3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Tue Mar 11 17:24:58 EDT 2008
6 The most recent version of this document can be viewed at
7 http://www.postgresql.org/docs/faqs.TODO.html.
9 #A hyphen, "-", marks changes that will appear in the upcoming 8.3 release.#
10 #A percent sign, "%", marks items that are easier to implement.#
12 Bracketed items, "[]", have more detail.
14 This list contains all known PostgreSQL bugs and feature requests. If
15 you would like to work on an item, please read the Developer's FAQ
16 first. There is also a developer's wiki at
17 http://developer.postgresql.org.
23 * Allow administrators to safely terminate individual sessions either
24 via an SQL function or SIGTERM
26 Lock table corruption following SIGTERM of an individual backend
27 has been reported in 8.0. A possible cause was fixed in 8.1, but
28 it is unknown whether other problems exist. This item mostly
29 requires additional testing rather than of writing any new code.
31 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00174.php
32 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00218.php
34 * Check for unreferenced table files created by transactions that were
35 in-progress when the server terminated abruptly
37 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
39 * Set proper permissions on non-system schemas during db creation
41 Currently all schemas are owned by the super-user because they are copied
42 from the template1 database. However, since all objects are inherited
43 from the template database, it is not clear that setting schemas to the db
46 * Add function to report the time of the most recent server reload
47 * Allow statistics collector information to be pulled from the collector
48 process directly, rather than requiring the collector to write a
49 filesystem file twice a second?
50 * Allow statistics last vacuum/analyze execution times to be displayed
51 without requiring stats_row_level to be enabled
52 * Allow log_min_messages to be specified on a per-module basis
54 This would allow administrators to see more detailed information from
55 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
56 Another idea is to allow separate configuration files for each module,
57 or allow arbitrary SET commands to be passed to them.
59 * Simplify ability to create partitioned tables
61 This would allow creation of partitioned tables without requiring
62 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
63 for rapid partition selection. Options could include range and hash
66 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
67 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
69 * Allow auto-selection of partitioned tables for min/max() operations
70 * Allow more complex user/database default GUC settings
72 Currently ALTER USER and ALTER DATABASE support per-user and
73 per-database defaults. Consider adding per-user-and-database
74 defaults so things like search_path can be defaulted for a
75 specific user connecting to a specific database.
77 * Allow custom variable classes that can restrict who can set the values
79 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
81 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
82 the privilege granted by the invoking role, and not those granted
85 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
87 * Allow SSL authentication/encryption over unix domain sockets
89 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
91 * Allow client certificate names to be checked against the client
94 This is already implemented in
95 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
100 o Allow pg_hba.conf to specify host names along with IP addresses
102 Host name lookup could occur when the postmaster reads the
103 pg_hba.conf file, or when the backend starts. Another
104 solution would be to reverse lookup the connection IP and
105 check that hostname against the host names in pg_hba.conf.
106 We could also then check that the host name maps to the IP
109 o %Allow postgresql.conf file values to be changed via an SQL
110 API, perhaps using SET GLOBAL
111 o Allow the server to be stopped/restarted via an SQL API
112 o Issue a warning if a change-on-restart-only postgresql.conf value
113 is modified and the server config files are reloaded
114 o Consider normalizing fractions in postgresql.conf, perhaps
117 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00550.php
122 o Allow a database in tablespace t1 with tables created in
123 tablespace t2 to be used as a template for a new database created
124 with default tablespace t2
126 Currently all objects in the default database tablespace must
127 have default tablespace specifications. This is because new
128 databases are created by copying directories. If you mix default
129 tablespace tables and tablespace-specified tables in the same
130 directory, creating a new database from such a mixed directory
131 would create a new database with tables that had incorrect
132 explicit tablespaces. To fix this would require modifying
133 pg_class in the newly copied database, which we don't currently
136 o Allow reporting of which objects are in which tablespaces
138 This item is difficult because a tablespace can contain objects
139 from multiple databases. There is a server-side function that
140 returns the databases which use a specific tablespace, so this
141 requires a tool that will call that function and connect to each
142 database to find the objects in each database for that tablespace.
144 o Allow WAL replay of CREATE TABLESPACE to work when the directory
145 structure on the recovery computer is different from the original
147 o Allow per-tablespace quotas
150 * Point-In-Time Recovery (PITR)
152 o Allow a warm standby system to also allow read-only statements
155 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
157 o %Create dump tool for write-ahead logs for use in determining
158 transaction id for point-in-time recovery
160 This is useful for checking PITR recovery.
162 o Allow recovery.conf to support the same syntax as
163 postgresql.conf, including quoting
165 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
171 * Change NUMERIC to enforce the maximum precision
172 * Reduce storage space for small NUMERICs
174 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
175 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
176 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00715.php
178 * Fix data types where equality comparison isn't intuitive, e.g. box
179 * Add support for public SYNONYMs
181 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
183 * Fix CREATE CAST on DOMAINs
185 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
186 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
188 * Add support for SQL-standard GENERATED/IDENTITY columns
190 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
191 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
192 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
193 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
194 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
196 * Improve XML support
198 http://developer.postgresql.org/index.php/XML_Support
200 * Consider placing all sequences in a single table, or create a system
203 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
205 * Allow the UUID type to accept non-standard formats
207 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
209 * Allow text search dictionary to filter out only stop words
211 http://archives.postgresql.org/pgsql-patches/2007-11/msg00081.php
213 * Consider a function-based API for '@@' full text searches
215 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00511.php
220 o Allow infinite dates and intervals just like infinite timestamps
221 o Merge hardwired timezone names with the TZ database; allow either
222 kind everywhere a TZ name is currently taken
223 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
224 information, either zone name or offset from UTC [timezone]
226 If the TIMESTAMP value is stored with a time zone name, interval
227 computations should adjust based on the time zone rules.
