3 Current maintainer: Bruce Momjian (bruce@momjian.us)
4 Last updated: Fri Jun 1 14:40:33 EDT 2007
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 major upgrades without dump/reload, perhaps using pg_upgrade
25 * Check for unreferenced table files created by transactions that were
26 in-progress when the server terminated abruptly
28 http://archives.postgresql.org/pgsql-patches/2006-06/msg00096.php
30 * Allow administrators to safely terminate individual sessions either
31 via an SQL function or SIGTERM
33 Lock table corruption following SIGTERM of an individual backend
34 has been reported in 8.0. A possible cause was fixed in 8.1, but
35 it is unknown whether other problems exist. This item mostly
36 requires additional testing rather than of writing any new code.
37 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00174.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 * Support table partitioning that allows a single table to be stored
47 in subtables that are partitioned based on the primary key or a WHERE
49 * Add function to report the time of the most recent server reload
50 * Allow statistics collector information to be pulled from the collector
51 process directly, rather than requiring the collector to write a
52 filesystem file twice a second?
53 * Allow log_min_messages to be specified on a per-module basis
55 This would allow administrators to see more detailed information from
56 specific sections of the backend, e.g. checkpoints, autovacuum, etc.
57 Another idea is to allow separate configuration files for each module,
58 or allow arbitrary SET commands to be passed to them.
60 * Simplify ability to create partitioned tables
62 This would allow creation of partitioned tables without requiring
63 creation of rules for INSERT/UPDATE/DELETE, and constraints for
64 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
88 * Improve replication solutions
92 You can use any of the master/slave replication servers to use a
93 standby server for data warehousing. To allow read/write queries to
94 multiple servers, you need multi-master replication like pgcluster.
96 o Allow replication over unreliable or non-persistent links
101 o -Allow commenting of variables in postgresql.conf to restore them
103 o Allow pg_hba.conf to specify host names along with IP addresses
105 Host name lookup could occur when the postmaster reads the
106 pg_hba.conf file, or when the backend starts. Another
107 solution would be to reverse lookup the connection IP and
108 check that hostname against the host names in pg_hba.conf.
109 We could also then check that the host name maps to the IP
112 o %Allow postgresql.conf file values to be changed via an SQL
113 API, perhaps using SET GLOBAL
114 o Allow the server to be stopped/restarted via an SQL API
115 o Issue a warning if a change-on-restart-only postgresql.conf value
116 is modified and the server config files are reloaded
117 o Mark change-on-restart-only values in postgresql.conf
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 All objects in the default database tablespace must have default
127 tablespace specifications. This is because new databases are
128 created by copying directories. If you mix default tablespace
129 tables and tablespace-specified tables in the same directory,
130 creating a new database from such a mixed directory would create a
131 new database with tables that had incorrect explicit tablespaces.
132 To fix this would require modifying pg_class in the newly copied
133 database, which we don't currently do.
135 o Allow reporting of which objects are in which tablespaces
137 This item is difficult because a tablespace can contain objects
138 from multiple databases. There is a server-side function that
139 returns the databases which use a specific tablespace, so this
140 requires a tool that will call that function and connect to each
141 database to find the objects in each database for that tablespace.
143 o -Add a GUC variable to control the tablespace for temporary objects
146 It could start with a random tablespace from a supplied list and
147 cycle through the list.
149 o Allow WAL replay of CREATE TABLESPACE to work when the directory
150 structure on the recovery computer is different from the original
152 o Allow per-tablespace quotas
155 * Point-In-Time Recovery (PITR)
157 o Allow a warm standby system to also allow read-only statements
160 This is useful for checking PITR recovery.
161 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00050.php
163 o %Create dump tool for write-ahead logs for use in determining
164 transaction id for point-in-time recovery
165 o Allow the PITR process to be debugged and data examined
166 o Allow recovery.conf to allow the same syntax as
167 postgresql.conf, including quoting
169 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00497.php
175 * Allow server log information to be output as INSERT statements
177 This would allow server log information to be easily loaded into
178 a database for analysis.
180 * -Add ability to monitor the use of temporary sort files
186 * -Make 64-bit version of the MONEY data type
187 * Add locale-aware MONEY type, and support multiple currencies
189 http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php
190 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php
192 * Change NUMERIC to enforce the maximum precision
193 * Reduce storage space for small NUMERICs
195 http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
196 http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php
198 * Fix data types where equality comparison isn't intuitive, e.g. box
199 * Allow user-defined types to specify a type modifier at table creation
201 * -Allow user-defined types to accept 'typmod' parameters
203 http://archives.postgresql.org/pgsql-hackers/2005-08/msg01142.php
204 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00012.php
205 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00149.php
207 * Add support for public SYNONYMs
209 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php
211 * Fix CREATE CAST on DOMAINs
213 http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php
214 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01681.php
216 * -Add Globally/Universally Unique Identifier (GUID/UUID)
218 http://archives.postgresql.org/pgsql-patches/2006-09/msg00209.php
219 http://archives.postgresql.org/pgsql-general/2007-01/msg00853.php
221 * Add support for SQL-standard GENERATED/IDENTITY columns
223 http://archives.postgresql.org/pgsql-hackers/2006-07/msg00543.php
224 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00038.php
225 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00344.php
226 http://archives.postgresql.org/pgsql-patches/2007-05/msg00076.php
228 * -Support a data type with specific enumerated values (ENUM)
229 * Improve XML support
231 http://developer.postgresql.org/index.php/XML_Support
233 * Consider placing all sequences in a single table, or create a system
235 * Allow all data types to cast to and from TEXT
237 http://archives.postgresql.org/pgsql-hackers/2007-04/msg00017.php
242 o Allow infinite dates and intervals just like infinite timestamps
243 o Merge hardwired timezone names with the TZ database; allow either
244 kind everywhere a TZ name is currently taken
245 o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
246 information, either zone name or offset from UTC [timezone]
248 If the TIMESTAMP value is stored with a time zone name, interval
249 computations should adjust based on the time zone rules.
