3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Fri Mar 7 14:18:54 EST 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
33 * Check for unreferenced table files created by transactions that were
34 in-progress when the server terminated abruptly
36 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
38 * Set proper permissions on non-system schemas during db creation
40 Currently all schemas are owned by the super-user because they are copied
41 from the template1 database. However, since all objects are inherited
42 from the template database, it is not clear that setting schemas to the db
45 * Add function to report the time of the most recent server reload
46 * Allow statistics collector information to be pulled from the collector
47 process directly, rather than requiring the collector to write a
48 filesystem file twice a second?
49 * Allow log_min_messages to be specified on a per-module basis
51 This would allow administrators to see more detailed information from
52 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
53 Another idea is to allow separate configuration files for each module,
54 or allow arbitrary SET commands to be passed to them.
56 * Simplify ability to create partitioned tables
58 This would allow creation of partitioned tables without requiring
59 creation of triggers or rules for INSERT/UPDATE/DELETE, and constraints
60 for rapid partition selection. Options could include range and hash
63 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00375.php
64 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00151.php
66 * Allow auto-selection of partitioned tables for min/max() operations
67 * Allow more complex user/database default GUC settings
69 Currently ALTER USER and ALTER DATABASE support per-user and
70 per-database defaults. Consider adding per-user-and-database
71 defaults so things like search_path can be defaulted for a
72 specific user connecting to a specific database.
74 * Allow custom variable classes that can restrict who can set the values
76 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00911.php
78 * Implement the SQL standard mechanism whereby REVOKE ROLE revokes only
79 the privilege granted by the invoking role, and not those granted
82 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00010.php
84 * Allow SSL authentication/encryption over unix domain sockets
86 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00924.php
88 * Allow client certificate names to be checked against the client
91 This is already implemented in
92 libpq/fe-secure.c::verify_peer_name_matches_certificate() but the code
97 o Allow pg_hba.conf to specify host names along with IP addresses
99 Host name lookup could occur when the postmaster reads the
100 pg_hba.conf file, or when the backend starts. Another
101 solution would be to reverse lookup the connection IP and
102 check that hostname against the host names in pg_hba.conf.
103 We could also then check that the host name maps to the IP
106 o %Allow postgresql.conf file values to be changed via an SQL
107 API, perhaps using SET GLOBAL
108 o Allow the server to be stopped/restarted via an SQL API
109 o Issue a warning if a change-on-restart-only postgresql.conf value
110 is modified and the server config files are reloaded
115 o Allow a database in tablespace t1 with tables created in
116 tablespace t2 to be used as a template for a new database created
117 with default tablespace t2
119 Currently all objects in the default database tablespace must
120 have default tablespace specifications. This is because new
121 databases are created by copying directories. If you mix default
122 tablespace tables and tablespace-specified tables in the same
123 directory, creating a new database from such a mixed directory
124 would create a new database with tables that had incorrect
125 explicit tablespaces. To fix this would require modifying
126 pg_class in the newly copied database, which we don't currently
129 o Allow reporting of which objects are in which tablespaces
131 This item is difficult because a tablespace can contain objects
132 from multiple databases. There is a server-side function that
133 returns the databases which use a specific tablespace, so this
134 requires a tool that will call that function and connect to each
135 database to find the objects in each database for that tablespace.
137 o Allow WAL replay of CREATE TABLESPACE to work when the directory
138 structure on the recovery computer is different from the original
140 o Allow per-tablespace quotas
143 * Point-In-Time Recovery (PITR)
145 o Allow a warm standby system to also allow read-only statements
148 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
150 o %Create dump tool for write-ahead logs for use in determining
151 transaction id for point-in-time recovery
153 This is useful for checking PITR recovery.
155 o Allow recovery.conf to support the same syntax as
156 postgresql.conf, including quoting
158 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
164 * Change NUMERIC to enforce the maximum precision
165 * Reduce storage space for small NUMERICs
167 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
168 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
170 * Fix data types where equality comparison isn't intuitive, e.g. box
171 * Add support for public SYNONYMs
173 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
175 * Fix CREATE CAST on DOMAINs
177 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
178 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
180 * Add support for SQL-standard GENERATED/IDENTITY columns
182 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
183 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
184 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
185 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
186 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00604.php
188 * Improve XML support
190 http://developer.postgresql.org/index.php/XML_Support
192 * Consider placing all sequences in a single table, or create a system
195 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
197 * Allow the UUID type to accept non-standard formats
199 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01214.php
203 o Allow infinite dates and intervals just like infinite timestamps
204 o Merge hardwired timezone names with the TZ database; allow either
205 kind everywhere a TZ name is currently taken
206 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
207 information, either zone name or offset from UTC [timezone]
209 If the TIMESTAMP value is stored with a time zone name, interval
210 computations should adjust based on the time zone rules.
212 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
213 o Add a GUC variable to allow output of interval values in ISO8601
215 o Have timestamp subtraction not call justify_hours()?
217 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
219 o Improve timestamptz subtraction to be DST-aware
221 Currently subtracting one date from another that crosses a
222 daylight savings time adjustment can return '1 day 1 hour', but
223 adding that back to the first date returns a time one hour in
224 the future. This is caused by the adjustment of '25 hours' to
225 '1 day 1 hour', and '1 day' is the same time the next day, even
226 if daylight savings adjustments are involved.