229 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
230 o Add a GUC variable to allow output of interval values in ISO8601
232 o Have timestamp subtraction not call justify_hours()?
234 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
236 o Improve timestamptz subtraction to be DST-aware
238 Currently subtracting one date from another that crosses a
239 daylight savings time adjustment can return '1 day 1 hour', but
240 adding that back to the first date returns a time one hour in
241 the future. This is caused by the adjustment of '25 hours' to
242 '1 day 1 hour', and '1 day' is the same time the next day, even
243 if daylight savings adjustments are involved.
245 o Fix interval display to support values exceeding 2^31 hours
246 o Add overflow checking to timestamp and interval arithmetic
247 o Extend timezone code to allow 64-bit values so we can
248 represent years beyond 2038
250 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
252 o Use LC_TIME for localized weekday/month names, rather than
255 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
257 o Add ISO INTERVAL handling
259 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
260 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
262 o Support ISO INTERVAL syntax if units cannot be determined from
263 the string, and are supplied after the string
265 The SQL standard states that the units after the string
266 specify the units of the string, e.g. INTERVAL '2' MINUTE
267 should return '00:02:00'. The current behavior has the units
268 restrict the interval value to the specified unit or unit
269 range, INTERVAL '70' SECOND returns '00:00:10'.
271 For syntax that isn't uniquely ISO or PG syntax, like '1' or
272 '1:30', treat as ISO if there is a range specification clause,
273 and as PG if there no clause is present, e.g. interpret '1:30'
274 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
275 '1:30' as '1 hour, 30 minutes'.
277 This makes common cases like SELECT INTERVAL '1' MONTH
278 SQL-standard results. The SQL standard supports a limited
279 number of unit combinations and doesn't support unit names in
280 the string. The PostgreSQL syntax is more flexible in the
281 range of units supported, e.g. PostgreSQL supports '1 year 1
282 hour', while the SQL standard does not.
284 o Add support for year-month syntax, INTERVAL '50-6' YEAR
286 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
287 year' AS INTERVAL MONTH), and this should return '12 months'
288 o Round or truncate values to the requested precision, e.g.
289 INTERVAL '11 months' AS YEAR should return one or zero
290 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
295 o Delay resolution of array expression's data type so assignment
296 coercion can be performed on empty array expressions
297 o Add support for arrays of domains
299 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
301 o Allow single-byte header storage for array elements
306 o Improve vacuum of large objects, like contrib/vacuumlo?
307 o Add security checking for large objects
308 o Auto-delete large objects when referencing row is deleted
310 contrib/lo offers this functionality.
312 o Allow read/write into TOAST values like large objects
314 This requires the TOAST column to be stored EXTERNAL.
316 o Add API for 64-bit large object access
318 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
322 * Add locale-aware MONEY type, and support multiple currencies
324 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
325 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
327 * MONEY dumps in a locale-specific format making it difficult to
328 restore to a system with a different locale
329 * Allow MONEY to be easily cast to/from other numeric data types
335 * Allow INET subnet tests using non-constants to be indexed
336 * Allow to_date() and to_timestamp() accept localized month names
337 * Fix to_date()-related functions to consistently issue errors
339 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
341 * Add missing parameter handling in to_char()
343 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
345 * Allow substring/replace() to get/set bit values
346 * Allow to_char() on interval values to accumulate the highest unit
349 Some special format flag would be required to request such
350 accumulation. Such functionality could also be added to EXTRACT.
351 Prevent accumulation that crosses the month/day boundary because of
352 the uneven number of days in a month.
354 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
355 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
356 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
357 o to_char(INTERVAL '3 years 5 months','MM') => 41
359 * Implement inlining of set-returning functions defined in SQL
360 * Allow SQL-language functions to return results from RETURNING queries
362 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
364 * Allow SQL-language functions to reference parameters by parameter name
366 Currently SQL-language functions can only refer to dollar parameters,
369 * Add SPI_gettypmod() to return the typemod for a TupleDesc
370 * Enforce typmod for function inputs, function results and parameters for
371 spi_prepare'd statements called from PLs
373 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
375 * Allow holdable cursors in SPI
376 * Tighten function permission checks
378 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
380 * Fix IS OF so it matches the ISO specification, and add documentation
382 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
383 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
385 * Add missing operators for geometric data types
387 Some geometric types do not have the full suite of geometric operators,
390 * Implement Boyer-Moore searching in strpos()
392 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
394 * Prevent malicious functions from being executed with the permissions
395 of unsuspecting users
397 Index functions are safe, so VACUUM and ANALYZE are safe too.
398 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
399 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
401 * Reduce memory usage of aggregates in set returning functions
403 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
405 * Add temporal versions of generate_series()
407 http://archives.postgresql.org/pgsql-hackers/2007-04/msg01180.php
411 Multi-Language Support
412 ======================
414 * Add NCHAR (as distinguished from ordinary varchar),
415 * Allow locale to be set at database creation
417 Currently locale can only be set during initdb. No global tables have
418 locale-aware columns. However, the database template used during
419 database creation might have locale-aware indexes. The indexes would
420 need to be reindexed to match the new locale.