251 o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
252 o Add a GUC variable to allow output of interval values in ISO8601
254 o Have timestamp subtraction not call justify_hours()?
256 http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
258 o Improve timestamptz subtraction to be DST-aware
260 Currently subtracting one date from another that crosses a
261 daylight savings time adjustment can return '1 day 1 hour', but
262 adding that back to the first date returns a time one hour in
263 the future. This is caused by the adjustment of '25 hours' to
264 '1 day 1 hour', and '1 day' is the same time the next day, even
265 if daylight savings adjustments are involved.
267 o Fix interval display to support values exceeding 2^31 hours
268 o Add overflow checking to timestamp and interval arithmetic
269 o Extend timezone code to allow 64-bit values so we can
270 represent years beyond 2038
272 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01363.php
274 o Use LC_TIME for localized weekday/month names, rather than
277 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php
279 o Add ISO INTERVAL handling
281 http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
282 http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php
284 o Support ISO INTERVAL syntax if units cannot be determined from
285 the string, and are supplied after the string
287 The SQL standard states that the units after the string
288 specify the units of the string, e.g. INTERVAL '2' MINUTE
289 should return '00:02:00'. The current behavior has the units
290 restrict the interval value to the specified unit or unit
291 range, INTERVAL '70' SECOND returns '00:00:10'.
293 For syntax that isn't uniquely ISO or PG syntax, like '1' or
294 '1:30', treat as ISO if there is a range specification clause,
295 and as PG if there no clause is present, e.g. interpret '1:30'
296 MINUTE TO SECOND as '1 minute 30 seconds', and interpret
297 '1:30' as '1 hour, 30 minutes'.
299 This makes common cases like SELECT INTERVAL '1' MONTH
300 SQL-standard results. The SQL standard supports a limited
301 number of unit combinations and doesn't support unit names in
302 the string. The PostgreSQL syntax is more flexible in the
303 range of units supported, e.g. PostgreSQL supports '1 year 1
304 hour', while the SQL standard does not.
306 o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
307 o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
308 INTERVAL MONTH), and this should return '12 months'
309 o Round or truncate values to the requested precision, e.g.
310 INTERVAL '11 months' AS YEAR should return one or zero
311 o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
316 o Delay resolution of array expression's data type so assignment
317 coercion can be performed on empty array expressions
318 o Add support for arrays of domains
320 http://archives.postgresql.org/pgsql-patches/2007-05/msg00114.php
322 o -Add support for arrays of complex types
323 o Allow single-byte header storage for array elements
328 o Improve vacuum of large objects, like /contrib/vacuumlo?
329 o Add security checking for large objects
330 o Auto-delete large objects when referencing row is deleted
332 /contrib/lo offers this functionality.
334 o Allow read/write into TOAST values like large objects
336 This requires the TOAST column to be stored EXTERNAL.
338 o Add API for 64-bit large object access
340 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00781.php
346 * Allow INET subnet tests using non-constants to be indexed
347 * Allow to_date() and to_timestamp() accept localized month names
348 * Fix to_date()-related functions to consistently issue errors
350 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00915.php
352 * Add missing parameter handling in to_char()
354 http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php
356 * Allow functions to have a schema search path specified at creation time
357 * Allow substring/replace() to get/set bit values
358 * Allow to_char() on interval values to accumulate the highest unit
361 Some special format flag would be required to request such
362 accumulation. Such functionality could also be added to EXTRACT.
363 Prevent accumulation that crosses the month/day boundary because of
364 the uneven number of days in a month.
366 o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
367 o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
368 o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
369 o to_char(INTERVAL '3 years 5 months','MM') => 41
371 * -Add ISO day of week format 'ID' to to_char() where Monday = 1
372 * -Add a field 'isoyear' to extract(), based on the ISO week
373 * Implement inlining of set-returning functions defined in SQL
374 * Allow SQL-language functions to return results from RETURNING queries
376 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00665.php
378 * Allow SQL-language functions to reference parameters by parameter name
380 Currently SQL-language functions can only refer to dollar parameters,
383 * Add SPI_gettypmod() to return the typemod for a TupleDesc
384 * Enforce typmod for function inputs, function results and parameters for
385 spi_prepare'd statements called from PLs
387 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01403.php
389 * Allow holdable cursors in SPI
390 * Tighten function permission checks
392 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00568.php
394 * Fix IS OF so it matches the ISO specification, and add documentation
396 http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
397 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php
399 * Add missing operators for geometric data types
401 Some geometric types do not have the full suite of geometric operators,
407 Multi-Language Support
408 ======================
410 * Add NCHAR (as distinguished from ordinary varchar),
411 * Allow locale to be set at database creation
413 Currently locale can only be set during initdb. No global tables have
414 locale-aware columns. However, the database template used during
415 database creation might have locale-aware indexes. The indexes would
416 need to be reindexed to match the new locale.
418 * Allow encoding on a per-column basis optionally using the ICU library:
420 Right now only one encoding is allowed per database. [locale]
421 http://archives.postgresql.org/pgsql-hackers/2005-03/msg00932.php
422 http://archives.postgresql.org/pgsql-patches/2005-08/msg00309.php
423 http://archives.postgresql.org/pgsql-patches/2006-03/msg00233.php
424 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00662.php
426 * Add CREATE COLLATE? [locale]
427 * Support multiple simultaneous character sets, per SQL92
428 * Improve UTF8 combined character handling?
429 * Add octet_length_server() and octet_length_client()
430 * Make octet_length_client() the same as octet_length()?
431 * Fix problems with wrong runtime encoding conversion for NLS message files
432 * Add URL to more complete multi-byte regression tests
434 http://archives.postgresql.org/pgsql-hackers/2005-07/msg00272.php
436 * Fix ILIKE and regular expressions to handle case insensitivity
437 properly in multibyte encodings
439 http://archives.postgresql.org/pgsql-bugs/2005-10/msg00001.php
440 http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php
442 * Set client encoding based on the client operating system encoding
444 Currently client_encoding is set in postgresql.conf, which
445 defaults to the server encoding.