228 o Fix interval display to support values exceeding 2^31 hours
229 o Add overflow checking to timestamp and interval arithmetic
230 o Extend timezone code to allow 64-bit values so we can
231 represent years beyond 2038
233 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
235 o Use LC_TIME for localized weekday/month names, rather than
238 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
240 o Add ISO INTERVAL handling
242 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
243 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
245 o Support ISO INTERVAL syntax if units cannot be determined from
246 the string, and are supplied after the string
248 The SQL standard states that the units after the string
249 specify the units of the string, e.g. INTERVAL '2' MINUTE
250 should return '00:02:00'. The current behavior has the units
251 restrict the interval value to the specified unit or unit
252 range, INTERVAL '70' SECOND returns '00:00:10'.
254 For syntax that isn't uniquely ISO or PG syntax, like '1' or
255 '1:30', treat as ISO if there is a range specification clause,
256 and as PG if there no clause is present, e.g. interpret '1:30'
257 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
258 '1:30' as '1 hour, 30 minutes'.
260 This makes common cases like SELECT INTERVAL '1' MONTH
261 SQL-standard results. The SQL standard supports a limited
262 number of unit combinations and doesn't support unit names in
263 the string. The PostgreSQL syntax is more flexible in the
264 range of units supported, e.g. PostgreSQL supports '1 year 1
265 hour', while the SQL standard does not.
267 o Add support for year-month syntax, INTERVAL '50-6' YEAR
269 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1
270 year' AS INTERVAL MONTH), and this should return '12 months'
271 o Round or truncate values to the requested precision, e.g.
272 INTERVAL '11 months' AS YEAR should return one or zero
273 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
278 o Delay resolution of array expression's data type so assignment
279 coercion can be performed on empty array expressions
280 o Add support for arrays of domains
282 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
284 o Allow single-byte header storage for array elements
289 o Improve vacuum of large objects, like contrib/vacuumlo?
290 o Add security checking for large objects
291 o Auto-delete large objects when referencing row is deleted
293 contrib/lo offers this functionality.
295 o Allow read/write into TOAST values like large objects
297 This requires the TOAST column to be stored EXTERNAL.
299 o Add API for 64-bit large object access
301 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
305 * Add locale-aware MONEY type, and support multiple currencies
307 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
308 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
310 * MONEY dumps in a locale-specific format making it difficult to
311 restore to a system with a different locale
312 * Allow MONEY to be easily cast to/from other numeric data types
318 * Allow INET subnet tests using non-constants to be indexed
319 * Allow to_date() and to_timestamp() accept localized month names
320 * Fix to_date()-related functions to consistently issue errors
322 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
324 * Add missing parameter handling in to_char()
326 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
328 * Allow substring/replace() to get/set bit values
329 * Allow to_char() on interval values to accumulate the highest unit
332 Some special format flag would be required to request such
333 accumulation. Such functionality could also be added to EXTRACT.
334 Prevent accumulation that crosses the month/day boundary because of
335 the uneven number of days in a month.
337 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
338 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
339 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
340 o to_char(INTERVAL '3 years 5 months','MM') => 41
342 * Implement inlining of set-returning functions defined in SQL
343 * Allow SQL-language functions to return results from RETURNING queries
345 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
347 * Allow SQL-language functions to reference parameters by parameter name
349 Currently SQL-language functions can only refer to dollar parameters,
352 * Add SPI_gettypmod() to return the typemod for a TupleDesc
353 * Enforce typmod for function inputs, function results and parameters for
354 spi_prepare'd statements called from PLs
356 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
358 * Allow holdable cursors in SPI
359 * Tighten function permission checks
361 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
363 * Fix IS OF so it matches the ISO specification, and add documentation
365 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
366 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
368 * Add missing operators for geometric data types
370 Some geometric types do not have the full suite of geometric operators,
373 * Implement Boyer-Moore searching in strpos()
375 http://archives.postgresql.org/pgsql-patches/2007-08/msg00012.php
377 * Prevent malicious functions from being executed with the permissions
378 of unsuspecting users
380 Index functions are safe, so VACUUM and ANALYZE are safe too.
381 Triggers, CHECK and DEFAULT expressions, and rules are still vulnerable.
382 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00268.php
384 * Reduce memory usage of aggregates in set returning functions
386 http://archives.postgresql.org/pgsql-performance/2008-01/msg00031.php
390 Multi-Language Support
391 ======================
393 * Add NCHAR (as distinguished from ordinary varchar),
394 * Allow locale to be set at database creation
396 Currently locale can only be set during initdb. No global tables have
397 locale-aware columns. However, the database template used during
398 database creation might have locale-aware indexes. The indexes would
399 need to be reindexed to match the new locale.
401 * Allow encoding on a per-column basis optionally using the ICU library:
403 Right now only one encoding is allowed per database. [locale]
405 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
406 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
407 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
408 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
410 * Add CREATE COLLATE? [locale]
411 * Support multiple simultaneous character sets, per SQL92
412 * Improve UTF8 combined character handling?
413 * Add octet_length_server() and octet_length_client()
414 * Make octet_length_client() the same as octet_length()?