422 * Allow encoding on a per-column basis optionally using the ICU library:
424 Right now only one encoding is allowed per database. [locale]
426 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
427 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
428 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
429 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
431 * Add CREATE COLLATE? [locale]
432 * Support multiple simultaneous character sets, per SQL92
433 * Improve UTF8 combined character handling?
434 * Add octet_length_server() and octet_length_client()
435 * Make octet_length_client() the same as octet_length()?
436 * Fix problems with wrong runtime encoding conversion for NLS message files
437 * Add URL to more complete multi-byte regression tests
439 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
441 * Fix ILIKE and regular expressions to handle case insensitivity
442 properly in multibyte encodings
444 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
445 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
447 * Set client encoding based on the client operating system encoding
449 Currently client_encoding is set in postgresql.conf, which
450 defaults to the server encoding.
451 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
453 * Change memory allocation for multi-byte functions so memory is
454 allocated inside conversion functions
456 Currently we preallocate memory based on worst-case usage.
463 * Automatically create rules on views so they are updateable, per SQL99
465 We can only auto-create rules for simple views. For more complex
466 cases users will still have to write rules manually.
468 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
469 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
471 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
472 * Allow NOTIFY in rules involving conditionals
473 * Allow VIEW/RULE recompilation when the underlying tables change
475 Another issue is whether underlying table changes should be reflected
476 in the view, e.g. should SELECT * show additional columns if they
477 are added after the view is created.
479 * Make it possible to use RETURNING together with conditional DO INSTEAD
480 rules, such as for partitioning setups
482 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
484 * Add the ability to automatically create materialized views
486 Right now materialized views require the user to create triggers on the
487 main table to keep the summary table current. SQL syntax should be able
488 to manager the triggers and summary table automatically. A more
489 sophisticated implementation would automatically retrieve from the
490 summary table when the main table is referenced, if possible.
497 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
498 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
499 * %Allow SET CONSTRAINTS to be qualified by schema/table name
500 * %Add a separate TRUNCATE permission
502 Currently only the owner can TRUNCATE a table because triggers are not
503 called, and the table is locked in exclusive mode.
505 * Allow PREPARE of cursors
506 * Allow finer control over the caching of prepared query plans
508 Currently queries prepared via the libpq API are planned on first
509 execute using the supplied parameters --- allow SQL PREPARE to do the
510 same. Also, allow control over replanning prepared queries either
511 manually or automatically when statistics for execute parameters
512 differ dramatically from those used during planning.
514 * Improve logging of prepared transactions recovered during startup
516 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
518 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
520 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
521 such information in memory would improve performance.
523 * Add optional textual message to NOTIFY
525 This would allow an informational message to be added to the notify
526 message, perhaps indicating the row modified or other custom
529 * Allow multiple identical NOTIFY events to always be communicated to the
530 client, rather than sent as a single notification to the listener
532 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
534 * Add a GUC variable to warn about non-standard SQL usage in queries
535 * Add SQL-standard MERGE command, typically used to merge two tables
538 This is similar to UPDATE, then for unmatched rows, INSERT.
539 Whether concurrent access allows modifications which could cause
540 row loss is implementation independent.
542 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
545 To implement this cleanly requires that the table have a unique index
546 so duplicate checking can be easily performed. It is possible to
547 do it without a unique index if we require the user to LOCK the table
550 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
551 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
553 * Add NOVICE output level for helpful messages like automatic sequence/index
555 * Add GUC to issue notice about statements that use unjoined tables
556 * Allow EXPLAIN to identify tables that were skipped because of
558 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
559 * Enable standard_conforming_strings
560 * Make standard_conforming_strings the default in 8.5?
562 When this is done, backslash-quote should be prohibited in non-E''
563 strings because of possible confusion over how such strings treat
564 backslashes. Basically, '' is always safe for a literal single
565 quote, while \' might or might not be based on the backslash
568 * Simplify dropping roles that have objects in several databases
569 * Allow COMMENT ON to accept an expression rather than just a string
570 * Allow the count returned by SELECT, etc to be represented as an int64
571 to allow a higher range of values
572 * Add SQL99 WITH clause to SELECT
573 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
575 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
576 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
577 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
579 * Add DEFAULT .. AS OWNER so permission checks are done as the table
582 This would be useful for SERIAL nextval() calls and CHECK constraints.
584 * Allow DISTINCT to work in multiple-argument aggregate calls
585 * Add column to pg_stat_activity that shows the progress of long-running
586 commands like CREATE INDEX and VACUUM
587 * Implement SQL:2003 window functions
588 * Improve failure message when DROP DATABASE is used on a database that
589 has prepared transactions
590 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
592 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
593 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
595 * Increase locking when DROPing objects so dependent objects cannot
596 get dropped while the DROP operation is happening
598 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
600 * -Allow AS in "SELECT col AS label" to be optional in certain cases
602 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
603 RETURNING to supply values to the INSERT
604 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
606 * Add comments on system tables/columns using the information in
609 Ideally the information would be pulled from the SGML file
615 o Allow CREATE TABLE AS to determine column lengths for complex
616 expressions like SELECT col1 || col2
618 o Have WITH CONSTRAINTS also create constraint indexes
620 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
622 o Have CONSTRAINT cname NOT NULL record the contraint name
624 Right now pg_attribute.attnotnull records the NOT NULL status
625 of the column, but does not record the contraint name
630 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
632 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
633 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
635 o Research self-referential UPDATEs that see inconsistent row versions
636 in read-committed mode
638 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
639 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
641 o Allow GLOBAL temporary tables to exist as empty by default in
644 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
649 o Have ALTER TABLE RENAME rename SERIAL sequence names
651 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
653 o Have ALTER SEQUENCE RENAME rename the sequence name stored
654 in the sequence table
656 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
657 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
658 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
660 o Add ALTER DOMAIN to modify the underlying data type
661 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
663 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
665 o %Allow ALTER TABLE to change constraint deferrability and actions
666 o Add missing object types for ALTER ... SET SCHEMA
667 o Allow ALTER TABLESPACE to move to different directories
668 o Allow databases to be moved to different tablespaces
669 o Allow moving system tables to other tablespaces, where possible
671 Currently non-global system tables must be in the default database
672 tablespace. Global system tables can never be moved.
