELECT ... FOR UPDATE on inherited tables
764 * Add UNIQUE capability to non-btree indexes
765 * Prevent index uniqueness checks when UPDATE does not modify the column
767 Uniqueness (index) checks are done when updating a column even if the
768 column is not modified by the UPDATE.
770 * Allow the creation of on-disk bitmap indexes which can be quickly
771 combined with other bitmap indexes
773 Such indexes could be more compact if there are only a few distinct values.
774 Such indexes can also be compressed. Keeping such indexes updated can be
777 * Allow use of indexes to search for NULLs
779 One solution is to create a partial index on an IS NULL expression.
781 * Allow accurate statistics to be collected on indexes with more than
782 one column or expression indexes, perhaps using per-index statistics
783 * Add fillfactor to control reserved free space during index creation
784 * Allow the creation of indexes with mixed ascending/descending specifiers
785 * Allow constraint_exclusion to work for UNIONs like it does for
786 inheritance, allow it to work for UPDATE and DELETE queries, and allow
787 it to be used for all queries with little performance impact
788 * Allow CREATE INDEX to take an additional parameter for use with
790 * Consider compressing indexes by storing key values duplicated in
791 several rows as a single index entry
793 This is difficult because it requires datatype-specific knowledge.
798 o Add more GIST index support for geometric data types
799 o Allow GIST indexes to create certain complex index types, like
800 digital trees (see Aoki)
804 o Pack hash index buckets onto disk pages more efficiently
806 Currently only one hash bucket can be stored on a page. Ideally
807 several hash buckets could be stored on a single page and greater
808 granularity used for the hash algorithm.
810 o Consider sorting hash buckets so entries can be found using a
811 binary search, rather than a linear scan
813 o In hash indexes, consider storing the hash value with or instead
816 o Add WAL logging for crash recovery
817 o Allow multi-column hash indexes
823 * Improve commit_delay handling to reduce fsync()
824 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
826 Ideally this requires a separate test program that can be run
827 at initdb time or optionally later.
829 * %Add an option to sync() before fsync()'ing checkpoint files
830 * Add program to test if fsync has a delay compared to non-fsync
836 * Allow free-behind capability for large sequential scans, perhaps using
839 Posix_fadvise() can control both sequential/random file caching and
840 free-behind behavior, but it is unclear how the setting affects other
841 backends that also have the file open, and the feature is not supported
842 on all operating systems.
846 We could use a fixed row count and a +/- count to follow MVCC
847 visibility rules, or a single cached value could be used and
848 invalidated if anyone modifies the table. Another idea is to
849 get a count directly from a unique index, but for this to be
850 faster than a sequential scan it must avoid access to the heap
851 to obtain tuple visibility information.
853 * Add estimated_count(*) to return an estimate of COUNT(*)
855 This would use the planner ANALYZE statistatics to return an estimated
858 * Allow data to be pulled directly from indexes
860 Currently indexes do not have enough tuple visibility information
861 to allow data to be pulled from the index without also accessing
862 the heap. One way to allow this is to set a bit on index tuples
863 to indicate if a tuple is currently visible to all transactions
864 when the first valid heap lookup happens. This bit would have to
865 be cleared when a heap tuple is expired.
867 Another idea is to maintain a bitmap of heap pages where all rows
868 are visible to all backends, and allow index lookups to reference
869 that bitmap to avoid heap lookups, perhaps the same bitmap we might
870 add someday to determine which heap pages need vacuuming. Frequently
871 accessed bitmaps would have to be stored in shared memory. One 8k
872 page of bitmaps could track 512MB of heap pages.
874 * Consider automatic caching of queries at various levels:
880 * Allow sequential scans to take advantage of other concurrent
881 sequentiqal scans, also called "Synchronised Scanning"
883 One possible implementation is to start sequential scans from the lowest
884 numbered buffer in the shared cache, and when reaching the end wrap
885 around to the beginning, rather than always starting sequential scans
886 at the start of the table.
892 * Improve speed with indexes
894 For large table adjustements during VACUUM FULL, it is faster to
895 reindex rather than update the index.
897 * Reduce lock time during VACUUM FULL by moving tuples with read lock,
898 then write lock and truncate table
900 Moved tuples are invisible to other backends so they don't require a
901 write lock. However, the read lock promotion to write lock could lead
902 to deadlock situations.
904 * Auto-fill the free space map by scanning the buffer cache or by
905 checking pages written by the background writer
906 * Create a bitmap of pages that need vacuuming
908 Instead of sequentially scanning the entire table, have the background
909 writer or some other process record pages that have expired rows, then
910 VACUUM can look at just those pages rather than the entire table. In
911 the event of a system crash, the bitmap would probably be invalidated.