415 * Fix problems with wrong runtime encoding conversion for NLS message files
416 * Add URL to more complete multi-byte regression tests
418 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
420 * Fix ILIKE and regular expressions to handle case insensitivity
421 properly in multibyte encodings
423 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
424 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
426 * Set client encoding based on the client operating system encoding
428 Currently client_encoding is set in postgresql.conf, which
429 defaults to the server encoding.
430 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
437 * Automatically create rules on views so they are updateable, per SQL99
439 We can only auto-create rules for simple views. For more complex
440 cases users will still have to write rules manually.
442 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
443 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
445 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
446 * Allow NOTIFY in rules involving conditionals
447 * Allow VIEW/RULE recompilation when the underlying tables change
449 Another issue is whether underlying table changes should be reflected
450 in the view, e.g. should SELECT * show additional columns if they
451 are added after the view is created.
453 * Make it possible to use RETURNING together with conditional DO INSTEAD
454 rules, such as for partitioning setups
456 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00577.php
458 * Add the ability to automatically create materialized views
460 Right now materialized views require the user to create triggers on the
461 main table to keep the summary table current. SQL syntax should be able
462 to manager the triggers and summary table automatically. A more
463 sophisticated implementation would automatically retrieve from the
464 summary table when the main table is referenced, if possible.
471 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
472 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
473 * %Allow SET CONSTRAINTS to be qualified by schema/table name
474 * %Add a separate TRUNCATE permission
476 Currently only the owner can TRUNCATE a table because triggers are not
477 called, and the table is locked in exclusive mode.
479 * Allow PREPARE of cursors
480 * Allow finer control over the caching of prepared query plans
482 Currently queries prepared via the libpq API are planned on first
483 execute using the supplied parameters --- allow SQL PREPARE to do the
484 same. Also, allow control over replanning prepared queries either
485 manually or automatically when statistics for execute parameters
486 differ dramatically from those used during planning.
488 * Improve logging of prepared transactions recovered during startup
490 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
492 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
494 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
495 such information in memory would improve performance.
497 * Add optional textual message to NOTIFY
499 This would allow an informational message to be added to the notify
500 message, perhaps indicating the row modified or other custom
503 * Allow multiple identical NOTIFY events to always be communicated to the
504 client, rather than sent as a single notification to the listener
506 http://archives.postgresql.org/pgsql-general/2008-01/msg00057.php
508 * Add a GUC variable to warn about non-standard SQL usage in queries
509 * Add SQL-standard MERGE command, typically used to merge two tables
512 This is similar to UPDATE, then for unmatched rows, INSERT.
513 Whether concurrent access allows modifications which could cause
514 row loss is implementation independent.
516 * Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
519 To implement this cleanly requires that the table have a unique index
520 so duplicate checking can be easily performed. It is possible to
521 do it without a unique index if we require the user to LOCK the table
524 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
525 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
527 * Add NOVICE output level for helpful messages like automatic sequence/index
529 * Add GUC to issue notice about statements that use unjoined tables
530 * Allow EXPLAIN to identify tables that were skipped because of
532 * Allow EXPLAIN output to be more easily processed by scripts, perhaps XML
533 * Enable standard_conforming_strings
534 * Make standard_conforming_strings the default in 8.5?
536 When this is done, backslash-quote should be prohibited in non-E''
537 strings because of possible confusion over how such strings treat
538 backslashes. Basically, '' is always safe for a literal single
539 quote, while \' might or might not be based on the backslash
542 * Simplify dropping roles that have objects in several databases
543 * Allow COMMENT ON to accept an expression rather than just a string
544 * Allow the count returned by SELECT, etc to be represented as an int64
545 to allow a higher range of values
546 * Add SQL99 WITH clause to SELECT
547 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
549 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01375.php
550 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00642.php
551 http://archives.postgresql.org/pgsql-patches/2007-03/msg00139.php
553 * Add DEFAULT .. AS OWNER so permission checks are done as the table
556 This would be useful for SERIAL nextval() calls and CHECK constraints.
558 * Allow DISTINCT to work in multiple-argument aggregate calls
559 * Add column to pg_stat_activity that shows the progress of long-running
560 commands like CREATE INDEX and VACUUM
561 * Implement SQL:2003 window functions
562 * Improve failure message when DROP DATABASE is used on a database that
563 has prepared transactions
564 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
566 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
567 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
569 * Increase locking when DROPing objects so dependent objects cannot
570 get dropped while the DROP operation is happening
572 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
574 * -Allow AS in "SELECT col AS label" to be optional in certain cases
576 * Allow INSERT ... DELETE ... RETURNING, namely allow the DELETE ...