446 http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php
448 * Fix cases where invalid byte encodings are accepted by the database,
449 but throw an error on SELECT
451 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00767.php
457 * Automatically create rules on views so they are updateable, per SQL99
459 We can only auto-create rules for simple views. For more complex
460 cases users will still have to write rules manually.
461 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php
462 http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php
464 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
465 * Allow NOTIFY in rules involving conditionals
466 * Allow VIEW/RULE recompilation when the underlying tables change
468 Another issue is whether underlying table changes should be reflected
469 in the view, e.g. should SELECT * show additional columns if they
470 are added after the view is created.
476 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
477 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
478 * %Allow SET CONSTRAINTS to be qualified by schema/table name
479 * %Add a separate TRUNCATE permission
481 Currently only the owner can TRUNCATE a table because triggers are not
482 called, and the table is locked in exclusive mode.
484 * Allow PREPARE of cursors
485 * Allow finer control over the caching of prepared query plans
487 Currently queries prepared via the libpq API are planned on first
488 execute using the supplied parameters --- allow SQL PREPARE to do the
489 same. Also, allow control over replanning prepared queries either
490 manually or automatically when statistics for execute parameters
491 differ dramatically from those used during planning.
493 * Improve logging of prepared statements recovered during startup
495 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00092.php
497 * Allow LISTEN/NOTIFY to store info in memory rather than tables?
499 Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
500 such information in memory would improve performance.
502 * Add optional textual message to NOTIFY
504 This would allow an informational message to be added to the notify
505 message, perhaps indicating the row modified or other custom
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
523 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00501.php
524 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00536.php
527 * Add NOVICE output level for helpful messages like automatic sequence/index
529 * -Add RESET SESSION command to reset all session state
530 * Add GUC to issue notice about statements that use unjoined tables
531 * Allow EXPLAIN to identify tables that were skipped because of
533 * Allow EXPLAIN output to be more easily processed by scripts
534 * Enable standard_conforming_strings
535 * Make standard_conforming_strings the default in 8.4?
537 When this is done, backslash-quote should be prohibited in non-E''
538 strings because of possible confusion over how such strings treat
539 backslashes. Basically, '' is always safe for a literal single
540 quote, while \' might or might not be based on the backslash
543 * Simplify dropping roles that have objects in several databases
544 * Allow COMMENT ON to accept an expression rather than just a string
545 * Allow the count returned by SELECT, etc to be to represent as an int64
546 to allow a higher range of values
547 * Add SQL99 WITH clause to SELECT
548 * Add SQL:2003 WITH RECURSIVE (hierarchical) queries to SELECT
549 * Add DEFAULT .. AS OWNER so permission checks are done as the table
552 This would be useful for SERIAL nextval() calls and CHECK constraints.
554 * Allow DISTINCT to work in multiple-argument aggregate calls
555 * Add column to pg_stat_activity that shows the progress of long-running
556 commands like CREATE INDEX and VACUUM
557 * Implement SQL:2003 window functions
558 * Improve failure message when DROP DATABASE is used on a database that
559 has prepared transactions
560 * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
562 http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
563 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
565 * Increase locking when DROPing objects so dependent objects cannot
566 get dropped while the DROP operation is happening
568 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php
572 o Allow CREATE TABLE AS to determine column lengths for complex
573 expressions like SELECT col1 || col2
574 o Use more reliable method for CREATE DATABASE to get a consistent
576 o Fix transaction restriction checks for CREATE DATABASE and
579 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00133.php
581 o Have WITH CONSTRAINTS also create constraint indexes
582 http://archives.postgresql.org/pgsql-patches/2007-04/msg00149.php
585 o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
587 http://archives.postgresql.org/pgsql-hackers/2006-07/msg01306.php
588 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00865.php
590 o Research self-referential UPDATEs that see inconsistent row versions
591 in read-committed mode
593 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00507.php
598 o %Have ALTER TABLE RENAME rename SERIAL sequence names
599 o Add ALTER DOMAIN to modify the underlying data type
600 o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
602 http://archives.postgresql.org/pgsql-patches/2006-02/msg00168.php
604 o %Allow ALTER TABLE to change constraint deferrability and actions
605 o Add missing object types for ALTER ... SET SCHEMA
606 o Allow ALTER TABLESPACE to move to different directories
607 o Allow databases to be moved to different tablespaces
608 o Allow moving system tables to other tablespaces, where possible
610 Currently non-global system tables must be in the default database
611 tablespace. Global system tables can never be moved.
613 o Prevent parent tables from altering or dropping constraints
614 like CHECK that are inherited by child tables unless CASCADE
616 o %Prevent child tables from altering or dropping constraints
617 like CHECK that were inherited from the parent table
618 o Have ALTER INDEX update the name of a constraint using that index
619 o Add ALTER TABLE RENAME CONSTRAINT, update index name also
620 o Allow column display reordering by recording a display,
621 storage, and permanent id for every column?
623 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00782.php
629 o -Make CLUSTER preserve recently-dead tuples per MVCC requirements
630 o Automatically maintain clustering on a table
632 This might require some background daemon to maintain clustering
633 during periods of low usage. It might also require tables to be only
634 partially filled for easier reorganization. Another idea would
635 be to create a merged heap/index data file so an index lookup would
636 automatically access the heap data too. A third idea would be to
637 store heap rows in hashed groups, perhaps using a user-supplied
639 http://archives.postgresql.org/pgsql-performance/2004-08/msg00349.php
641 o %Add default clustering to system tables
643 To do this, determine the ideal cluster index for each system
644 table and set the cluster setting during initdb.
646 o %Add VERBOSE option to report tables as they are processed,
648 o -Add more logical syntax CLUSTER table USING index;
649 support current syntax for backward compatibility
654 o Allow COPY to report error lines and continue
656 This requires the use of a savepoint before each COPY line is
657 processed, with ROLLBACK on COPY failure.