674 o Prevent parent tables from altering or dropping constraints
675 like CHECK that are inherited by child tables unless CASCADE
677 o %Prevent child tables from altering or dropping constraints
678 like CHECK that were inherited from the parent table
679 o Have ALTER INDEX update the name of a constraint using that index
680 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
681 o Allow column display reordering by recording a display,
682 storage, and permanent id for every column?
684 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
690 o Automatically maintain clustering on a table
692 This might require some background daemon to maintain clustering
693 during periods of low usage. It might also require tables to be only
694 partially filled for easier reorganization. Another idea would
695 be to create a merged heap/index data file so an index lookup would
696 automatically access the heap data too. A third idea would be to
697 store heap rows in hashed groups, perhaps using a user-supplied
699 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
701 o %Add default clustering to system tables
703 To do this, determine the ideal cluster index for each system
704 table and set the cluster setting during initdb.
706 o %Add VERBOSE option to report tables as they are processed,
712 o Allow COPY to report error lines and continue
714 This requires the use of a savepoint before each COPY line is
715 processed, with ROLLBACK on COPY failure.
717 o Allow COPY on a newly-created table to skip WAL logging
719 On crash recovery, the table involved in the COPY would
720 be removed or have its heap and index files truncated. One
721 issue is that no other backend should be able to add to
722 the table at the same time, which is something that is
723 currently allowed. This currently is done if the table is
724 created inside the same transaction block as the COPY because
725 no other backends can see the table.
727 o Consider using a ring buffer for COPY FROM
729 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
730 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
732 o Allow COPY FROM to create index entries in bulk
734 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
740 o Allow column-level privileges
741 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
744 The proposed syntax is:
745 GRANT SELECT ON ALL TABLES IN public TO phpuser;
746 GRANT SELECT ON NEW TABLES IN public TO phpuser;
748 o Allow GRANT/REVOKE permissions to be inherited by objects based on
751 o Allow SERIAL sequences to inherit permissions from the base table?
756 o Prevent DROP TABLE from dropping a row referenced by its own open
762 o Allow INSERT/UPDATE of the system-generated oid value for a row
763 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
769 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
774 Referential Integrity
775 =====================
777 * Add MATCH PARTIAL referential integrity
778 * Change foreign key constraint for array -> element to mean element
780 * Fix problem when cascading referential triggers make changes on
781 cascaded tables, seeing the tables in an intermediate state
783 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
784 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
786 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
788 This would allow UPDATE tab SET col = col + 1 to work if col has
789 a unique index. Currently, uniqueness checks are done while the
790 command is being executed, rather than at the end of the statement
792 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
793 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
795 * Optimize referential integrity checks
797 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
798 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00744.php
801 Server-Side Languages
802 =====================
805 o Fix RENAME to work on variables other than OLD/NEW
807 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
808 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
809 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
811 o Allow function parameters to be passed by name,
812 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
813 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
814 o Allow listing of record column names, and access to
815 record columns via variables, e.g. columns := r.(*),
818 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
819 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
820 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
822 o Add support for SCROLL cursors
823 o Add support for WITH HOLD cursors
824 o Allow row and record variables to be set to NULL constants,
825 and allow NULL tests on such variables
827 Because a row is not scalar, do not allow assignment
828 from NULL-valued scalars.
830 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
832 o Review handling of MOVE and FETCH
834 http://archives.postgresql.org/pgsql-patches/2007-04/msg00527.php
838 o Add table function support to pltcl, plpythonu
839 o Add support for polymorphic arguments and return types to
840 languages other than PL/PgSQL
841 o Add capability to create and call PROCEDURES
842 o Add support for OUT and INOUT parameters to languages other
844 o Add PL/PythonU tracebacks
846 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
848 o Allow data to be passed in native language formats, rather
851 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
858 * Have pg_ctl look at PGHOST in case it is a socket directory?
859 * Allow pg_ctl to work properly with configuration files located outside
862 pg_ctl can not read the pid file because it isn't located in the
863 config directory but in the PGDATA directory. The solution is to
864 allow pg_ctl to read and understand postgresql.conf to find the
865 data_directory value.
867 * Add a function like pg_get_indexdef() that report more detailed index
870 http://archives.postgresql.org/pgsql-bugs/2007-12/msg00166.php
872 * Prefix command-line utilities like createuser with 'pg_'
874 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00025.php
878 o Have psql show current values for a sequence
879 o Move psql backslash database information into the backend, use
880 mnemonic commands? [psql]
882 This would allow non-psql clients to pull the same information out
883 of the database as psql.