577 RETURNING to supply values to the INSERT
578 http://archives.postgresql.org/pgsql-hackers/2008-02/thrd2.php#00979
580 * Add comments on system tables/columns using the information in
583 Ideally the information would be pulled from the SGML file
589 o Allow CREATE TABLE AS to determine column lengths for complex
590 expressions like SELECT col1 || col2
592 o Have WITH CONSTRAINTS also create constraint indexes
594 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
596 o Have CONSTRAINT cname NOT NULL record the contraint name
598 Right now pg_attribute.attnotnull records the NOT NULL status
599 of the column, but does not record the contraint name
604 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
606 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
607 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
609 o Research self-referential UPDATEs that see inconsistent row versions
610 in read-committed mode
612 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
613 http://archives.postgresql.org/pgsql-hackers/2007-06/msg00016.php
615 o Allow GLOBAL temporary tables to exist as empty by default in
618 http://archives.postgresql.org/pgsql-hackers/2007-07/msg00006.php
623 o Have ALTER TABLE RENAME rename SERIAL sequence names
625 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
627 o Have ALTER SEQUENCE RENAME rename the sequence name stored
628 in the sequence table
630 http://archives.postgresql.org/pgsql-bugs/2007-09/msg00092.php
631 http://archives.postgresql.org/pgsql-bugs/2007-10/msg00007.php
632 http://archives.postgresql.org/pgsql-hackers/2008-03/msg00008.php
634 o Add ALTER DOMAIN to modify the underlying data type
635 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
637 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
639 o %Allow ALTER TABLE to change constraint deferrability and actions
640 o Add missing object types for ALTER ... SET SCHEMA
641 o Allow ALTER TABLESPACE to move to different directories
642 o Allow databases to be moved to different tablespaces
643 o Allow moving system tables to other tablespaces, where possible
645 Currently non-global system tables must be in the default database
646 tablespace. Global system tables can never be moved.
648 o Prevent parent tables from altering or dropping constraints
649 like CHECK that are inherited by child tables unless CASCADE
651 o %Prevent child tables from altering or dropping constraints
652 like CHECK that were inherited from the parent table
653 o Have ALTER INDEX update the name of a constraint using that index
654 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
655 o Allow column display reordering by recording a display,
656 storage, and permanent id for every column?
658 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
664 o Automatically maintain clustering on a table
666 This might require some background daemon to maintain clustering
667 during periods of low usage. It might also require tables to be only
668 partially filled for easier reorganization. Another idea would
669 be to create a merged heap/index data file so an index lookup would
670 automatically access the heap data too. A third idea would be to
671 store heap rows in hashed groups, perhaps using a user-supplied
673 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
675 o %Add default clustering to system tables
677 To do this, determine the ideal cluster index for each system
678 table and set the cluster setting during initdb.
680 o %Add VERBOSE option to report tables as they are processed,
686 o Allow COPY to report error lines and continue
688 This requires the use of a savepoint before each COPY line is
689 processed, with ROLLBACK on COPY failure.
691 o Allow COPY on a newly-created table to skip WAL logging
693 On crash recovery, the table involved in the COPY would
694 be removed or have its heap and index files truncated. One
695 issue is that no other backend should be able to add to
696 the table at the same time, which is something that is
697 currently allowed. This currently is done if the table is
698 created inside the same transaction block as the COPY because
699 no other backends can see the table.
701 o Consider using a ring buffer for COPY FROM
703 http://archives.postgresql.org/pgsql-patches/2008-02/msg00140.php
704 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01080.php
706 o Allow COPY FROM to create index entries in bulk
708 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00811.php
714 o Allow column-level privileges
715 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
718 The proposed syntax is:
719 GRANT SELECT ON ALL TABLES IN public TO phpuser;
720 GRANT SELECT ON NEW TABLES IN public TO phpuser;
722 o Allow GRANT/REVOKE permissions to be inherited by objects based on
725 o Allow SERIAL sequences to inherit permissions from the base table?
730 o Prevent DROP TABLE from dropping a row referenced by its own open
736 o Allow INSERT/UPDATE of the system-generated oid value for a row
737 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
743 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
748 Referential Integrity
749 =====================
751 * Add MATCH PARTIAL referential integrity
752 * Change foreign key constraint for array -> element to mean element
754 * Fix problem when cascading referential triggers make changes on
755 cascaded tables, seeing the tables in an intermediate state
757 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
758 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
760 * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
762 This would allow UPDATE tab SET col = col + 1 to work if col has
763 a unique index. Currently, uniqueness checks are done while the
764 command is being executed, rather than at the end of the statement
766 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
767 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
769 * Improve referential integrity checks
771 http://archives.postgresql.org/pgsql-performance/2005-10/msg00458.php
775 Server-Side Languages
776 =====================
779 o Fix RENAME to work on variables other than OLD/NEW
781 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
782 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
783 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
785 o Allow function parameters to be passed by name,
786 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
787 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
788 o Allow listing of record column names, and access to
789 record columns via variables, e.g. columns := r.(*),
792 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
793 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
794 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
796 o Add support for SCROLL cursors
797 o Add support for WITH HOLD cursors
798 o Allow row and record variables to be set to NULL constants,
799 and allow NULL tests on such variables
801 Because a row is not scalar, do not allow assignment
802 from NULL-valued scalars.
804 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
807 o Add table function support to pltcl, plpythonu
808 o Add support for polymorphic arguments and return types to
809 languages other than PL/PgSQL
810 o Add capability to create and call PROCEDURES
811 o Add support for OUT and INOUT parameters to languages other
813 o Add PL/PythonU tracebacks
815 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
817 o Allow data to be passed in native language formats, rather
820 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
827 * Have pg_ctl look at PGHOST in case it is a socket directory?
828 * Allow pg_ctl to work properly with configuration files located outside
831 pg_ctl can not read the pid file because it isn't located in the
832 config directory but in the PGDATA directory. The solution is to
833 allow pg_ctl to read and understand postgresql.conf to find the
834 data_directory value.