659 o Allow COPY on a newly-created table to skip WAL logging
661 On crash recovery, the table involved in the COPY would
662 be removed or have its heap and index files truncated. One
663 issue is that no other backend should be able to add to
664 the table at the same time, which is something that is
670 o Allow column-level privileges
671 o %Allow GRANT/REVOKE permissions to be applied to all schema objects
674 The proposed syntax is:
675 GRANT SELECT ON ALL TABLES IN public TO phpuser;
676 GRANT SELECT ON NEW TABLES IN public TO phpuser;
678 o Allow GRANT/REVOKE permissions to be inherited by objects based on
681 o Allow SERIAL sequences to inherit permissions from the base table?
686 o Allow UPDATE/DELETE WHERE CURRENT OF cursor
688 This requires using the row ctid to map cursor rows back to the
689 original heap row. This become more complicated if WITH HOLD cursors
690 are to be supported because WITH HOLD cursors have a copy of the row
691 and no FOR UPDATE lock.
692 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01014.php
694 o Prevent DROP TABLE from dropping a row referenced by its own open
700 o Allow INSERT/UPDATE of the system-generated oid value for a row
701 o In rules, allow VALUES() to contain a mixture of 'old' and 'new'
707 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
709 o Add SET PATH for schemas?
711 This is basically the same as SET search_path.
714 * Referential Integrity
716 o Add MATCH PARTIAL referential integrity
717 o Change foreign key constraint for array -> element to mean element
719 o Enforce referential integrity for system tables
720 o Fix problem when cascading referential triggers make changes on
721 cascaded tables, seeing the tables in an intermediate state
723 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
724 http://archives.postgresql.org/pgsql-hackers/2005-09/msg00174.php
726 o Allow DEFERRABLE and end-of-statement UNIQUE constraints?
728 This would allow UPDATE tab SET col = col + 1 to work if col has
729 a unique index. Currently, uniqueness checks are done while the
730 command is being executed, rather than at the end of the statement
732 http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
733 http://archives.postgresql.org/pgsql-hackers/2006-09/msg01458.php
736 * Server-Side Languages
739 o Fix RENAME to work on variables other than OLD/NEW
741 http://archives.postgresql.org/pgsql-hackers/2002-03/msg00591.php
742 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01615.php
743 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01587.php
745 o Allow function parameters to be passed by name,
746 get_employee_salary(12345 AS emp_id, 2001 AS tax_year)
747 o Add Oracle-style packages (Pavel)
749 A package would be a schema with session-local variables,
750 public/private functions, and initialization functions. It
751 is also possible to implement these capabilities
752 in all schemas and not use a separate "packages"
754 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00384.php
756 o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
757 o Allow listing of record column names, and access to
758 record columns via variables, e.g. columns := r.(*),
761 http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
762 http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
763 http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
765 o Add single-step debugging of functions
766 o -Add support for MOVE cursors
767 o Add support for SCROLL cursors
768 o Add support for WITH HOLD cursors
769 o Allow RETURN to return row or record functions
771 http://archives.postgresql.org/pgsql-patches/2005-11/msg00045.php
772 http://archives.postgresql.org/pgsql-patches/2006-08/msg00397.php
773 http://archives.postgresql.org/pgsql-hackers/2006-09/msg00388.php
775 o Fix problems with RETURN NEXT on tables with
776 dropped/added columns after function creation
778 http://archives.postgresql.org/pgsql-patches/2006-02/msg00165.php
780 o Allow row and record variables to be set to NULL constants,
781 and allow NULL tests on such variables
783 Because a row is not scalar, do not allow assignment
784 from NULL-valued scalars.
785 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00070.php
788 o Add table function support to pltcl, plpythonu
789 o Add support for polymorphic arguments and return types to
790 languages other than PL/PgSQL
791 o Add capability to create and call PROCEDURES
792 o Add support for OUT and INOUT parameters to languages other
794 o Add PL/PythonU tracebacks
796 http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php
798 o -Allow PL/PythonU to return boolean rather than 1/0
799 o Allow data to be passed in native language formats, rather
802 http://archives.postgresql.org/pgsql-hackers/2007-05/msg00289.php
808 * Have pg_ctl look at PGHOST in case it is a socket directory?
809 * Allow pg_ctl to work properly with configuration files located outside
812 pg_ctl can not read the pid file because it isn't located in the
813 config directory but in the PGDATA directory. The solution is to
814 allow pg_ctl to read and understand postgresql.conf to find the
815 data_directory value.
817 * Make consistent use of long/short command options --- pg_ctl needs
818 long ones, pg_config doesn't have short ones, postgres doesn't have
819 enough long ones, etc.
824 o Have psql show current values for a sequence
825 o Move psql backslash database information into the backend, use
826 mnemonic commands? [psql]
828 This would allow non-psql clients to pull the same information out
829 of the database as psql.
831 o Fix psql's \d commands more consistent
833 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
834 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00014.php
836 o -Allow psql \pset boolean variables to set to fixed values, rather
838 o Consistently display privilege information for all objects in psql
839 o Add auto-expanded mode so expanded output is used if the row
840 length is wider than the screen width.
842 Consider using auto-expanded mode for backslash commands like \df+.
844 o Prevent tab completion of SET TRANSACTION from querying the
845 database and therefore preventing the transaction isolation
846 level from being set.
848 Currently SET <tab> causes a database lookup to check all
849 supported session variables. This query causes problems
850 because setting the transaction isolation level must be the
851 first statement of a transaction.
853 o Consider parsing the -c string into individual queries so each
854 is run in its own transaction
856 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00291.php
858 o Add a \set variable to control whether \s displays line numbers
860 Another option is to add \# which lists line numbers, and
861 allows command execution.
863 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00255.php
868 o %Add dumping of comments on index columns and composite type columns
869 o %Add full object name to the tag field. eg. for operators we need
870 '=(integer, integer)', instead of just '='.
871 o Add pg_dumpall custom format dumps?
872 o Remove unnecessary function pointer abstractions in pg_dump source
874 o Allow selection of individual object(s) of all types, not just
876 o In a selective dump, allow dumping of an object and all its
878 o Add options like pg_restore -l and -L to pg_dump
879 o Stop dumping CASCADE on DROP TYPE commands in clean mode
880 o Allow pg_dump --clean to drop roles that own objects or have
882 o -Add -f to pg_dumpall
889 Document differences between ecpg and the SQL standard and
890 information about the Informix-compatibility module.
892 o Solve cardinality > 1 for input descriptors / variables?
893 o Add a semantic check level, e.g. check if a table really exists
894 o fix handling of DB attributes that are arrays
895 o Use backend PREPARE/EXECUTE facility for ecpg where possible
897 o Fix nested C comments
898 o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
899 o Make SET CONNECTION thread-aware, non-standard?
900 o Allow multidimensional arrays
901 o Add internationalized message strings
902 o Implement COPY FROM STDIN
907 o Add PQescapeIdentifierConn()
908 o Prevent PQfnumber() from lowercasing unquoted the column name
910 PQfnumber() should never have been doing lowercasing, but
911 historically it has so we need a way to prevent it
913 o Allow statement results to be automatically batched to the client
915 Currently all statement results are transferred to the libpq
916 client before libpq makes the results available to the
917 application. This feature would allow the application to make
918 use of the first result rows while the rest are transferred, or
919 held on the server waiting for them to be requested by libpq.
920 One complexity is that a statement like SELECT 1/col could error
921 out mid-way through the result set.
922 o Fix SSL retry to avoid useless repeated connection attempts and
923 ensuing misleading error messages
924 o Consider disallowing multiple queries in PQexec() as an
925 additional barrier to SQL injection attacks
927 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00184.php
929 o Add PQexecf() that allows complex parameter substitution
931 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php
936 * Add deferred trigger queue file
938 Right now all deferred trigger information is stored in backend
939 memory. This could exhaust memory for very large trigger queues.
940 This item involves dumping large queues into files.
942 * Allow triggers to be disabled in only the current session.
944 This is currently possible by starting a multi-statement transaction,
945 modifying the system tables, performing the desired SQL, restoring the
946 system tables, and committing the transaction. ALTER TABLE ...
947 TRIGGER requires a table lock so it is not ideal for this usage.
949 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
951 If the dump is known to be valid, allow foreign keys to be added
952 without revalidating the data.
954 * Allow statement-level triggers to access modified rows
955 * Support triggers on columns
957 http://archives.postgresql.org/pgsql-patches/2005-07/msg00107.php
959 * Allow AFTER triggers on system tables
961 System tables are modified in many places in the backend without going
962 through the executor and therefore not causing triggers to fire. To
963 complete this item, the functions that modify system tables will have
966 * Tighten trigger permission checks
968 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00564.php
970 * Allow BEFORE INSERT triggers on views
972 http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php
979 * -Flush cached query plans when the dependent objects change or
980 when new ANALYZE statistics are available
981 * -Track dependencies in function bodies and recompile/invalidate
982 * -Invalidate prepared queries, like INSERT, when the table definition
989 * Add UNIQUE capability to non-btree indexes
990 * Prevent index uniqueness checks when UPDATE does not modify the column
992 Uniqueness (index) checks are done when updating a column even if the
993 column is not modified by the UPDATE.
995 * Allow the creation of on-disk bitmap indexes which can be quickly
996 combined with other bitmap indexes
998 Such indexes could be more compact if there are only a few distinct values.
999 Such indexes can also be compressed. Keeping such indexes updated can be
1002 http://archives.postgresql.org/pgsql-patches/2005-07/msg00512.php
1003 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01107.php
1005 * -Allow use of indexes to search for NULLs
1006 * Allow accurate statistics to be collected on indexes with more than
1007 one column or expression indexes, perhaps using per-index statistics
1009 http://archives.postgresql.org/pgsql-performance/2006-10/msg00222.php
1010 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01131.php
1012 * -Allow the creation of indexes with mixed ascending/descending
1014 * Consider compressing indexes by storing key values duplicated in
1015 several rows as a single index entry
1017 This is difficult because it requires datatype-specific knowledge.
1019 * Add REINDEX CONCURRENTLY, like CREATE INDEX CONCURRENTLY
1021 This is difficult because you must upgrade to an exclusive table lock
1022 to replace the existing index file. CREATE INDEX CONCURRENTLY does not
1023 have this complication. This would allow index compaction without
1029 o Allow inherited tables to inherit indexes, UNIQUE constraints,
1030 and primary/foreign keys
1031 o Honor UNIQUE INDEX on base column in INSERTs/UPDATEs
1032 on inherited table, e.g. INSERT INTO inherit_table
1033 (unique_index_col) VALUES (dup) should fail
1035 The main difficulty with this item is the problem of
1036 creating an index that can span multiple tables.
1038 o Allow SELECT ... FOR UPDATE on inherited tables
1043 o Add more GIST index support for geometric data types
1044 o Allow GIST indexes to create certain complex index types, like
1045 digital trees (see Aoki)
1050 o Pack hash index buckets onto disk pages more efficiently
1052 Currently only one hash bucket can be stored on a page. Ideally
1053 several hash buckets could be stored on a single page and greater
1054 granularity used for the hash algorithm.
1056 http://archives.postgresql.org/pgsql-hackers/2004-06/msg00168.php
1058 o Consider sorting hash buckets so entries can be found using a
1059 binary search, rather than a linear scan
1060 o In hash indexes, consider storing the hash value with or instead
1062 o Add WAL logging for crash recovery
1063 o Allow multi-column hash indexes
1064 o During index creation, pre-sort the tuples to improve build speed
1066 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
1073 * Improve commit_delay handling to reduce fsync()
1074 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
1076 Ideally this requires a separate test program that can be run
1077 at initdb time or optionally later. Consider O_SYNC when
1080 * %Add an option to sync() before fsync()'ing checkpoint files
1081 * Add program to test if fsync has a delay compared to non-fsync
1082 * Reduce checkpoint performance degredation by forcing data to disk
1085 http://archives.postgresql.org/pgsql-patches/2006-12/msg00104.php
1086 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00337.php
1087 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00079.php
1093 * Allow free-behind capability for large sequential scans, perhaps using
1096 Posix_fadvise() can control both sequential/random file caching and
1097 free-behind behavior, but it is unclear how the setting affects other
1098 backends that also have the file open, and the feature is not supported
1099 on all operating systems.
1103 We could use a fixed row count and a +/- count to follow MVCC
1104 visibility rules, or a single cached value could be used and
1105 invalidated if anyone modifies the table. Another idea is to
1106 get a count directly from a unique index, but for this to be
1107 faster than a sequential scan it must avoid access to the heap
1108 to obtain tuple visibility information.
1110 * Provide a way to calculate an "estimated COUNT(*)"
1112 Perhaps by using the optimizer's cardinality estimates or random
1115 http://archives.postgresql.org/pgsql-hackers/2005-11/msg00943.php
1117 * Allow data to be pulled directly from indexes
1119 Currently indexes do not have enough tuple visibility information
1120 to allow data to be pulled from the index without also accessing
1121 the heap. One way to allow this is to set a bit on index tuples
1122 to indicate if a tuple is currently visible to all transactions
1123 when the first valid heap lookup happens. This bit would have to
1124 be cleared when a heap tuple is expired.
1126 Another idea is to maintain a bitmap of heap pages where all rows
1127 are visible to all backends, and allow index lookups to reference
1128 that bitmap to avoid heap lookups, perhaps the same bitmap we might
1129 add someday to determine which heap pages need vacuuming. Frequently
1130 accessed bitmaps would have to be stored in shared memory. One 8k
1131 page of bitmaps could track 512MB of heap pages.
1133 * Consider automatic caching of statements at various levels:
1136 o Query execute plan
1139 * Allow sequential scans to take advantage of other concurrent
1140 sequential scans, also called "Synchronised Scanning"
1142 One possible implementation is to start sequential scans from the lowest
1143 numbered buffer in the shared cache, and when reaching the end wrap
1144 around to the beginning, rather than always starting sequential scans
1145 at the start of the table.
1147 http://archives.postgresql.org/pgsql-patches/2006-12/msg00076.php
1148 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00408.php
1149 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00784.php
1150 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00415.php
1152 * Consider increasing internal areas when shared buffers is increased
1154 http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php
1156 * Consider decreasing the amount of memory used by PrivateRefCount
1158 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
1159 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1166 * Improve speed with indexes
1168 For large table adjustments during VACUUM FULL, it is faster to cluster
1169 or reindex rather than update the index. Also, index updates can bloat
1171 http://archives.postgresql.org/pgsql-hackers/2007-03/msg00024.php
1172 http://archives.postgresql.org/pgsql-performance/2007-05/msg00296.php
1174 * Reduce lock time during VACUUM FULL by moving tuples with read lock,
1175 then write lock and truncate table
1177 Moved tuples are invisible to other backends so they don't require a
1178 write lock. However, the read lock promotion to write lock could lead
1179 to deadlock situations.
1181 * Auto-fill the free space map by scanning the buffer cache or by
1182 checking pages written by the background writer
1184 http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
1185 http://archives.postgresql.org/pgsql-hackers/2006-03/msg00011.php
1187 * Create a bitmap of pages that need vacuuming
1189 Instead of sequentially scanning the entire table, have the background
1190 writer or some other process record pages that have expired rows, then
1191 VACUUM can look at just those pages rather than the entire table. In
1192 the event of a system crash, the bitmap would probably be invalidated.
1193 One complexity is that index entries still have to be vacuumed, and
1194 doing this without an index scan (by using the heap values to find the
1195 index entry) might be slow and unreliable, especially for user-defined
1198 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01188.php
1199 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00121.php
1201 * Allow FSM to return free space toward the beginning of the heap file,
1202 in hopes that empty pages at the end can be truncated by VACUUM
1203 * Allow FSM page return free space based on table clustering, to assist
1204 in maintaining clustering?
1205 * Consider shrinking expired tuples to just their headers
1207 http://archives.postgresql.org/pgsql-patches/2006-03/msg00142.php
1208 http://archives.postgresql.org/pgsql-hackers/2007-01/msg01025.php
1210 * Allow heap reuse of UPDATEd rows if no indexed columns are changed,
1211 and old and new versions are on the same heap page?
1213 While vacuum handles DELETEs fine, updating of non-indexed columns, like
1214 counters, are difficult for VACUUM to handle efficiently. This method
1215 is possible for same-page updates because a single index row can be
1216 used to point to both old and new values.
1217 http://archives.postgresql.org/pgsql-hackers/2006-06/msg01305.php
1218 http://archives.postgresql.org/pgsql-hackers/2006-06/msg01534.php
1220 * Reuse index tuples that point to heap tuples that are not visible to
1222 * Improve dead row detection during multi-statement transactions usage
1224 http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
1228 o Use free-space map information to guide refilling
1229 o %Issue log message to suggest VACUUM FULL if a table is nearly
1231 o Consider logging activity either to the logs or a system view
1232 o -Turn on by default
1233 o -Allow multiple vacuums so large tables do not starve small
1235 o Improve control of auto-vacuum
1237 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00876.php
1243 * Fix priority ordering of read and write light-weight locks (Neil)
1245 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00893.php
1246 http://archives.postgresql.org/pgsql-hackers/2004-11/msg00905.php
1248 * Fix problem when multiple subtransactions of the same outer transaction
1249 hold different types of locks, and one subtransaction aborts
1251 http://archives.postgresql.org/pgsql-hackers/2006-11/msg01011.php
1252 http://archives.postgresql.org/pgsql-hackers/2006-12/msg00001.php
1253 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00435.php
1255 * Allow UPDATEs on only non-referential integrity columns not to conflict
1256 with referential integrity locks
1258 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php
1260 * Add idle_in_transaction_timeout GUC so locks are not held for long
1265 Startup Time Improvements
1266 =========================
1268 * Experiment with multi-threaded backend for backend creation [thread]
1270 This would prevent the overhead associated with process creation. Most
1271 operating systems have trivial process creation time compared to
1272 database startup overhead, but a few operating systems (Win32,
1273 Solaris) might benefit from threading. Also explore the idea of
1274 a single session using multiple threads to execute a statement faster.
1276 * Experiment with multi-threaded backend better resource utilization
1278 This would allow a single query to make use of multiple CPU's or
1279 multiple I/O channels simultaneously. One idea is to create a
1280 background reader that can pre-fetch sequential and index scan
1281 pages needed by other backends. This could be expanded to allow
1282 concurrent reads from multiple devices in a partitioned table.
1284 * Add connection pooling
1286 It is unclear if this should be done inside the backend code or done
1287 by something external like pgpool. The passing of file descriptors to
1288 existing backends is one of the difficulties with a backend approach.
1294 * Eliminate need to write full pages to WAL before page modification [wal]
1296 Currently, to protect against partial disk page writes, we write
1297 full page images to WAL before they are modified so we can correct any
1298 partial page writes during recovery. These pages can also be
1299 eliminated from point-in-time archive files.
1301 o When off, write CRC to WAL and check file system blocks
1304 If CRC check fails during recovery, remember the page in case
1305 a later CRC for that page properly matches.
1307 o Write full pages during file system write and not when
1308 the page is modified in the buffer cache
1310 This allows most full page writes to happen in the background
1311 writer. It might cause problems for applying WAL on recovery
1312 into a partially-written page, but later the full page will be
1315 * Allow WAL traffic to be streamed to another server for stand-by
1317 * Reduce WAL traffic so only modified values are written rather than
1320 http://archives.postgresql.org/pgsql-hackers/2007-03/msg01589.php
1322 * -Allow the pg_xlog directory location to be specified during initdb
1323 with a symlink back to the /data location
1324 * Allow WAL information to recover corrupted pg_controldata
1326 http://archives.postgresql.org/pgsql-patches/2006-06/msg00025.php
1328 * Find a way to reduce rotational delay when repeatedly writing
1331 Currently fsync of WAL requires the disk platter to perform a full
1332 rotation to fsync again. One idea is to write the WAL to different
1333 offsets that might reduce the rotational delay.
1335 * Allow buffered WAL writes and fsync
1337 Instead of guaranteeing recovery of all committed transactions, this
1338 would provide improved performance by delaying WAL writes and fsync
1339 so an abrupt operating system restart might lose a few seconds of
1340 committed transactions but still be consistent. We could perhaps
1341 remove the 'fsync' parameter (which results in an an inconsistent
1342 database) in favor of this capability.
1344 * Allow WAL logging to be turned off for a table, but the table
1345 might be dropped or truncated during crash recovery [walcontrol]
1347 Allow tables to bypass WAL writes and just fsync() dirty pages on
1348 commit. This should be implemented using ALTER TABLE, e.g. ALTER
1349 TABLE PERSISTENCE [ DROP | TRUNCATE | DEFAULT ]. Tables using
1350 non-default logging should not use referential integrity with
1351 default-logging tables. A table without dirty buffers during a
1352 crash could perhaps avoid the drop/truncate.
1354 * Allow WAL logging to be turned off for a table, but the table would
1355 avoid being truncated/dropped [walcontrol]
1357 To do this, only a single writer can modify the table, and writes
1358 must happen only on new pages so the new pages can be removed during
1359 crash recovery. Readers can continue accessing the table. Such
1360 tables probably cannot have indexes. One complexity is the handling
1361 of indexes on TOAST tables.
1364 Optimizer / Executor
1365 ====================
1367 * Improve selectivity functions for geometric operators
1368 * -Allow ORDER BY ... LIMIT # to select high/low value without sort or
1369 index using a sequential scan for highest/lowest values
1370 * Precompile SQL functions to avoid overhead
1371 * Create utility to compute accurate random_page_cost value
1372 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
1373 * Have EXPLAIN ANALYZE issue NOTICE messages when the estimated and
1374 actual row counts differ by a specified percentage
1375 * Consider using hash buckets to do DISTINCT, rather than sorting
1377 This would be beneficial when there are few distinct values. This is
1378 already used by GROUP BY.
1380 * Log statements where the optimizer row estimates were dramatically
1381 different from the number of rows actually found?
1382 * Consider compressed annealing to search for query plans
1384 This might replace GEQO, http://sixdemonbag.org/Djinni.
1386 * Improve merge join performance by allowing mark/restore of
1389 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00096.php
1393 Miscellaneous Performance
1394 =========================
1396 * Do async I/O for faster random read-ahead of data
1398 Async I/O allows multiple I/O requests to be sent to the disk with
1399 results coming back asynchronously.
1400 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00820.php
1402 * Use mmap() rather than SYSV shared memory or to write WAL files?
1404 This would remove the requirement for SYSV SHM but would introduce
1405 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1406 to prevent I/O overhead.
1408 * Consider mmap()'ing files into a backend?
1410 Doing I/O to large tables would consume a lot of address space or
1411 require frequent mapping/unmapping. Extending the file also causes
1412 mapping problems that might require mapping only individual pages,
1413 leading to thousands of mappings. Another problem is that there is no
1414 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1415 could hit disk before WAL is written.
1417 * Add a script to ask system configuration questions and tune postgresql.conf
1418 * -Merge xmin/xmax/cmin/cmax back into three header fields
1419 * Consider ways of storing rows more compactly on disk
1421 o -Support a smaller header for short variable-length fields
1422 o Reduce the row header size?
1423 o Consider reducing on-disk varlena length from four bytes to
1424 two because a heap row cannot be more than 64k in length
1426 * Consider increasing NUM_CLOG_BUFFERS
1427 * Consider having the background writer update the transaction status
1428 hint bits before writing out the page
1429 * Allow user configuration of TOAST thresholds
1431 http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
1433 * Allow configuration of backend priorities via the operating system
1435 Though backend priorities make priority inversion during lock
1436 waits possible, research shows that this is not a huge problem.
1437 http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
1439 * Consider reducing memory used for shared buffer reference count
1441 http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
1448 * Add use of 'const' for variables in source tree
1449 * Move some things from /contrib into main tree
1450 * %Remove warnings created by -Wcast-align
1451 * Move platform-specific ps status display info from ps_status.c to ports
1452 * Add optional CRC checksum to heap and index pages
1453 * Improve documentation to build only interfaces (Marc)
1454 * Remove or relicense modules that are not under the BSD license, if possible
1455 * %Remove memory/file descriptor freeing before ereport(ERROR)
1456 * Acquire lock on a relation before building a relcache entry for it
1457 * %Promote debug_query_string into a server-side function current_query()
1458 * Allow cross-compiling by generating the zic database on the target system
1459 * Improve NLS maintenance of libpgport messages linked onto applications
1460 * Allow ecpg to work with MSVC and BCC
1461 * Add xpath_array() to /contrib/xml2 to return results as an array
1462 * Clean up casting in /contrib/isn
1464 http://archives.postgresql.org/pgsql-hackers/2006-11/msg00245.php
1466 * Allow building in directories containing spaces
1468 This is probably not possible because 'gmake' and other compiler tools
1469 do not fully support quoting of paths with spaces.
1471 * Fix sgmltools so PDFs can be generated with bookmarks
1472 * Use UTF8 encoding for NLS messages so all server encodings can
1474 * Update Bonjour to work with newer cross-platform SDK
1476 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02238.php
1477 http://archives.postgresql.org/pgsql-patches/2006-10/msg00048.php
1479 * Split out libpq pgpass and environment documentation sections to make
1480 it easier for non-developers to find
1481 * Consider detoasting keys before sorting
1482 * Consider GnuTLS if OpenSSL license becomes a problem
1484 http://archives.postgresql.org/pgsql-patches/2006-05/msg00040.php
1485 http://archives.postgresql.org/pgsql-hackers/2006-12/msg01213.php
1487 * Use strlcpy() rather than our StrNCpy() macro
1489 http://archives.postgresql.org/pgsql-hackers/2006-09/msg02108.php
1491 * Consider changing documentation format from SGML to XML
1493 http://archives.postgresql.org/pgsql-docs/2006-12/msg00152.php
1495 * -Move NAMEDATALEN from postgres_ext.h to pg_config_manual.h
1496 * Consider making NAMEDATALEN more configurable in future releases
1497 * Update our code to handle 64-bit timezone files to match the zic
1498 source code, which now uses them
1499 * Have configure choose integer datetimes by default
1501 http://archives.postgresql.org/pgsql-patches/2007-05/msg00046.php
1503 * -Fix problem with excessive logging during SSL disconnection
1505 http://archives.postgresql.org/pgsql-bugs/2006-12/msg00122.php
1506 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00065.php
1508 * Support scoped IPv6 addresses
1510 http://archives.postgresql.org/pgsql-bugs/2007-05/msg00111.php
1512 * Consider allowing 64-bit integers and floats to be passed by value on
1515 Also change 32-bit floats (float4) to be passed by value at the same
1522 o Remove configure.in check for link failure when cause is found
1523 o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1525 o Remove psql newline patch when we find out why mingw outputs an
1527 o Allow psql to use readline once non-US code pages work with
1529 o Re-enable timezone output on log_line_prefix '%t' when a
1530 shorter timezone string is available
1531 o Fix problem with shared memory on the Win32 Terminal Server
1532 o Improve signal handling
1534 http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
1536 o -Add long file support for binary pg_dump output
1537 o Check WSACancelBlockingCall() for interrupts [win32intr]
1540 * Wire Protocol Changes
1542 o Allow dynamic character set handling
1543 o Add decoded type, length, precision
1545 o Update clients to use data types, typmod, schema.table.column names
1546 of result sets using new statement protocol
1552 * Add pre-parsing phase that converts non-ISO syntax to supported
1555 This could allow SQL written for other databases to run without
1558 * Allow plug-in modules to emulate features from other databases
1559 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
1561 * Allow statements across databases or servers with transaction
1564 This can be done using dblink and two-phase commit.
1566 * Add the features of packages
1568 o Make private objects accessible only to objects in the same schema
1569 o Allow current_schema.objname to access current schema objects
1570 o Add session variables
1571 o Allow nested schemas
1573 * Consider allowing control of upper/lower case folding of unquoted
1576 http://archives.postgresql.org/pgsql-hackers/2004-04/msg00818.php
1577 http://archives.postgresql.org/pgsql-hackers/2006-10/msg01527.php
1580 Features We Do _Not_ Want
1581 =========================
1583 * All backends running as threads in a single process (not wanted)
1585 This eliminates the process protection we get from the current setup.
1586 Thread creation is usually the same overhead as process creation on
1587 modern systems, so it seems unwise to use a pure threaded model.
1589 * Optimizer hints (not wanted)
1591 Optimizer hints are used to work around problems in the optimizer. We
1592 would rather have the problems reported and fixed.
1593 http://archives.postgresql.org/pgsql-hackers/2006-08/msg00506.php
1594 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00517.php
1595 http://archives.postgresql.org/pgsql-hackers/2006-10/msg00663.php
1597 * Allow AS in "SELECT col AS label" to be optional (not wanted)
1599 Because we support postfix operators, it isn't possible to make AS
1600 optional and continue to use bison.
1601 http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php
1603 * Embedded server (not wanted)
1605 While PostgreSQL clients runs fine in limited-resource environments, the
1606 server requires multiple processes and a stable pool of resources to
1607 run reliabily and efficiently. Stripping down the PostgreSQL server
1608 to run in the same process address space as the client application
1609 would add too much complexity and failure cases.