885 o Fix psql's \d commands more consistent
887 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
888 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
890 o Consistently display privilege information for all objects in psql
891 o Add auto-expanded mode so expanded output is used if the row
892 length is wider than the screen width.
894 Consider using auto-expanded mode for backslash commands like \df+.
896 o Prevent tab completion of SET TRANSACTION from querying the
897 database and therefore preventing the transaction isolation
898 level from being set.
900 Currently SET <tab> causes a database lookup to check all
901 supported session variables. This query causes problems
902 because setting the transaction isolation level must be the
903 first statement of a transaction.
905 o Add a \set variable to control whether \s displays line numbers
907 Another option is to add \# which lists line numbers, and
908 allows command execution.
910 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
912 o Prevent escape string warnings when object names have
915 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
917 o Have \d show foreign keys that reference a table's primary key
919 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
921 o Have \d show child tables that inherit from the specified parent
922 o Have \l+ show database size, if permissions allow
924 Ideally it will not generate an error for invalid permissions
928 * pg_dump / pg_restore
929 o %Add dumping of comments on index columns and composite type columns
930 o %Add full object name to the tag field. eg. for operators we need
931 '=(integer, integer)', instead of just '='.
932 o Add pg_dumpall custom format dumps?
933 o Allow selection of individual object(s) of all types, not just
935 o In a selective dump, allow dumping of an object and all its
937 o Add options like pg_restore -l and -L to pg_dump
938 o Stop dumping CASCADE on DROP TYPE commands in clean mode
939 o Allow pg_dump --clean to drop roles that own objects or have
941 o Change pg_dump so that a comment on the dumped database is
942 applied to the loaded database, even if the database has a
943 different name. This will require new backend syntax, perhaps
944 COMMENT ON CURRENT DATABASE.
945 o Remove unnecessary function pointer abstractions in pg_dump source
947 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
948 multiple objects simultaneously
950 The difficulty with this is getting multiple dump processes to
951 produce a single dump output file. It also would require
952 several sessions to share the same snapshot.
953 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
955 o Allow pg_restore to utilize multiple CPUs and I/O channels by
956 restoring multiple objects simultaneously
958 This might require a pg_restore flag to indicate how many
959 simultaneous operations should be performed. Only pg_dump's
960 -Fc format has the necessary dependency information.
961 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
963 o To better utilize resources, allow pg_restore to check foreign
964 keys simultaneously, where possible
965 o Allow pg_restore to create all indexes of a table
966 concurrently, via a single heap scan
968 This requires a pg_dump -Fc file because that format contains
969 the required dependency information.
970 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
972 o Allow pg_restore to load different parts of the COPY data
974 o Prevent pg_dump/pg_restore from being affected by
977 Using psql to restore a pg_dump dump is also affected.
983 Document differences between ecpg and the SQL standard and
984 information about the Informix-compatibility module.
986 o Solve cardinality > 1 for input descriptors / variables?
987 o Add a semantic check level, e.g. check if a table really exists
988 o fix handling of DB attributes that are arrays
989 o Use backend PREPARE/EXECUTE facility for ecpg where possible
991 o Fix nested C comments
992 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
993 o Make SET CONNECTION thread-aware, non-standard?
994 o Allow multidimensional arrays
995 o Add internationalized message strings
996 o Implement COPY FROM STDIN
1000 o Add PQescapeIdentifierConn()
1001 o Prevent PQfnumber() from lowercasing unquoted the column name
1003 PQfnumber() should never have been doing lowercasing, but
1004 historically it has so we need a way to prevent it
1006 o Allow statement results to be automatically batched to the client
1008 Currently all statement results are transferred to the libpq
1009 client before libpq makes the results available to the
1010 application. This feature would allow the application to make
1011 use of the first result rows while the rest are transferred, or
1012 held on the server waiting for them to be requested by libpq.
1013 One complexity is that a statement like SELECT 1/col could error
1014 out mid-way through the result set.
1016 o Consider disallowing multiple queries in PQexec() as an
1017 additional barrier to SQL injection attacks
1019 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
1021 o Add PQexecf() that allows complex parameter substitution
1023 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
1025 o Add SQLSTATE severity to PGconn return status
1027 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
1033 * Add deferred trigger queue file
1035 Right now all deferred trigger information is stored in backend
1036 memory. This could exhaust memory for very large trigger queues.
1037 This item involves dumping large queues into files.
1039 * Allow triggers to be disabled in only the current session.
1041 This is currently possible by starting a multi-statement transaction,
1042 modifying the system tables, performing the desired SQL, restoring the
1043 system tables, and committing the transaction. ALTER TABLE ...
1044 TRIGGER requires a table lock so it is not ideal for this usage.
1046 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1048 If the dump is known to be valid, allow foreign keys to be added
1049 without revalidating the data.
1051 * Allow statement-level triggers to access modified rows
1052 * Support triggers on columns
1054 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1056 * Allow AFTER triggers on system tables
1058 System tables are modified in many places in the backend without going
1059 through the executor and therefore not causing triggers to fire. To
1060 complete this item, the functions that modify system tables will have
1063 * Tighten trigger permission checks
1065 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1067 * Allow BEFORE INSERT triggers on views
1069 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1071 * Add ability to trigger on TRUNCATE
1073 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
1080 * Add UNIQUE capability to non-btree indexes
1081 * Prevent index uniqueness checks when UPDATE does not modify the column
1083 Uniqueness (index) checks are done when updating a column even if the
1084 column is not modified by the UPDATE.
1086 * Allow the creation of on-disk bitmap indexes which can be quickly
1087 combined with other bitmap indexes
1089 Such indexes could be more compact if there are only a few distinct values.
1090 Such indexes can also be compressed. Keeping such indexes updated can be
1093 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1094 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1096 * Allow accurate statistics to be collected on indexes with more than
1097 one column or expression indexes, perhaps using per-index statistics
1099 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1100 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1102 * Consider increasing the number of default statistics target, and
1103 reduce statistics target overhead
1105 Also consider having a larger statistics target for indexed columns
1106 and expression indexes
1107 http://archives.postgresql.org/pgsql-general/2007-05/msg01228.php
1108 http://archives.postgresql.org/pgsql-general/2007-06/msg00542.php
1110 * Consider compressing indexes by storing key values duplicated in
1111 several rows as a single index entry
1113 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1114 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1115 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1116 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1118 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1120 This is difficult because you must upgrade to an exclusive table lock
1121 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1122 have this complication. This would allow index compaction without
1125 * Allow multiple indexes to be created concurrently, ideally via a
1126 single heap scan, and have pg_restore use it
1132 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1133 and primary/foreign keys
1134 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1135 on inherited table, e.g. INSERT INTO inherit_table
1136 (unique_index_col) VALUES (dup) should fail
1138 The main difficulty with this item is the problem of
1139 creating an index that can span multiple tables.
1141 o Allow SELECT ... FOR UPDATE on inherited tables
1142 o Require all CHECK constraints to be inherited
1144 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1149 o Add more GIST index support for geometric data types
1150 o Allow GIST indexes to create certain complex index types, like
1151 digital trees (see Aoki)
1156 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1158 o Pack hash index buckets onto disk pages more efficiently
1160 Currently only one hash bucket can be stored on a page. Ideally
1161 several hash buckets could be stored on a single page and greater
1162 granularity used for the hash algorithm.
1164 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1166 o Consider sorting hash buckets so entries can be found using a
1167 binary search, rather than a linear scan
1168 o In hash indexes, consider storing the hash value with or instead
1170 o Add WAL logging for crash recovery
1171 o Allow multi-column hash indexes
1172 o During index creation, pre-sort the tuples to improve build speed
1174 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
1181 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1183 Ideally this requires a separate test program that can be run
1184 at initdb time or optionally later. Consider O_SYNC when
1187 * Add program to test if fsync has a delay compared to non-fsync
1188 * Consider sorting writes during checkpoint
1190 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00541.php
1199 We could use a fixed row count and a +/- count to follow MVCC
1200 visibility rules, or a single cached value could be used and
1201 invalidated if anyone modifies the table. Another idea is to
1202 get a count directly from a unique index, but for this to be
1203 faster than a sequential scan it must avoid access to the heap
1204 to obtain tuple visibility information.
1206 * Provide a way to calculate an "estimated COUNT(*)"
1208 Perhaps by using the optimizer's cardinality estimates or random
1211 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1213 * Allow data to be pulled directly from indexes
1215 Currently indexes do not have enough tuple visibility information
1216 to allow data to be pulled from the index without also accessing
1217 the heap. One way to allow this is to set a bit on index tuples
1218 to indicate if a tuple is currently visible to all transactions
1219 when the first valid heap lookup happens. This bit would have to
1220 be cleared when a heap tuple is expired.
1222 Another idea is to maintain a bitmap of heap pages where all rows
1223 are visible to all backends, and allow index lookups to reference
1224 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1225 add someday to determine which heap pages need vacuuming. Frequently
1226 accessed bitmaps would have to be stored in shared memory. One 8k
1227 page of bitmaps could track 512MB of heap pages.
1229 A third idea would be for a heap scan to check if all rows are visible
1230 and if so set a per-table flag which can be checked by index scans.
1231 Any change to the table would have to clear the flag. To detect
1232 changes during the heap scan a counter could be set at the start and
1233 checked at the end --- if it is the same, the table has not been
1234 modified --- any table change would increment the counter.
1236 * Consider automatic caching of statements at various levels:
1239 o Query execute plan
1242 * Consider increasing internal areas when shared buffers is increased
1244 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1246 * Consider decreasing the amount of memory used by PrivateRefCount
1248 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1249 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1251 * Consider allowing higher priority queries to have referenced buffer
1252 cache pages stay in memory longer
1254 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00562.php
1261 * Improve speed with indexes
1263 For large table adjustments during VACUUM FULL, it is faster to cluster
1264 or reindex rather than update the index. Also, index updates can bloat
1267 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1268 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1270 * Auto-fill the free space map by scanning the buffer cache or by
1271 checking pages written by the background writer
1273 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1274 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1276 * Create a bitmap of pages that need vacuuming
1278 Instead of sequentially scanning the entire table, have the background
1279 writer or some other process record pages that have expired rows, then
1280 VACUUM can look at just those pages rather than the entire table. In
1281 the event of a system crash, the bitmap would probably be invalidated.
1282 One complexity is that index entries still have to be vacuumed, and
1283 doing this without an index scan (by using the heap values to find the
1284 index entry) might be slow and unreliable, especially for user-defined
1287 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1288 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1290 * Allow FSM to return free space toward the beginning of the heap file,
1291 in hopes that empty pages at the end can be truncated by VACUUM
1292 * Allow FSM page return free space based on table clustering, to assist
1293 in maintaining clustering?