837 o Have psql show current values for a sequence
838 o Move psql backslash database information into the backend, use
839 mnemonic commands? [psql]
841 This would allow non-psql clients to pull the same information out
842 of the database as psql.
844 o Fix psql's \d commands more consistent
846 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
847 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
849 o Consistently display privilege information for all objects in psql
850 o Add auto-expanded mode so expanded output is used if the row
851 length is wider than the screen width.
853 Consider using auto-expanded mode for backslash commands like \df+.
855 o Prevent tab completion of SET TRANSACTION from querying the
856 database and therefore preventing the transaction isolation
857 level from being set.
859 Currently SET <tab> causes a database lookup to check all
860 supported session variables. This query causes problems
861 because setting the transaction isolation level must be the
862 first statement of a transaction.
864 o Add a \set variable to control whether \s displays line numbers
866 Another option is to add \# which lists line numbers, and
867 allows command execution.
869 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
871 o Prevent escape string warnings when object names have
874 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00227.php
876 o Have \d show foreign keys that reference a table's primary key
878 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00424.php
880 o Have \d show child tables that inherit from the specified parent
881 o Have \l+ show database size, if permissions allow
883 Ideally it will not generate an error for invalid permissions
887 * pg_dump / pg_restore
888 o %Add dumping of comments on index columns and composite type columns
889 o %Add full object name to the tag field. eg. for operators we need
890 '=(integer, integer)', instead of just '='.
891 o Add pg_dumpall custom format dumps?
892 o Allow selection of individual object(s) of all types, not just
894 o In a selective dump, allow dumping of an object and all its
896 o Add options like pg_restore -l and -L to pg_dump
897 o Stop dumping CASCADE on DROP TYPE commands in clean mode
898 o Allow pg_dump --clean to drop roles that own objects or have
900 o Change pg_dump so that a comment on the dumped database is
901 applied to the loaded database, even if the database has a
902 different name. This will require new backend syntax, perhaps
903 COMMENT ON CURRENT DATABASE.
904 o Remove unnecessary function pointer abstractions in pg_dump source
906 o Allow pg_dump to utilize multiple CPUs and I/O channels by dumping
907 multiple objects simultaneously
909 The difficulty with this is getting multiple dump processes to
910 produce a single dump output file. It also would require
911 several sessions to share the same snapshot.
912 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00205.php
914 o Allow pg_restore to utilize multiple CPUs and I/O channels by
915 restoring multiple objects simultaneously
917 This might require a pg_restore flag to indicate how many
918 simultaneous operations should be performed. Only pg_dump's
919 -Fc format has the necessary dependency information.
920 http://archives.postgresql.org/pgsql-hackers/2008-02/msg00963.php
922 o To better utilize resources, allow pg_restore to check foreign
923 keys simultaneously, where possible
924 o Allow pg_restore to create all indexes of a table
925 concurrently, via a single heap scan
927 This requires a pg_dump -Fc file because that format contains
928 the required dependency information.
929 http://archives.postgresql.org/pgsql-general/2007-05/msg01274.php
931 o Allow pg_restore to load different parts of the COPY data
938 Document differences between ecpg and the SQL standard and
939 information about the Informix-compatibility module.
941 o Solve cardinality > 1 for input descriptors / variables?
942 o Add a semantic check level, e.g. check if a table really exists
943 o fix handling of DB attributes that are arrays
944 o Use backend PREPARE/EXECUTE facility for ecpg where possible
946 o Fix nested C comments
947 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
948 o Make SET CONNECTION thread-aware, non-standard?
949 o Allow multidimensional arrays
950 o Add internationalized message strings
951 o Implement COPY FROM STDIN
955 o Add PQescapeIdentifierConn()
956 o Prevent PQfnumber() from lowercasing unquoted the column name
958 PQfnumber() should never have been doing lowercasing, but
959 historically it has so we need a way to prevent it
961 o Allow statement results to be automatically batched to the client
963 Currently all statement results are transferred to the libpq
964 client before libpq makes the results available to the
965 application. This feature would allow the application to make
966 use of the first result rows while the rest are transferred, or
967 held on the server waiting for them to be requested by libpq.
968 One complexity is that a statement like SELECT 1/col could error
969 out mid-way through the result set.
971 o Consider disallowing multiple queries in PQexec() as an
972 additional barrier to SQL injection attacks
974 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
976 o Add PQexecf() that allows complex parameter substitution
978 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
980 o Add SQLSTATE severity to PGconn return status
982 http://archives.postgresql.org/pgsql-interfaces/2007-11/msg00015.php
988 * Add deferred trigger queue file
990 Right now all deferred trigger information is stored in backend
991 memory. This could exhaust memory for very large trigger queues.
992 This item involves dumping large queues into files.
994 * Allow triggers to be disabled in only the current session.
996 This is currently possible by starting a multi-statement transaction,
997 modifying the system tables, performing the desired SQL, restoring the
998 system tables, and committing the transaction. ALTER TABLE ...
999 TRIGGER requires a table lock so it is not ideal for this usage.
1001 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
1003 If the dump is known to be valid, allow foreign keys to be added
1004 without revalidating the data.
1006 * Allow statement-level triggers to access modified rows
1007 * Support triggers on columns
1009 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
1011 * Allow AFTER triggers on system tables
1013 System tables are modified in many places in the backend without going
1014 through the executor and therefore not causing triggers to fire. To
1015 complete this item, the functions that modify system tables will have
1018 * Tighten trigger permission checks
1020 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
1022 * Allow BEFORE INSERT triggers on views
1024 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
1026 * Add ability to trigger on TRUNCATE
1028 http://archives.postgresql.org/pgsql-sql/2008-01/msg00050.php
1035 * Add UNIQUE capability to non-btree indexes
1036 * Prevent index uniqueness checks when UPDATE does not modify the column
1038 Uniqueness (index) checks are done when updating a column even if the
1039 column is not modified by the UPDATE.
1041 * Allow the creation of on-disk bitmap indexes which can be quickly
1042 combined with other bitmap indexes
1044 Such indexes could be more compact if there are only a few distinct values.
1045 Such indexes can also be compressed. Keeping such indexes updated can be
1048 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1049 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1051 * Allow accurate statistics to be collected on indexes with more than
1052 one column or expression indexes, perhaps using per-index statistics
1054 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1055 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1057 * Consider compressing indexes by storing key values duplicated in
1058 several rows as a single index entry
1060 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00341.php
1061 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01264.php
1062 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00465.php
1063 http://archives.postgresql.org/pgsql-patches/2007-03/msg00163.php
1065 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1067 This is difficult because you must upgrade to an exclusive table lock
1068 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1069 have this complication. This would allow index compaction without
1072 * Allow multiple indexes to be created concurrently, ideally via a
1073 single heap scan, and have pg_restore use it
1079 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1080 and primary/foreign keys
1081 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1082 on inherited table, e.g. INSERT INTO inherit_table
1083 (unique_index_col) VALUES (dup) should fail
1085 The main difficulty with this item is the problem of
1086 creating an index that can span multiple tables.
1088 o Allow SELECT ... FOR UPDATE on inherited tables
1089 o Require all CHECK constraints to be inherited
1091 http://archives.postgresql.org/pgsql-bugs/2007-04/msg00026.php
1096 o Add more GIST index support for geometric data types
1097 o Allow GIST indexes to create certain complex index types, like
1098 digital trees (see Aoki)
1103 http://archives.postgresql.org/pgsql-hackers/2007-09/msg00051.php
1105 o Pack hash index buckets onto disk pages more efficiently
1107 Currently only one hash bucket can be stored on a page. Ideally
1108 several hash buckets could be stored on a single page and greater
1109 granularity used for the hash algorithm.
1111 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1113 o Consider sorting hash buckets so entries can be found using a
1114 binary search, rather than a linear scan
1115 o In hash indexes, consider storing the hash value with or instead
1117 o Add WAL logging for crash recovery
1118 o Allow multi-column hash indexes
1119 o During index creation, pre-sort the tuples to improve build speed
1121 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
1128 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1130 Ideally this requires a separate test program that can be run
1131 at initdb time or optionally later. Consider O_SYNC when
1134 * Add program to test if fsync has a delay compared to non-fsync
1143 We could use a fixed row count and a +/- count to follow MVCC
1144 visibility rules, or a single cached value could be used and
1145 invalidated if anyone modifies the table. Another idea is to
1146 get a count directly from a unique index, but for this to be
1147 faster than a sequential scan it must avoid access to the heap
1148 to obtain tuple visibility information.
1150 * Provide a way to calculate an "estimated COUNT(*)"
1152 Perhaps by using the optimizer's cardinality estimates or random
1155 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1157 * Allow data to be pulled directly from indexes
1159 Currently indexes do not have enough tuple visibility information
1160 to allow data to be pulled from the index without also accessing
1161 the heap. One way to allow this is to set a bit on index tuples
1162 to indicate if a tuple is currently visible to all transactions
1163 when the first valid heap lookup happens. This bit would have to
1164 be cleared when a heap tuple is expired.
1166 Another idea is to maintain a bitmap of heap pages where all rows
1167 are visible to all backends, and allow index lookups to reference
1168 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1169 add someday to determine which heap pages need vacuuming. Frequently
1170 accessed bitmaps would have to be stored in shared memory. One 8k
1171 page of bitmaps could track 512MB of heap pages.
1173 A third idea would be for a heap scan to check if all rows are visible
1174 and if so set a per-table flag which can be checked by index scans.
1175 Any change to the table would have to clear the flag. To detect
1176 changes during the heap scan a counter could be set at the start and
1177 checked at the end --- if it is the same, the table has not been
1178 modified --- any table change would increment the counter.
1180 * Consider automatic caching of statements at various levels:
1183 o Query execute plan
1186 * Consider increasing internal areas when shared buffers is increased
1188 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1190 * Consider decreasing the amount of memory used by PrivateRefCount
1192 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1193 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1200 * Improve speed with indexes
1202 For large table adjustments during VACUUM FULL, it is faster to cluster
1203 or reindex rather than update the index. Also, index updates can bloat
1206 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1207 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1209 * Auto-fill the free space map by scanning the buffer cache or by
1210 checking pages written by the background writer
1212 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1213 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1215 * Create a bitmap of pages that need vacuuming
1217 Instead of sequentially scanning the entire table, have the background
1218 writer or some other process record pages that have expired rows, then
1219 VACUUM can look at just those pages rather than the entire table. In
1220 the event of a system crash, the bitmap would probably be invalidated.