912 One complexity is that index entries still have to be vacuumed, and
913 doing this without an index scan (by using the heap values to find the
914 index entry) might be slow and unreliable, especially for user-defined
917 * %Add system view to show free space map contents
922 o Use free-space map information to guide refilling
923 o %Issue log message to suggest VACUUM FULL if a table is nearly
925 o Improve xid wraparound detection by recording per-table rather
932 * Fix priority ordering of read and write light-weight locks (Neil)
935 Startup Time Improvements
936 =========================
938 * Experiment with multi-threaded backend [thread]
940 This would prevent the overhead associated with process creation. Most
941 operating systems have trivial process creation time compared to
942 database startup overhead, but a few operating systems (WIn32,
943 Solaris) might benefit from threading. Also explore the idea of
944 a single session using multiple threads to execute a query faster.
946 * Add connection pooling
948 It is unclear if this should be done inside the backend code or done
949 by something external like pgpool. The passing of file descriptors to
950 existing backends is one of the difficulties with a backend approach.
956 * Eliminate need to write full pages to WAL before page modification [wal]
958 Currently, to protect against partial disk page writes, we write
959 full page images to WAL before they are modified so we can correct any
960 partial page writes during recovery. These pages can also be
961 eliminated from point-in-time archive files.
963 o When off, write CRC to WAL and check file system blocks
966 If CRC check fails during recovery, remember the page in case
967 a later CRC for that page properly matches.
969 o Write full pages during file system write and not when
970 the page is modified in the buffer cache
972 This allows most full page writes to happen in the background
973 writer. It might cause problems for applying WAL on recovery
974 into a partially-written page, but later the full page will be
977 * Allow WAL traffic to be streamed to another server for stand-by
979 * Reduce WAL traffic so only modified values are written rather than
981 * Allow the pg_xlog directory location to be specified during initdb
982 with a symlink back to the /data location
983 * Allow WAL information to recover corrupted pg_controldata
984 * Find a way to reduce rotational delay when repeatedly writing
987 Currently fsync of WAL requires the disk platter to perform a full
988 rotation to fsync again. One idea is to write the WAL to different
989 offsets that might reduce the rotational delay.
991 * Allow buffered WAL writes and fsync
993 Instead of guaranteeing recovery of all committed transactions, this
994 would provide improved performance by delaying WAL writes and fsync
995 so an abrupt operating system restart might lose a few seconds of
996 committed transactions but still be consistent. We could perhaps
997 remove the 'fsync' parameter (which results in an an inconsistent
998 database) in favor of this capability.
1001 Optimizer / Executor
1002 ====================
1004 * Add missing optimizer selectivities for date, r-tree, etc
1005 * Allow ORDER BY ... LIMIT # to select high/low value without sort or
1006 index using a sequential scan for highest/lowest values
1008 Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
1009 all values to return the high/low value. Instead The idea is to do a
1010 sequential scan to find the high/low value, thus avoiding the sort.
1011 MIN/MAX already does this, but not for LIMIT > 1.
1013 * Precompile SQL functions to avoid overhead
1014 * Create utility to compute accurate random_page_cost value
1015 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1016 * Have EXPLAIN ANALYZE highlight poor optimizer estimates
1017 * Consider using hash buckets to do DISTINCT, rather than sorting
1019 This would be beneficial when there are few distinct values. This is
1020 already used by GROUP BY.
1022 * Log queries where the optimizer row estimates were dramatically
1023 different from the number of rows actually found?
1026 Miscellaneous Performance
1027 =========================
1029 * Do async I/O for faster random read-ahead of data
1031 Async I/O allows multiple I/O requests to be sent to the disk with
1032 results coming back asynchronously.
1034 * Use mmap() rather than SYSV shared memory or to write WAL files?
1036 This would remove the requirement for SYSV SHM but would introduce
1037 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1038 to prevent I/O overhead.
1040 * Consider mmap()'ing files into a backend?
1042 Doing I/O to large tables would consume a lot of address space or
1043 require frequent mapping/unmapping. Extending the file also causes
1044 mapping problems that might require mapping only individual pages,
1045 leading to thousands of mappings. Another problem is that there is no
1046 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1047 could hit disk before WAL is written.
1049 * Add a script to ask system configuration questions and tune postgresql.conf
1050 * Merge xmin/xmax/cmin/cmax back into three header fields
1052 Before subtransactions, there used to be only three fields needed to
1053 store these four values. This was possible because only the current
1054 transaction looks at the cmin/cmax values. If the current transaction
1055 created and expired the row the fields stored where xmin (same as
1056 xmax), cmin, cmax, and if the transaction was expiring a row from a
1057 another transaction, the fields stored were xmin (cmin was not
1058 needed), xmax, and cmax. Such a system worked because a transaction
1059 could only see rows from another completed transaction. However,
1060 subtransactions can see rows from outer transactions, and once the
1061 subtransaction completes, the outer transaction continues, requiring
1062 the storage of all four fields. With subtransactions, an outer
1063 transaction can create a row, a subtransaction expire it, and when the
1064 subtransaction completes, the outer transaction still has to have
1065 proper visibility of the row's cmin, for example, for cursors.
1067 One possible solution is to create a phantom cid which represents a
1068 cmin/cmax pair and is stored in local memory. Another idea is to
1069 store both cmin and cmax only in local memory.
1071 * Research storing disk pages with no alignment/padding
1077 * Add use of 'const' for variables in source tree
1078 * Rename some /contrib modules from pg* to pg_*
1079 * Move some things from /contrib into main tree
1080 * Move some /contrib modules out to their own project sites
1081 * %Remove warnings created by -Wcast-align
1082 * Move platform-specific ps status display info from ps_status.c to ports
1083 * Add optional CRC checksum to heap and index pages
1084 * Improve documentation to build only interfaces (Marc)
1085 * Remove or relicense modules that are not under the BSD license, if possible
1086 * %Remove memory/file descriptor freeing before ereport(ERROR)
1087 * Acquire lock on a relation before building a relcache entry for it
1088 * %Promote debug_query_string into a server-side function current_query()
1089 * %Allow the identifier length to be increased via a configure option
1090 * Allow cross-compiling by generating the zic database on the target system
1091 * Improve NLS maintenace of libpgport messages linked onto applications
1092 * Allow ecpg to work with MSVC and BCC
1093 * Add xpath_array() to /contrib/xml2 to return results as an array
1094 * Allow building in directories containing spaces
1096 This is probably not possible because 'gmake' and other compiler tools
1097 do not fully support quoting of paths with spaces.
1099 * Allow installing to directories containing spaces
1101 This is possible if proper quoting is added to the makefiles for the
1102 install targets. Because PostgreSQL supports relocatable installs, it
1103 is already possible to install into a directory that doesn't contain
1104 spaces and then copy the install to a directory with spaces.
1106 * Fix sgmltools so PDFs can be generated with bookmarks
1107 * %Clean up compiler warnings (especially with gcc version 4)
1108 * Add function to return the thread safety status of libpq and ecpg
1109 * Use UTF8 encoding for NLS messages so all server encodings can
1111 * Update Bonjour to work with newer cross-platform SDK
1112 * Remove BeOS and QNX-specific code
1117 o Remove configure.in check for link failure when cause is found
1118 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1120 o Remove psql newline patch when we find out why mingw outputs an
1122 o Allow psql to use readline once non-US code pages work with
1124 o Re-enable timezone output on log_line_prefix '%t' when a
1125 shorter timezone string is available
1126 o Fix problem with shared memory on the Win32 Terminal Server
1127 o Improve signal handling,
1128 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1131 * Wire Protocol Changes
1133 o Allow dynamic character set handling
1134 o Add decoded type, length, precision
1136 o Update clients to use data types, typmod, schema.table.column names
1137 of result sets using new query protocol
1140 ---------------------------------------------------------------------------
1143 Developers who have claimed items are:
1144 --------------------------------------
1145 * Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
1146 * Andrew is Andrew Dunstan <andrew@dunslane.net>
1147 * Bruce is Bruce Momjian <pgman@candle.pha.pa.us> of Software Research Assoc.
1148 * Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
1149 Family Health Network
1150 * Claudio is Claudio Natoli <claudio.natoli@memetrics.com>
1151 * D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
1152 * Fabien is Fabien Coelho <coelho@cri.ensmp.fr>
1153 * Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
1154 * Greg is Greg Sabino Mullane <greg@turnstep.com>
1155 * Hiroshi is Hiroshi Inoue <Inoue@tpf.co.jp>
1156 * Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
1157 * Joe is Joe Conway <mail@joeconway.com>
1158 * Karel is Karel Zak <zakkr@zf.jcu.cz>
1159 * Magnus is Magnus Hagander <mha@sollentuna.net>
1160 * Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
1161 * Matthew T. O'Connor <matthew@zeut.net>
1162 * Michael is Michael Meskes <meskes@postgresql.org> of Credativ
1163 * Neil is Neil Conway <neilc@samurai.com>
1164 * Oleg is Oleg Bartunov <oleg@sai.msu.su>
1165 * Peter is Peter Eisentraut <peter_e@gmx.net>
1166 * Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
1167 * Rod is Rod Taylor <pg@rbt.ca>
1168 * Simon is Simon Riggs <simon@2ndquadrant.com>
1169 * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
1170 * Tatsuo is Tatsuo Ishii <t-ishii@sra.co.jp> of Software Research Assoc.
1171 * Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat