4 Current maintainer: Bruce Momjian (pgman@candle.pha.pa.us)
5 Last updated: Tue Sep 27 11:18:04 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.#
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 * Have sequence dependency track use of DEFAULT sequences,
186 * %Disallow changing default expression of a SERIAL column?
187 * Fix data types where equality comparison isn't intuitive, e.g. box
188 * %Prevent INET cast to CIDR if the unmasked bits are not zero, or
190 * %Prevent INET cast to CIDR from droping netmask, SELECT '1.1.1.1'::inet::cidr
191 * Allow INET + INT4 to increment the host part of the address, or
192 throw an error on overflow
193 * %Add 'tid != tid ' operator for use in corruption recovery
194 * Allow user-defined types to specify a type modifier at table creation
200 o Allow infinite dates just like infinite timestamps
201 o Add a GUC variable to allow output of interval values in ISO8601
203 o Merge hardwired timezone names with the TZ database; allow either
204 kind everywhere a TZ name is currently taken
205 o Allow customization of the known set of TZ names (generalize the
206 present australian_timezones hack)
207 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
208 information, either zone name or offset from UTC [timezone]
210 If the TIMESTAMP value is stored with a time zone name, interval
211 computations should adjust based on the time zone rules.
213 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
214 o Fix SELECT INTERVAL '1' MONTH
215 o Add ISO INTERVAL handling
216 o Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
218 o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
219 o For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
220 '1', treat as ISO if there is a range specification clause,
221 and as PG if there no clause is present, e.g. interpret
222 '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
223 interpret '1:30' as '1 hour, 30 minutes'
224 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
225 INTERVAL MONTH), and this should return '12 months'
226 o Round or truncate values to the requested precision, e.g.
227 INTERVAL '11 months' AS YEAR should return one or zero
228 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
233 o Allow NULLs in arrays
234 o Delay resolution of array expression's data type so assignment
235 coercion can be performed on empty array expressions
240 o Improve vacuum of large objects, like /contrib/vacuumlo?
241 o Add security checking for large objects
242 o Auto-delete large objects when referencing row is deleted
244 /contrib/lo offers this functionality.
246 o Allow read/write into TOAST values like large objects
248 This requires the TOAST column to be stored EXTERNAL.
254 * Allow INET subnet tests using non-constants to be indexed
255 * Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
258 Current CURRENT_TIMESTAMP returns the start time of the current
259 transaction, and gettimeofday() returns the wallclock time. This will
260 make time reporting more consistent and will allow reporting of
261 the statement start time.
263 * %Add pg_get_acldef(), pg_get_typedefault(), and pg_get_attrdef()
264 * Allow to_char() to print localized month names
265 * Allow functions to have a schema search path specified at creation time
266 * Allow substring/replace() to get/set bit values
267 * Allow to_char() on interval values to accumulate the highest unit
270 Some special format flag would be required to request such
271 accumulation. Such functionality could also be added to EXTRACT.
272 Prevent accumulation that crosses the month/day boundary because of
273 the uneven number of days in a month.
275 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
276 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
277 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
278 o to_char(INTERVAL '3 years 5 months','MM') => 41
280 * Add sleep() function, remove from regress.c
283 Multi-Language Support
284 ======================
286 * Add NCHAR (as distinguished from ordinary varchar),
287 * Allow locale to be set at database creation
289 Currently locale can only be set during initdb. No global tables have
290 locale-aware columns. However, the database template used during
291 database creation might have locale-aware indexes. The indexes would
292 need to be reindexed to match the new locale.
294 * Allow encoding on a per-column basis
296 Right now only one encoding is allowed per database.
298 * Support multiple simultaneous character sets, per SQL92
299 * Improve UTF8 combined character handling?
300 * Add octet_length_server() and octet_length_client()
301 * Make octet_length_client() the same as octet_length()?
302 * Fix problems with wrong runtime encoding conversion for NLS message files
308 * %Automatically create rules on views so they are updateable, per SQL99
310 We can only auto-create rules for simple views. For more complex
311 cases users will still have to write rules.
313 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
314 * Allow NOTIFY in rules involving conditionals
315 * Allow VIEW/RULE recompilation when the underlying tables change
317 Another issue is whether underlying table changes should be reflected
318 in the view, e.g. should SELECT * show additional columns if they
319 are added after the view is created.