1294 * Improve dead row detection during multi-statement transactions usage
1296 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1298 * Consider a more compact data representation for dead tuples
1300 http://archives.postgresql.org/pgsql-patches/2007-05/msg00143.php
1305 o %Issue log message to suggest VACUUM FULL if a table is nearly
1307 o Improve control of auto-vacuum
1309 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1311 o Prevent long-lived temporary tables from causing frozen-xid
1312 advancement starvation
1314 The problem is that autovacuum cannot vacuum them to set frozen xids;
1315 only the session that created them can do that.
1316 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1318 o Store per-table autovacuum settings in pg_class.reloptions.
1320 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1321 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1323 o Prevent autovacuum from running if an old transaction is still
1324 running from the last vacuum
1326 http://archives.postgresql.org/pgsql-hackers/2007-11/msg00899.php
1333 * Fix priority ordering of read and write light-weight locks (Neil)
1335 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1336 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1338 * Fix problem when multiple subtransactions of the same outer transaction
1339 hold different types of locks, and one subtransaction aborts
1341 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1342 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1343 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1344 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00773.php
1346 * Allow UPDATEs on only non-referential integrity columns not to conflict
1347 with referential integrity locks
1349 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1351 * Add idle_in_transaction_timeout GUC so locks are not held for long
1354 * Improve deadlock detection when a page cleaning lock conflicts
1355 with a shared buffer that is pinned
1357 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1358 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1359 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1364 Startup Time Improvements
1365 =========================
1367 * Experiment with multi-threaded backend for backend creation [thread]
1369 This would prevent the overhead associated with process creation. Most
1370 operating systems have trivial process creation time compared to
1371 database startup overhead, but a few operating systems (Win32,
1372 Solaris) might benefit from threading. Also explore the idea of
1373 a single session using multiple threads to execute a statement faster.
1380 * Eliminate need to write full pages to WAL before page modification [wal]
1382 Currently, to protect against partial disk page writes, we write
1383 full page images to WAL before they are modified so we can correct any
1384 partial page writes during recovery. These pages can also be
1385 eliminated from point-in-time archive files.
1387 o When off, write CRC to WAL and check file system blocks
1390 If CRC check fails during recovery, remember the page in case
1391 a later CRC for that page properly matches.
1393 o Write full pages during file system write and not when
1394 the page is modified in the buffer cache
1396 This allows most full page writes to happen in the background
1397 writer. It might cause problems for applying WAL on recovery
1398 into a partially-written page, but later the full page will be
1401 * Allow WAL traffic to be streamed to another server for stand-by
1403 * Reduce WAL traffic so only modified values are written rather than
1406 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1408 * Allow WAL information to recover corrupted pg_controldata
1410 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1412 * Find a way to reduce rotational delay when repeatedly writing
1415 Currently fsync of WAL requires the disk platter to perform a full
1416 rotation to fsync again. One idea is to write the WAL to different
1417 offsets that might reduce the rotational delay.
1419 * Allow WAL logging to be turned off for a table, but the table
1420 might be dropped or truncated during crash recovery [walcontrol]
1422 Allow tables to bypass WAL writes and just fsync() dirty pages on
1423 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1424 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1425 non-default logging should not use referential integrity with
1426 default-logging tables. A table without dirty buffers during a
1427 crash could perhaps avoid the drop/truncate.
1429 * Allow WAL logging to be turned off for a table, but the table would
1430 avoid being truncated/dropped [walcontrol]
1432 To do this, only a single writer can modify the table, and writes
1433 must happen only on new pages so the new pages can be removed during
1434 crash recovery. Readers can continue accessing the table. Such
1435 tables probably cannot have indexes. One complexity is the handling
1436 of indexes on TOAST tables.
1438 * Speed WAL recovery by allowing more than one page to be prefetched
1440 This involves having a separate process that can be told which pages
1441 the recovery process will need in the near future.
1442 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1446 Optimizer / Executor
1447 ====================
1449 * Improve selectivity functions for geometric operators
1450 * Precompile SQL functions to avoid overhead
1451 * Create utility to compute accurate random_page_cost value
1452 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1453 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1454 actual row counts differ by a specified percentage
1455 * Consider using hash buckets to do DISTINCT, rather than sorting
1457 This would be beneficial when there are few distinct values. This is
1458 already used by GROUP BY.
1460 * Log statements where the optimizer row estimates were dramatically
1461 different from the number of rows actually found?
1462 * Consider compressed annealing to search for query plans
1464 This might replace GEQO, http://sixdemonbag.org/Djinni.
1466 * Improve merge join performance by allowing mark/restore of
1469 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1471 * Consider using a hash for joining to a large IN (VALUES ...) list
1473 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00450.php
1480 * Consider having the background writer update the transaction status
1481 hint bits before writing out the page
1483 Implementing this requires the background writer to have access to system
1484 catalogs and the transaction status log.
1486 * Consider adding buffers the background writer finds reusable to the
1489 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1491 * Automatically tune bgwriter_delay based on activity rather then using a
1494 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00781.php
1496 * Consider wither increasing BM_MAX_USAGE_COUNT improves performance
1498 http://archives.postgresql.org/pgsql-hackers/2007-06/msg01007.php
1500 * Test to see if calling PreallocXlogFiles() from the background writer
1501 will help with WAL segment creation latency
1503 http://archives.postgresql.org/pgsql-patches/2007-06/msg00340.php
1507 Miscellaneous Performance
1508 =========================
1510 * Do async I/O for faster random read-ahead of data
1512 Async I/O allows multiple I/O requests to be sent to the disk with
1513 results coming back asynchronously.
1515 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1516 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1518 * Use mmap() rather than SYSV shared memory or to write WAL files?
1520 This would remove the requirement for SYSV SHM but would introduce
1521 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1522 to prevent I/O overhead.
1524 * Consider mmap()'ing files into a backend?
1526 Doing I/O to large tables would consume a lot of address space or
1527 require frequent mapping/unmapping. Extending the file also causes
1528 mapping problems that might require mapping only individual pages,
1529 leading to thousands of mappings. Another problem is that there is no
1530 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1531 could hit disk before WAL is written.
1533 * Add a script to ask system configuration questions and tune postgresql.conf
1534 * Consider ways of storing rows more compactly on disk
1536 o Reduce the row header size?
1537 o Consider reducing on-disk varlena length from four bytes to
1538 two because a heap row cannot be more than 64k in length
1540 * Consider increasing NUM_CLOG_BUFFERS
1541 * Allow user configuration of TOAST thresholds
1543 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1545 * Allow configuration of backend priorities via the operating system
1547 Though backend priorities make priority inversion during lock
1548 waits possible, research shows that this is not a huge problem.
1550 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1552 * Experiment with multi-threaded backend better resource utilization
1554 This would allow a single query to make use of multiple CPU's or
1555 multiple I/O channels simultaneously. One idea is to create a
1556 background reader that can pre-fetch sequential and index scan
1557 pages needed by other backends. This could be expanded to allow
1558 concurrent reads from multiple devices in a partitioned table.
1560 * Consider increasing the minimum allowed number of shared buffers
1562 http://archives.postgresql.org/pgsql-bugs/2008-02/msg00157.php
1570 * Add use of 'const' for variables in source tree
1571 * Move some things from contrib into main tree
1572 * %Remove warnings created by -Wcast-align
1573 * Move platform-specific ps status display info from ps_status.c to ports
1574 * Add optional CRC checksum to heap and index pages
1575 * Improve documentation to build only interfaces (Marc)
1576 * Remove or relicense modules that are not under the BSD license, if possible
1577 * Acquire lock on a relation before building a relcache entry for it
1578 * Allow cross-compiling by generating the zic database on the target system
1579 * Improve NLS maintenance of libpgport messages linked onto applications
1580 * Clean up casting in contrib/isn
1582 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1584 * Use UTF8 encoding for NLS messages so all server encodings can
1586 * Update Bonjour to work with newer cross-platform SDK
1588 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1589 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1591 * Consider detoasting keys before sorting
1592 * Consider GnuTLS if OpenSSL license becomes a problem
1594 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1595 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1597 * Consider changing documentation format from SGML to XML
1599 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1601 * Consider making NAMEDATALEN more configurable in future releases
1602 * Update our code to handle 64-bit timezone files to match the zic
1603 source code, which now uses them
1604 * Have configure choose integer datetimes by default
1606 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1608 * Support scoped IPv6 addresses
1610 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1612 * Consider allowing 64-bit integers and floats to be passed by value on
1615 Also change 32-bit floats (float4) to be passed by value at the same
1618 * Research use of signals and sleep wake ups
1620 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00003.php
1622 * Add automated check for invalid C++ source code constructs
1624 http://archives.postgresql.org/pgsql-patches/2007-07/msg00056.php
1629 o Remove configure.in check for link failure when cause is found
1630 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1632 o Remove psql newline patch when we find out why mingw outputs an
1634 o Allow psql to use readline once non-US code pages work with
1636 o Fix problem with shared memory on the Win32 Terminal Server
1637 o Improve signal handling
1639 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1643 * Wire Protocol Changes
1645 o Allow dynamic character set handling
1646 o Add decoded type, length, precision
1648 o Update clients to use data types, typmod, schema.table.column names
1649 of result sets using new statement protocol
1655 * Add pre-parsing phase that converts non-ISO syntax to supported
1658 This could allow SQL written for other databases to run without
1661 * Allow plug-in modules to emulate features from other databases
1662 * Add features of Oracle-style packages (Pavel)
1664 A package would be a schema with session-local variables,
1665 public/private functions, and initialization functions. It
1666 is also possible to implement these capabilities
1667 in any schema and not use a separate "packages"
1670 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1672 * Consider allowing control of upper/lower case folding of unquoted
1675 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1676 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1678 * Add autonomous transactions
1680 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1684 Features We Do _Not_ Want
1685 =========================
1687 * All backends running as threads in a single process (not wanted)
1689 This eliminates the process protection we get from the current setup.
1690 Thread creation is usually the same overhead as process creation on
1691 modern systems, so it seems unwise to use a pure threaded model.
1693 * Optimizer hints (not wanted)
1695 Optimizer hints are used to work around problems in the optimizer. We
1696 would rather have the problems reported and fixed.
1698 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1699 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1700 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1703 Because we support postfix operators, it isn't possible to make AS
1704 optional and continue to use bison.
1705 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1707 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1709 * Embedded server (not wanted)
1711 While PostgreSQL clients runs fine in limited-resource environments, the
1712 server requires multiple processes and a stable pool of resources to
1713 run reliabily and efficiently. Stripping down the PostgreSQL server
1714 to run in the same process address space as the client application
1715 would add too much complexity and failure cases.