1221 One complexity is that index entries still have to be vacuumed, and
1222 doing this without an index scan (by using the heap values to find the
1223 index entry) might be slow and unreliable, especially for user-defined
1226 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1227 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1229 * Allow FSM to return free space toward the beginning of the heap file,
1230 in hopes that empty pages at the end can be truncated by VACUUM
1231 * Allow FSM page return free space based on table clustering, to assist
1232 in maintaining clustering?
1233 * Improve dead row detection during multi-statement transactions usage
1235 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1239 o %Issue log message to suggest VACUUM FULL if a table is nearly
1241 o Improve control of auto-vacuum
1243 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1245 o Prevent long-lived temporary tables from causing frozen-xid
1246 advancement starvation
1248 The problem is that autovacuum cannot vacuum them to set frozen xids;
1249 only the session that created them can do that.
1250 http://archives.postgresql.org/pgsql-general/2007-06/msg01645.php
1252 o Store per-table autovacuum settings in pg_class.reloptions.
1254 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01440.php
1255 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00724.php
1261 * Fix priority ordering of read and write light-weight locks (Neil)
1263 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1264 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1266 * Fix problem when multiple subtransactions of the same outer transaction
1267 hold different types of locks, and one subtransaction aborts
1269 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1270 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1271 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1273 * Allow UPDATEs on only non-referential integrity columns not to conflict
1274 with referential integrity locks
1276 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1278 * Add idle_in_transaction_timeout GUC so locks are not held for long
1281 * Improve deadlock detection when a page cleaning lock conflicts
1282 with a shared buffer that is pinned
1284 http://archives.postgresql.org/pgsql-bugs/2008-01/msg00138.php
1285 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00873.php
1286 http://archives.postgresql.org/pgsql-committers/2008-01/msg00365.php
1291 Startup Time Improvements
1292 =========================
1294 * Experiment with multi-threaded backend for backend creation [thread]
1296 This would prevent the overhead associated with process creation. Most
1297 operating systems have trivial process creation time compared to
1298 database startup overhead, but a few operating systems (Win32,
1299 Solaris) might benefit from threading. Also explore the idea of
1300 a single session using multiple threads to execute a statement faster.
1307 * Eliminate need to write full pages to WAL before page modification [wal]
1309 Currently, to protect against partial disk page writes, we write
1310 full page images to WAL before they are modified so we can correct any
1311 partial page writes during recovery. These pages can also be
1312 eliminated from point-in-time archive files.
1314 o When off, write CRC to WAL and check file system blocks
1317 If CRC check fails during recovery, remember the page in case
1318 a later CRC for that page properly matches.
1320 o Write full pages during file system write and not when
1321 the page is modified in the buffer cache
1323 This allows most full page writes to happen in the background
1324 writer. It might cause problems for applying WAL on recovery
1325 into a partially-written page, but later the full page will be
1328 * Allow WAL traffic to be streamed to another server for stand-by
1330 * Reduce WAL traffic so only modified values are written rather than
1333 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1335 * Allow WAL information to recover corrupted pg_controldata
1337 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1339 * Find a way to reduce rotational delay when repeatedly writing
1342 Currently fsync of WAL requires the disk platter to perform a full
1343 rotation to fsync again. One idea is to write the WAL to different
1344 offsets that might reduce the rotational delay.
1346 * Allow WAL logging to be turned off for a table, but the table
1347 might be dropped or truncated during crash recovery [walcontrol]
1349 Allow tables to bypass WAL writes and just fsync() dirty pages on
1350 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1351 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1352 non-default logging should not use referential integrity with
1353 default-logging tables. A table without dirty buffers during a
1354 crash could perhaps avoid the drop/truncate.
1356 * Allow WAL logging to be turned off for a table, but the table would
1357 avoid being truncated/dropped [walcontrol]
1359 To do this, only a single writer can modify the table, and writes
1360 must happen only on new pages so the new pages can be removed during
1361 crash recovery. Readers can continue accessing the table. Such
1362 tables probably cannot have indexes. One complexity is the handling
1363 of indexes on TOAST tables.
1365 * Speed WAL recovery by allowing more than one page to be prefetched
1367 This involves having a separate process that can be told which pages
1368 the recovery process will need in the near future.
1369 http://archives.postgresql.org/pgsql-hackers/2008-02/msg01279.php
1373 Optimizer / Executor
1374 ====================
1376 * Improve selectivity functions for geometric operators
1377 * Precompile SQL functions to avoid overhead
1378 * Create utility to compute accurate random_page_cost value
1379 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1380 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1381 actual row counts differ by a specified percentage
1382 * Consider using hash buckets to do DISTINCT, rather than sorting
1384 This would be beneficial when there are few distinct values. This is
1385 already used by GROUP BY.
1387 * Log statements where the optimizer row estimates were dramatically
1388 different from the number of rows actually found?
1389 * Consider compressed annealing to search for query plans
1391 This might replace GEQO, http://sixdemonbag.org/Djinni.
1393 * Improve merge join performance by allowing mark/restore of
1396 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1400 Miscellaneous Performance
1401 =========================
1403 * Do async I/O for faster random read-ahead of data
1405 Async I/O allows multiple I/O requests to be sent to the disk with
1406 results coming back asynchronously.
1408 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1409 http://archives.postgresql.org/pgsql-hackers/2007-12/msg00027.php
1411 * Use mmap() rather than SYSV shared memory or to write WAL files?
1413 This would remove the requirement for SYSV SHM but would introduce
1414 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1415 to prevent I/O overhead.
1417 * Consider mmap()'ing files into a backend?
1419 Doing I/O to large tables would consume a lot of address space or
1420 require frequent mapping/unmapping. Extending the file also causes
1421 mapping problems that might require mapping only individual pages,
1422 leading to thousands of mappings. Another problem is that there is no
1423 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1424 could hit disk before WAL is written.
1426 * Add a script to ask system configuration questions and tune postgresql.conf
1427 * Consider ways of storing rows more compactly on disk
1429 o Reduce the row header size?
1430 o Consider reducing on-disk varlena length from four bytes to
1431 two because a heap row cannot be more than 64k in length
1433 * Consider increasing NUM_CLOG_BUFFERS
1434 * Consider having the background writer update the transaction status
1435 hint bits before writing out the page
1437 Implementing this requires the background writer to have access to system
1438 catalogs and the transaction status log.
1440 * Allow user configuration of TOAST thresholds
1442 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1444 * Allow configuration of backend priorities via the operating system
1446 Though backend priorities make priority inversion during lock
1447 waits possible, research shows that this is not a huge problem.
1449 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1451 * Experiment with multi-threaded backend better resource utilization
1453 This would allow a single query to make use of multiple CPU's or
1454 multiple I/O channels simultaneously. One idea is to create a
1455 background reader that can pre-fetch sequential and index scan
1456 pages needed by other backends. This could be expanded to allow
1457 concurrent reads from multiple devices in a partitioned table.
1464 * Add use of 'const' for variables in source tree
1465 * Move some things from contrib into main tree
1466 * %Remove warnings created by -Wcast-align
1467 * Move platform-specific ps status display info from ps_status.c to ports
1468 * Add optional CRC checksum to heap and index pages
1469 * Improve documentation to build only interfaces (Marc)
1470 * Remove or relicense modules that are not under the BSD license, if possible
1471 * Acquire lock on a relation before building a relcache entry for it
1472 * Allow cross-compiling by generating the zic database on the target system
1473 * Improve NLS maintenance of libpgport messages linked onto applications
1474 * Clean up casting in contrib/isn
1476 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1478 * Use UTF8 encoding for NLS messages so all server encodings can
1480 * Update Bonjour to work with newer cross-platform SDK
1482 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1483 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1485 * Consider detoasting keys before sorting
1486 * Consider GnuTLS if OpenSSL license becomes a problem
1488 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1489 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1491 * Consider changing documentation format from SGML to XML
1493 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1495 * Consider making NAMEDATALEN more configurable in future releases
1496 * Update our code to handle 64-bit timezone files to match the zic
1497 source code, which now uses them
1498 * Have configure choose integer datetimes by default
1500 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1502 * Support scoped IPv6 addresses
1504 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1506 * Consider allowing 64-bit integers and floats to be passed by value on
1509 Also change 32-bit floats (float4) to be passed by value at the same
1516 o Remove configure.in check for link failure when cause is found
1517 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1519 o Remove psql newline patch when we find out why mingw outputs an
1521 o Allow psql to use readline once non-US code pages work with
1523 o Fix problem with shared memory on the Win32 Terminal Server
1524 o Improve signal handling
1526 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1530 * Wire Protocol Changes
1532 o Allow dynamic character set handling
1533 o Add decoded type, length, precision
1535 o Update clients to use data types, typmod, schema.table.column names
1536 of result sets using new statement protocol
1542 * Add pre-parsing phase that converts non-ISO syntax to supported
1545 This could allow SQL written for other databases to run without
1548 * Allow plug-in modules to emulate features from other databases
1549 * Add features of Oracle-style packages (Pavel)
1551 A package would be a schema with session-local variables,
1552 public/private functions, and initialization functions. It
1553 is also possible to implement these capabilities
1554 in any schema and not use a separate "packages"
1557 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
1559 * Consider allowing control of upper/lower case folding of unquoted
1562 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1563 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1565 * Add autonomous transactions
1567 http://archives.postgresql.org/pgsql-hackers/2008-01/msg00893.php
1571 Features We Do _Not_ Want
1572 =========================
1574 * All backends running as threads in a single process (not wanted)
1576 This eliminates the process protection we get from the current setup.
1577 Thread creation is usually the same overhead as process creation on
1578 modern systems, so it seems unwise to use a pure threaded model.
1580 * Optimizer hints (not wanted)
1582 Optimizer hints are used to work around problems in the optimizer. We
1583 would rather have the problems reported and fixed.
1585 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1586 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1587 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1590 Because we support postfix operators, it isn't possible to make AS
1591 optional and continue to use bison.
1592 http://archives.postgresql.org/pgsql-hackers/2003-04/msg00436.php
1594 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1596 * Embedded server (not wanted)
1598 While PostgreSQL clients runs fine in limited-resource environments, the
1599 server requires multiple processes and a stable pool of resources to
1600 run reliabily and efficiently. Stripping down the PostgreSQL server
1601 to run in the same process address space as the client application
1602 would add too much complexity and failure cases.