325 * Change LIMIT/OFFSET and FETCH/MOVE to use int8
326 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
327 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
328 * %Allow SET CONSTRAINTS to be qualified by schema/table name
329 * %Allow TRUNCATE ... CASCADE/RESTRICT
331 This is like DELETE CASCADE, but truncates.
333 * %Add a separate TRUNCATE permission
335 Currently only the owner can TRUNCATE a table because triggers are not
336 called, and the table is locked in exclusive mode.
338 * Allow PREPARE of cursors
339 * Allow PREPARE to automatically determine parameter types based on the SQL
341 * Allow finer control over the caching of prepared query plans
343 Currently, queries prepared via the libpq API are planned on first
344 execute using the supplied parameters --- allow SQL PREPARE to do the
345 same. Also, allow control over replanning prepared queries either
346 manually or automatically when statistics for execute parameters
347 differ dramatically from those used during planning.
349 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
351 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
352 such information in memory would improve performance.
354 * Add optional textual message to NOTIFY
356 This would allow an informational message to be added to the notify
357 message, perhaps indicating the row modified or other custom
360 * Add a GUC variable to warn about non-standard SQL usage in queries
361 * Add MERGE command that does UPDATE/DELETE, or on failure, INSERT (rules,
363 * Add NOVICE output level for helpful messages like automatic sequence/index
365 * %Add COMMENT ON for all cluster global objects (roles, databases
367 * %Make row-wise comparisons work per SQL spec
368 * Add RESET CONNECTION command to reset all session state
370 This would include resetting of all variables (RESET ALL), dropping of
371 temporary tables, removing any NOTIFYs, cursors, open transactions,
372 prepared queries, currval()s, etc. This could be used for connection
373 pooling. We could also change RESET ALL to have this functionality.
374 The difficult of this features is allowing RESET ALL to not affect
375 changes made by the interface driver for its internal use. One idea
376 is for this to be a protocol-only feature. Another approach is to
377 notify the protocol when a RESET CONNECTION command is used.
379 * Add GUC to issue notice about queries that use unjoined tables
380 * Allow EXPLAIN to identify tables that were skipped because of
382 * Allow EXPLAIN output to be more easily processed by scripts
383 * Eventually enable escape_string_warning and standard_conforming_strings
384 * Simplify dropping roles that have objects in several databases
389 o Allow CREATE TABLE AS to determine column lengths for complex
390 expressions like SELECT col1 || col2
391 o Use more reliable method for CREATE DATABASE to get a consistent
393 o Add ON COMMIT capability to CREATE TABLE AS ... SELECT
397 o Allow UPDATE to handle complex aggregates [update]?
398 o Allow an alias to be provided for the target table in
401 This is not SQL-spec but many DBMSs allow it.
403 o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
409 o %Have ALTER TABLE RENAME rename SERIAL sequence names
410 o Add ALTER DOMAIN to modify the underlying data type
411 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
412 o %Allow ALTER TABLE to change constraint deferrability and actions
413 o Add missing object types for ALTER ... SET SCHEMA
414 o Allow ALTER TABLESPACE to move to different directories
415 o Allow databases to be moved to different tablespaces
416 o Allow moving system tables to other tablespaces, where possible
418 Currently non-global system tables must be in the default database
419 tablespace. Global system tables can never be moved.
421 o %Disallow dropping of an inherited constraint
422 o %Prevent child tables from altering or dropping constraints
423 like CHECK that were inherited from the parent table
428 o Automatically maintain clustering on a table
430 This might require some background daemon to maintain clustering
431 during periods of low usage. It might also require tables to be only
432 paritally filled for easier reorganization. Another idea would
433 be to create a merged heap/index data file so an index lookup would
434 automatically access the heap data too. A third idea would be to
435 store heap rows in hashed groups, perhaps using a user-supplied
438 o %Add default clustering to system tables
440 To do this, determine the ideal cluster index for each system
441 table and set the cluster setting during initdb.
446 o Allow COPY to report error lines and continue
448 This requires the use of a savepoint before each COPY line is
449 processed, with ROLLBACK on COPY failure.
451 o %Have COPY return the number of rows loaded/unloaded?
452 o Allow COPY on a newly-created table to skip WAL logging
454 On crash recovery, the table involved in the COPY would
455 be removed or have its heap and index files truncated. One
456 issue is that no other backend should be able to add to
457 the table at the same time, which is something that is
460 o Allow COPY to output from views
462 Another idea would be to allow actual SELECT queries in a COPY.
467 o Allow column-level privileges
468 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
471 The proposed syntax is:
472 GRANT SELECT ON ALL TABLES IN public TO phpuser;
473 GRANT SELECT ON NEW TABLES IN public TO phpuser;
475 * Allow GRANT/REVOKE permissions to be inherited by objects based on
481 o Allow UPDATE/DELETE WHERE CURRENT OF cursor
483 This requires using the row ctid to map cursor rows back to the
484 original heap row. This become more complicated if WITH HOLD cursors
485 are to be supported because WITH HOLD cursors have a copy of the row
486 and no FOR UPDATE lock.
488 o Prevent DROP TABLE from dropping a row referenced by its own open
491 o %Allow pooled connections to list all open WITH HOLD cursors
493 Because WITH HOLD cursors exist outside transactions, this allows
494 them to be listed so they can be closed.
499 o Allow INSERT/UPDATE of the system-generated oid value for a row
500 o Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
501 o Allow INSERT/UPDATE ... RETURNING new.col or old.col
503 This is useful for returning the auto-generated key for an INSERT.
504 One complication is how to handle rules that run as part of
510 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
512 o Add SET PATH for schemas?
514 This is basically the same as SET search_path.
517 * Server-Side Languages
519 o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
520 o Allow function parameters to be passed by name,
521 get_employee_salary(emp_id => 12345, tax_year => 2001)
522 o Add Oracle-style packages
523 o Add table function support to pltcl, plpython
524 o Add capability to create and call PROCEDURES
525 o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
526 o Allow function argument names to be queries from PL/PgSQL
527 o Add MOVE to PL/pgSQL
528 o Add support for polymorphic arguments and return types to
529 languages other than PL/PgSQL
530 o Add support for OUT and INOUT parameters to languages other
532 o Add single-step debugging of PL/PgSQL functions
533 o Allow PL/PgSQL to support WITH HOLD cursors
539 * Add a libpq function to support Parse/DescribeStatement capability
540 * Prevent libpq's PQfnumber() from lowercasing the column name?
541 * Add PQescapeIdentifier() to libpq
542 * Have initdb set the input DateStyle (MDY or DMY) based on locale?
543 * Have pg_ctl look at PGHOST in case it is a socket directory?
544 * Allow pg_ctl to work properly with configuration files located outside
547 pg_ctl can not read the pid file because it isn't located in the
548 config directory but in the PGDATA directory. The solution is to
549 allow pg_ctl to read and understand postgresql.conf to find the
550 data_directory value.
555 o Have psql show current values for a sequence
556 o Move psql backslash database information into the backend, use
557 mnemonic commands? [psql]
559 This would allow non-psql clients to pull the same information out
560 of the database as psql.
562 o Fix psql's display of schema information (Neil)
563 o Allow psql \pset boolean variables to set to fixed values, rather
565 o Consistently display privilege information for all objects in psql
566 o Improve psql's handling of multi-line queries
568 Currently, while \e saves a single query as one entry, interactive
569 queries are saved one line at a time. Ideally all queries
570 whould be saved like \e does.
572 o Allow multi-line column values to align in the proper columns
574 If the second output column value is 'a\nb', the 'b' should appear
575 in the second display column, rather than the first column as it
578 o Display IN, INOUT, and OUT parameters in \df+
580 It probably requires psql to output newlines in the proper
581 column, which is already on the TODO list.
586 o %Have pg_dump use multi-statement transactions for INSERT dumps
587 o %Allow pg_dump to use multiple -t and -n switches [pg_dump]
588 o %Add dumping of comments on composite type columns
589 o %Add dumping of comments on index columns
590 o %Replace crude DELETE FROM method of pg_dumpall --clean for
591 cleaning of roles with separate DROP commands
592 o Stop dumping CASCADE on DROP TYPE commands in clean mode
593 o %Add full object name to the tag field. eg. for operators we need
594 '=(integer, integer)', instead of just '='.
595 o Add pg_dumpall custom format dumps?
596 o %Add CSV output format
597 o Update pg_dump and psql to use the new COPY libpq API (Christopher)
598 o Remove unnecessary function pointer abstractions in pg_dump source
606 Document differences between ecpg and the SQL standard and
607 information about the Informix-compatibility module.
609 o Solve cardinality > 1 for input descriptors / variables?
610 o Add a semantic check level, e.g. check if a table really exists
611 o fix handling of DB attributes that are arrays
612 o Use backend PREPARE/EXECUTE facility for ecpg where possible
614 o Fix nested C comments
615 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
616 o Make SET CONNECTION thread-aware, non-standard?
617 o Allow multidimensional arrays
618 o Add internationalized message strings
621 Referential Integrity
622 =====================
624 * Add MATCH PARTIAL referential integrity
625 * Add deferred trigger queue file
627 Right now all deferred trigger information is stored in backend
628 memory. This could exhaust memory for very large trigger queues.
629 This item involves dumping large queues into files.
631 * Change foreign key constraint for array -> element to mean element
633 * Allow DEFERRABLE UNIQUE constraints?
634 * Allow triggers to be disabled in only the current session.
636 This is currently possible by starting a multi-statement transaction,
637 modifying the system tables, performing the desired SQL, restoring the
638 system tables, and committing the transaction. ALTER TABLE ...
639 TRIGGER requires a table lock so it is not ideal for this usage.
641 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
643 If the dump is known to be valid, allow foreign keys to be added
644 without revalidating the data.
646 * Allow statement-level triggers to access modified rows
647 * Support triggers on columns (Greg Sabino Mullane)
648 * Enforce referential integrity for system tables
649 * Allow AFTER triggers on system tables
651 System tables are modified in many places in the backend without going
652 through the executor and therefore not causing triggers to fire. To
653 complete this item, the functions that modify system tables will have
660 * Flush cached query plans when the dependent objects change
661 * Track dependencies in function bodies and recompile/invalidate
663 This is particularly important for references to temporary tables
664 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
665 in PL/PgSQL is to use EXECUTE. One complexity is that a function
666 might itself drop and recreate dependent tables, causing it to
667 invalidate its own query plan.
673 * Add SQL99 WITH clause to SELECT
674 * Add SQL99 WITH RECURSIVE to SELECT
675 * Add pre-parsing phase that converts non-ISO syntax to supported
678 This could allow SQL written for other databases to run without
681 * Allow plug-in modules to emulate features from other databases
682 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
684 * Allow queries across databases or servers with transaction
687 This can be done using dblink and two-phase commit.
689 * Add the features of packages
691 o Make private objects accessable only to objects in the same schema
692 o Allow current_schema.objname to access current schema objects
693 o Add session variables
694 o Allow nested schemas
700 * Allow inherited tables to inherit index, UNIQUE constraint, and primary
702 * UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
703 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
706 The main difficulty with this item is the problem of creating an index
707 that can span more than one table.
709 * Allow SELECT ... FOR UPDATE on inherited tables
710 * Add UNIQUE capability to non-btree indexes
711 * Prevent index uniqueness checks when UPDATE does not modify the column
713 Uniqueness (index) checks are done when updating a column even if the
714 column is not modified by the UPDATE.
716 * Allow the creation of on-disk bitmap indexes which can be quickly
717 combined with other bitmap indexes
719 Such indexes could be more compact if there are only a few distinct values.
720 Such indexes can also be compressed. Keeping such indexes updated can be
723 * Allow use of indexes to search for NULLs
725 One solution is to create a partial index on an IS NULL expression.
727 * Allow accurate statistics to be collected on indexes with more than
728 one column or expression indexes, perhaps using per-index statistics
729 * Add fillfactor to control reserved free space during index creation
730 * Allow the creation of indexes with mixed ascending/descending specifiers
731 * Allow constraint_exclusion to work for UNIONs like it does for
732 inheritance, allow it to work for UPDATE and DELETE queries, and allow
733 it to be used for all queries with little performance impact
734 * Allow CREATE INDEX to take an additional parameter for use with
741 o Add more GIST index support for geometric data types
742 o Allow GIST indexes to create certain complex index types, like
743 digital trees (see Aoki)
747 o Pack hash index buckets onto disk pages more efficiently
749 Currently only one hash bucket can be stored on a page. Ideally
750 several hash buckets could be stored on a single page and greater
751 granularity used for the hash algorithm.
753 o Consider sorting hash buckets so entries can be found using a
754 binary search, rather than a linear scan
756 o In hash indexes, consider storing the hash value with or instead
759 o Add WAL logging for crash recovery
760 o Allow multi-column hash indexes
766 * Improve commit_delay handling to reduce fsync()
767 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
769 Ideally this requires a separate test program that can be run
770 at initdb time or optionally later.
772 * %Add an option to sync() before fsync()'ing checkpoint files
773 * Add program to test if fsync has a delay compared to non-fsync
779 * Allow free-behind capability for large sequential scans, perhaps using
782 Posix_fadvise() can control both sequential/random file caching and
783 free-behind behavior, but it is unclear how the setting affects other
784 backends that also have the file open, and the feature is not supported
785 on all operating systems.
789 We could use a fixed row count and a +/- count to follow MVCC
790 visibility rules, or a single cached value could be used and
791 invalidated if anyone modifies the table. Another idea is to
792 get a count directly from a unique index, but for this to be
793 faster than a sequential scan it must avoid access to the heap
794 to obtain tuple visibility information.
796 * Allow data to be pulled directly from indexes
798 Currently indexes do not have enough tuple visibility information
799 to allow data to be pulled from the index without also accessing
800 the heap. One way to allow this is to set a bit to index tuples
801 to indicate if a tuple is currently visible to all transactions
802 when the first valid heap lookup happens. This bit would have to
803 be cleared when a heap tuple is expired.
806 * Consider automatic caching of queries at various levels:
812 * Allow sequential scans to take advantage of other concurrent
813 sequentiqal scans, also called "Synchronised Scanning"
815 One possible implementation is to start sequential scans from the lowest
816 numbered buffer in the shared cache, and when reaching the end wrap
817 around to the beginning, rather than always starting sequential scans
818 at the start of the table.
824 * Improve speed with indexes
826 For large table adjustements during VACUUM FULL, it is faster to
827 reindex rather than update the index.
829 * Reduce lock time during VACUUM FULL by moving tuples with read lock,
830 then write lock and truncate table
832 Moved tuples are invisible to other backends so they don't require a
833 write lock. However, the read lock promotion to write lock could lead
834 to deadlock situations.
836 * Auto-fill the free space map by scanning the buffer cache or by
837 checking pages written by the background writer
838 * Create a bitmap of pages that need vacuuming
840 Instead of sequentially scanning the entire table, have the background
841 writer or some other process record pages that have expired rows, then
842 VACUUM can look at just those pages rather than the entire table. In
843 the event of a system crash, the bitmap would probably be invalidated.
844 One complexity is that index entries still have to be vacuumed, and
845 doing this without an index scan (by using the heap values to find the
846 index entry) might be slow and unreliable, especially for user-defined
849 * %Add system view to show free space map contents
854 o Use free-space map information to guide refilling
855 o %Issue log message to suggest VACUUM FULL if a table is nearly
857 o Improve xid wraparound detection by recording per-table rather
864 * Add code to detect an SMP machine and handle spinlocks accordingly
865 from distributted.net, http://www1.distributed.net/source,
866 in client/common/cpucheck.cpp
868 On SMP machines, it is possible that locks might be released shortly,
869 while on non-SMP machines, the backend should sleep so the process
870 holding the lock can complete and release it.
872 * Fix priority ordering of read and write light-weight locks (Neil)
875 Startup Time Improvements
876 =========================
878 * Experiment with multi-threaded backend [thread]
880 This would prevent the overhead associated with process creation. Most
881 operating systems have trivial process creation time compared to
882 database startup overhead, but a few operating systems (WIn32,
883 Solaris) might benefit from threading. Also explore the idea of
884 a single session using multiple threads to execute a query faster.
886 * Add connection pooling
888 It is unclear if this should be done inside the backend code or done
889 by something external like pgpool. The passing of file descriptors to
890 existing backends is one of the difficulties with a backend approach.
896 * Eliminate need to write full pages to WAL before page modification [wal]
898 Currently, to protect against partial disk page writes, we write
899 full page images to WAL before they are modified so we can correct any
900 partial page writes during recovery. These pages can also be
901 eliminated from point-in-time archive files.
903 o When off, write CRC to WAL and check file system blocks
906 If CRC check fails during recovery, remember the page in case
907 a later CRC for that page properly matches.
909 o Write full pages during file system write and not when
910 the page is modified in the buffer cache
912 This allows most full page writes to happen in the background
913 writer. It might cause problems for applying WAL on recovery
914 into a partially-written page, but later the full page will be
917 * Allow WAL traffic to be streamed to another server for stand-by
919 * Reduce WAL traffic so only modified values are written rather than
921 * Allow the pg_xlog directory location to be specified during initdb
922 with a symlink back to the /data location
923 * Allow WAL information to recover corrupted pg_controldata
924 * Find a way to reduce rotational delay when repeatedly writing
927 Currently fsync of WAL requires the disk platter to perform a full
928 rotation to fsync again. One idea is to write the WAL to different
929 offsets that might reduce the rotational delay.
931 * Allow buffered WAL writes and fsync
933 Instead of guaranteeing recovery of all committed transactions, this
934 would provide improved performance by delaying WAL writes and fsync
935 so an abrupt operating system restart might lose a few seconds of
936 committed transactions but still be consistent. We could perhaps
937 remove the 'fsync' parameter (which results in an an inconsistent
938 database) in favor of this capability.
944 * Add missing optimizer selectivities for date, r-tree, etc
945 * Allow ORDER BY ... LIMIT # to select high/low value without sort or
946 index using a sequential scan for highest/lowest values
948 Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
949 all values to return the high/low value. Instead The idea is to do a
950 sequential scan to find the high/low value, thus avoiding the sort.
951 MIN/MAX already does this, but not for LIMIT > 1.
953 * Precompile SQL functions to avoid overhead
954 * Create utility to compute accurate random_page_cost value
955 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
956 * Have EXPLAIN ANALYZE highlight poor optimizer estimates
957 * Consider using hash buckets to do DISTINCT, rather than sorting
959 This would be beneficial when there are few distinct values. This is
960 already used by GROUP BY.
962 * Log queries where the optimizer row estimates were dramatically
963 different from the number of rows actually found?
966 Miscellaneous Performance
967 =========================
969 * Do async I/O for faster random read-ahead of data
971 Async I/O allows multiple I/O requests to be sent to the disk with
972 results coming back asynchronously.
974 * Use mmap() rather than SYSV shared memory or to write WAL files?
976 This would remove the requirement for SYSV SHM but would introduce
977 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
978 to prevent I/O overhead.
980 * Consider mmap()'ing files into a backend?
982 Doing I/O to large tables would consume a lot of address space or
983 require frequent mapping/unmapping. Extending the file also causes
984 mapping problems that might require mapping only individual pages,
985 leading to thousands of mappings. Another problem is that there is no
986 way to _prevent_ I/O to disk from the dirty shared buffers so changes
987 could hit disk before WAL is written.
989 * Add a script to ask system configuration questions and tune postgresql.conf
990 * Merge xmin/xmax/cmin/cmax back into three header fields
992 Before subtransactions, there used to be only three fields needed to
993 store these four values. This was possible because only the current
994 transaction looks at the cmin/cmax values. If the current transaction
995 created and expired the row the fields stored where xmin (same as
996 xmax), cmin, cmax, and if the transaction was expiring a row from a
997 another transaction, the fields stored were xmin (cmin was not
998 needed), xmax, and cmax. Such a system worked because a transaction
999 could only see rows from another completed transaction. However,
1000 subtransactions can see rows from outer transactions, and once the
1001 subtransaction completes, the outer transaction continues, requiring
1002 the storage of all four fields. With subtransactions, an outer
1003 transaction can create a row, a subtransaction expire it, and when the
1004 subtransaction completes, the outer transaction still has to have
1005 proper visibility of the row's cmin, for example, for cursors.
1007 One possible solution is to create a phantom cid which represents a
1008 cmin/cmax pair and is stored in local memory. Another idea is to
1009 store both cmin and cmax only in local memory.
1011 * Research storing disk pages with no alignment/padding
1017 * Add use of 'const' for variables in source tree
1018 * Rename some /contrib modules from pg* to pg_*
1019 * Move some things from /contrib into main tree
1020 * Move some /contrib modules out to their own project sites
1021 * %Remove warnings created by -Wcast-align
1022 * Move platform-specific ps status display info from ps_status.c to ports
1023 * Add optional CRC checksum to heap and index pages
1024 * Improve documentation to build only interfaces (Marc)
1025 * Remove or relicense modules that are not under the BSD license, if possible
1026 * %Remove memory/file descriptor freeing before ereport(ERROR)
1027 * Acquire lock on a relation before building a relcache entry for it
1028 * %Promote debug_query_string into a server-side function current_query()
1029 * %Allow the identifier length to be increased via a configure option
1030 * Remove Win32 rename/unlink looping if unnecessary
1031 * Allow cross-compiling by generating the zic database on the target system
1032 * Improve NLS maintenace of libpgport messages linked onto applications
1033 * Allow ecpg to work with MSVC and BCC
1034 * Add xpath_array() to /contrib/xml2 to return results as an array
1035 * Allow building in directories containing spaces
1037 This is probably not possible because 'gmake' and other compiler tools
1038 do not fully support quoting of paths with spaces.
1040 * Allow installing to directories containing spaces
1042 This is possible if proper quoting is added to the makefiles for the
1043 install targets. Because PostgreSQL supports relocatable installs, it
1044 is already possible to install into a directory that doesn't contain
1045 spaces and then copy the install to a directory with spaces.
1047 * Fix sgmltools so PDFs can be generated with bookmarks
1048 * %Clean up compiler warnings (especially with gcc version 4)
1049 * Add function to return the thread safety status of libpq and ecpg
1050 * Use UTF8 encoding for NLS messages so all server encodings can
1052 * Update Bonjour to work with newer cross-platform SDK
1053 * Remove BeOS and QNX-specific code
1054 * +Add options to pg_config to show the share_dir, sysconfdir,
1055 pkgincludedir, and localedir
1060 o Remove configure.in check for link failure when cause is found
1061 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1063 o Remove psql newline patch when we find out why mingw outputs an
1065 o Allow psql to use readline once non-US code pages work with
1067 o Re-enable timezone output on log_line_prefix '%t' when a
1068 shorter timezone string is available
1069 o Fix problem with shared memory on the Win32 Terminal Server
1070 o Improve signal handling,
1071 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1074 * Wire Protocol Changes
1076 o Allow dynamic character set handling
1077 o Add decoded type, length, precision
1079 o Update clients to use data types, typmod, schema.table.column names
1080 of result sets using new query protocol
1083 ---------------------------------------------------------------------------
1086 Developers who have claimed items are:
1087 --------------------------------------
1088 * Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
1089 * Andrew is Andrew Dunstan <andrew@dunslane.net>
1090 * Bruce is Bruce Momjian <pgman@candle.pha.pa.us> of Software Research Assoc.
1091 * Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
1092 Family Health Network
1093 * Claudio is Claudio Natoli <claudio.natoli@memetrics.com>
1094 * D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
1095 * Fabien is Fabien Coelho <coelho@cri.ensmp.fr>
1096 * Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
1097 * Greg is Greg Sabino Mullane <greg@turnstep.com>
1098 * Hiroshi is Hiroshi Inoue <Inoue@tpf.co.jp>
1099 * Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
1100 * Joe is Joe Conway <mail@joeconway.com>
1101 * Karel is Karel Zak <zakkr@zf.jcu.cz>
1102 * Magnus is Magnus Hagander <mha@sollentuna.net>
1103 * Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
1104 * Matthew T. O'Connor <matthew@zeut.net>
1105 * Michael is Michael Meskes <meskes@postgresql.org> of Credativ
1106 * Neil is Neil Conway <neilc@samurai.com>
1107 * Oleg is Oleg Bartunov <oleg@sai.msu.su>
1108 * Peter is Peter Eisentraut <peter_e@gmx.net>
1109 * Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
1110 * Rod is Rod Taylor <pg@rbt.ca>
1111 * Simon is Simon Riggs <simon@2ndquadrant.com>
1112 * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
1113 * Tatsuo is Tatsuo Ishii <t-ishii@sra.co.jp> of Software Research Assoc.
1114 * Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat