1 TODO list for PostgreSQL
2 ========================
3 Last updated: Sun May 2 08:26:56 EDT 2004
5 Current maintainer: Bruce Momjian (pgman@candle.pha.pa.us)
7 The most recent version of this document can be viewed at
8 the PostgreSQL web site, http://www.PostgreSQL.org.
10 A dash (-) marks changes that will appear in the upcoming 7.5 release.
12 Bracketed items "[]" have more detailed.
18 * Add replication of distributed databases
21 o Master/slave replication
22 o Multi-master replication
23 o Partition data across servers
24 o Queries across databases or servers (two-phase commit)
25 o Allow replication over unreliable or non-persistent links
26 * Point-in-time data recovery using backup and write-ahead log,
27 http://momjian.postgresql.org/main/writings/pgsql/project/pitr.html
28 * Create native Win32 port, http://momjian.postgresql.org/main/writings/pgsql/project/win32.html
35 * Remove unreferenced table files and temp tables during database vacuum
36 or postmaster startup (Bruce)
37 * Remove behavior of postmaster -o after making postmaster/postgres
39 * Allow configuration files to be specified in a different directory
40 * Allow limits on per-db/user connections
41 * Add group object ownership, so groups can rename/drop/grant on objects,
42 so we can implement roles
43 * Add the concept of dataspaces/tablespaces [tablespaces] (Gavin)
44 * -Allow logging of only data definition(DDL), or DDL and modification statements
45 * -Allow log lines to include session-level information, like database and user
46 * Allow server log information to be output as INSERT statements
47 * Prevent default re-use of sysids for dropped users and groups
48 * Prevent dropping user that still owns objects, or auto-drop the objects
49 * Allow pooled connections to query prepared queries
50 * Allow pooled connections to close all open WITH HOLD cursors
51 * Allow major upgrades without dump/reload, perhaps using pg_upgrade
52 * Have SHOW ALL and pg_settings show descriptions for server-side variables(Joe)
53 * Allow external interfaces to extend the GUC variable set
59 * Remove Money type, add money formatting for decimal type
60 * -Change factorial to return a numeric (Gavin)
61 * Change NUMERIC to enforce the maximum precision, and increase it
62 * Add function to return compressed length of TOAST data values (Tom)
63 * Allow INET subnet tests using non-constants to be indexed
64 * Add transaction_timestamp(), statement_timestamp(), clock_timestamp() functionality
65 * Have sequence dependency track use of DEFAULT sequences, seqname.nextval
66 * Disallow changing default expression of a SERIAL column
67 * Allow infinite dates just like infinite timestamps
68 * Allow pg_dump to dump sequences using NO_MAXVALUE and NO_MINVALUE
69 * Allow backend to output result sets in XML
70 * Prevent whole-row references from leaking memory, e.g. SELECT COUNT(tab.*)
71 * Have initdb set DateStyle based on locale?
72 * Add pg_get_acldef(), pg_get_typedefault(), and pg_get_attrdef()
73 * Add ALTER DOMAIN, AGGREGATE, CONVERSION, SEQUENCE ... OWNER TO
74 * Allow to_char to print localized month names (Karel)
75 * Allow functions to have a search path specified at creation time
76 * Make LENGTH() of CHAR() not count trailing spaces
77 * Allow substring/replace() to get/set bit values
78 * Add GUC variable to allow output of interval values in ISO8601 format
82 o Allow nulls in arrays
83 o Allow MIN()/MAX() on arrays
84 o Delay resolution of array expression type so assignment coercion
85 can be performed on empty array expressions (Joe)
86 o Modify array literal representation to handle array index lower bound
91 o Improve vacuum of large objects, like /contrib/vacuumlo
92 o Add security checking for large objects
93 o Make file in/out interface for TOAST columns, similar to large object
94 interface (force out-of-line storage and no compression)
95 o Auto-delete large objects when referencing row is deleted
98 Multi-Language Support
99 ======================
101 * Add NCHAR (as distinguished from ordinary varchar),
102 * Allow locale to be set at database creation
103 * Allow locale on a per-column basis, default to ASCII
104 * Optimize locale to have minimal performance impact when not used (Peter E)
105 * Support multiple simultaneous character sets, per SQL92
106 * Improve Unicode combined character handling
107 * Add octet_length_server() and octet_length_client() (Thomas, Tatsuo)
108 * Make octet_length_client the same as octet_length() (?)
109 * Prevent mismatch of frontend/backend encodings from converting bytea
110 data from being interpreted as encoded strings
111 * Fix upper()/lower() to work for multibyte encodings
117 * Automatically create rules on views so they are updateable, per SQL92 [view]
118 * Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
119 * Allow NOTIFY in rules involving conditionals
120 * Have views on temporary tables exist in the temporary namespace
121 * Allow temporary views on non-temporary tables
122 * Allow RULE recompilation
128 * -Order duplicate index entries on creation by tid for faster heap lookups
129 * Allow inherited tables to inherit index, UNIQUE constraint, and primary
130 key, foreign key [inheritance]
131 * UNIQUE INDEX on base column not honored on inserts from inherited table
132 INSERT INTO inherit_table (unique_index_col) VALUES (dup) should fail
134 * Add UNIQUE capability to non-btree indexes
135 * Add rtree index support for line, lseg, path, point
136 * Use indexes for min() and max() or convert to SELECT col FROM tab ORDER
137 BY col DESC LIMIT 1 if appropriate index exists and WHERE clause acceptible
138 * Use index to restrict rows returned by multi-key index when used with
139 non-consecutive keys or OR clauses, so fewer heap accesses
140 * Be smarter about insertion of already-ordered data into btree index
141 * Prevent index uniqueness checks when UPDATE does not modify the column
142 * Use bitmaps to fetch heap pages in sequential order [performance]
143 * Use bitmaps to combine existing indexes [performance]
144 * Allow use of indexes to search for NULLs
145 * Allow SELECT * FROM tab WHERE int2col = 4 to use int2col index, int8,
146 float4, numeric/decimal too
147 * Add FILLFACTOR to btree index creation
148 * Add concurrency to GIST
149 * Allow a single index to index multiple tables (for inheritance and subtables)
155 * Add BETWEEN ASYMMETRIC/SYMMETRIC (Christopher)
156 * Change LIMIT/OFFSET to use int8
157 * CREATE TABLE AS can not determine column lengths from expressions [atttypmod]
158 * Allow UPDATE to handle complex aggregates [update]
159 * Allow command blocks to ignore certain types of errors
160 * Allow backslash handling in quoted strings to be disabled for portability
161 * Allow UPDATE, DELETE to handle table aliases for self-joins [delete]
162 * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
163 * Allow REINDEX to rebuild all indexes, remove /contrib/reindex
164 * Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
165 * Add schema option to createlang
166 * Allow savepoints / nested transactions [transactions] (Alvaro)
167 * Use nested transactions to prevent syntax errors from aborting a transaction
168 * Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple columns
169 * Allow SET CONSTRAINTS to be qualified by schema/table
170 * Prevent COMMENT ON DATABASE from using a database name
172 * Allow TRUNCATE ... CASCADE/RESTRICT
173 * Allow PREPARE of cursors
174 * Allow LISTEN/NOTIFY to store info in memory rather than tables
175 * -COMMENT ON [ CAST | CONVERSION | OPERATOR CLASS | LARGE OBJECT | LANGUAGE ]
177 * Dump large object comments in custom dump format
178 * Add optional textual message to NOTIFY
179 * Allow more ISOLATION LEVELS to be accepted, but issue a warning for them
180 * Allow CREATE TABLE foo (f1 INT CHECK (f1 > 0) CHECK (f1 < 10)) to work
181 by searching for non-conflicting constraint names, and prefix with table name
182 * Use more reliable method for CREATE DATABASE to get a consistent copy of db
183 * -Have psql \dn show only visible temp schemas using current_schemas()
184 * -Have psql '\i ~/<tab><tab>' actually load files it displays from home dir
185 * Ignore temporary tables from other session when processing inheritance
186 * Add GUC setting to make created tables default to WITHOUT OIDS
187 * Have pg_ctl look at PGHOST in case it is a socket directory
188 * Allow column-level privileges
189 * Add a session mode to warn about non-standard SQL usage
192 o ALTER TABLE ADD COLUMN does not honor DEFAULT and non-CHECK CONSTRAINT
193 o ALTER TABLE ADD COLUMN column DEFAULT should fill existing
194 rows with DEFAULT value
195 o ALTER TABLE ADD COLUMN column SERIAL doesn't create sequence because
197 o Have ALTER TABLE rename SERIAL sequences
198 o Allow columns to be reordered using ALTER ... POSITION i col1 [,col2];
199 have SELECT * and INSERT honor such ordering
200 o Allow ALTER TABLE to modify column lengths and change to binary
202 o Add ALTER DATABASE ... OWNER TO newowner
203 o Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
204 o Allow ALTER TABLE to change constraint deferrability and actions
205 o Disallow dropping of an inherited constraint
206 o Allow the schema of objects to be changed
209 o Automatically maintain clustering on a table
210 o Add ALTER TABLE table SET WITHOUT CLUSTER
213 o -Allow dump/load of CSV format
214 o Allow COPY to report error lines and continue; optionally
215 allow error codes to be specified; requires savepoints or can
216 not be run in a multi-statement transaction
217 o Allow COPY to understand \x as hex
218 o Have COPY return number of rows loaded/unloaded
221 o Allow UPDATE/DELETE WHERE CURRENT OF cursor using per-cursor tid
222 stored in the backend (Gavin)
223 o Prevent DROP of table being referenced by our own open cursor
226 o Allow INSERT/UPDATE of system-generated oid value for a row
227 o Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
228 o Allow INSERT/UPDATE ... RETURNING new.col or old.col; handle
232 o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
234 o Add SET PATH for schemas
236 * SERVER-SIDE LANGUAGES
237 o Allow PL/PgSQL's RAISE function to take expressions
238 o Change PL/PgSQL to use palloc() instead of malloc()
239 o Allow Java server-side programming [java] (Dave)
240 o Fix problems with complex temporary table creation/destruction
241 without using PL/PgSQL EXECUTE, needs cache prevention/invalidation
242 o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
243 o Improve PL/PgSQL exception handling
244 o Allow parameters to be specified by name and type during definition
245 o Allow function parameters to be passed by name,
246 get_employee_salary(emp_id => 12345, tax_year => 2001)
247 o Add PL/PgSQL packages
248 o Add table function support to pltcl, plperl, plpython
250 o Allow PL/pgSQL to name columns by ordinal position, e.g. rec.(3)
251 o Allow PL/pgSQL EXECUTE query_var INTO record_var;
252 o Add capability to create and call PROCEDURES
253 o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
259 * Add XML capability to pg_dump and COPY, when backend XML capability
260 * -Allow psql \du to show users, and add \dg for groups
261 * Allow clients to query a list of WITH HOLD cursors and prepared statements
262 * Add a libpq function to support Parse/DescribeStatement capability
263 * Prevent libpq's PQfnumber() from lowercasing the column name
264 * -Allow pg_dump to dump CREATE CONVERSION (Christopher)
265 * Allow fastpast to pass values in portable format
266 * Allow libpq to return information about prepared queries
267 * Make pg_restore continue after errors, so it acts more like pg_dump scripts
268 * Have psql show more information about sequences
269 * Allow pg_dumpall to use non-text output formats
270 * Have pg_dump use multi-statement transactions for INSERT dumps
271 * Move psql backslash database information into the backend, use nmumonic
276 o Comprehensive test suite. This may be available already.
277 o JDBC-standard BLOB support
278 o Error Codes (pending backend implementation)
279 o Support both 'make' and 'ant'
280 o Fix LargeObject API to handle OIDs as unsigned ints
281 o Use cursors implicitly to avoid large results (see setCursorName())
282 o Add LISTEN/NOTIFY support to the JDBC driver (Barry)
286 o Implement set descriptor, using descriptor
287 o Solve cardinality > 1 for input descriptors / variables
288 o Improve error handling
289 o Add a semantic check level, e.g. check if a table really exists
290 o fix handling of DB attributes that are arrays
291 o Use backend prepare/execute facility for ecpg where possible
293 o Fix nested C comments
294 o sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
295 o Make SET CONNECTION thread-aware, non-standard?
296 o Allow multidimensional arrays
299 o Allow users to register their own types with pg_
300 o Allow SELECT to return a dictionary of dictionaries
301 o Allow COPY BINARY FROM
304 Referential Integrity
305 =====================
307 * Add MATCH PARTIAL referential integrity
308 * Add deferred trigger queue file (Jan)
309 * Implement dirty reads or shared row locks and use them in RI triggers
310 * Enforce referential integrity for system tables
311 * Change foreign key constraint for array -> element to mean element
313 * Allow DEFERRABLE UNIQUE constraints
314 * Allow triggers to be disabled [trigger]
315 * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
316 * Allow statement-level triggers to access modified rows
317 * Support triggers on columns (Neil)
318 * Have AFTER triggers execute after the appropriate SQL statement in a
319 function, not at the end of the function
320 * Print table names with constraint names in error messages, or make constraint
321 names unique within a schema
322 * -Issue NOTICE if foreign key data requires costly test to match primary key
323 * Remove CREATE CONSTRAINT TRIGGER
324 * Allow AFTER triggers on system tables
330 * Flush cached query plans when their underlying catalog data changes
331 * -Use dependency information to dump data in proper order
332 * Have pg_dump -c clear the database using dependency information
338 * Add SQL99 WITH clause to SELECT (Tom, Fernando)
339 * Add SQL99 WITH RECURSIVE to SELECT (Tom, Fernando)
340 * Add pre-parsing phase that converts non-ANSI features to supported features
341 * Allow plug-in modules to emulate features from other databases
342 * SQL*Net listener that makes PostgreSQL appear as an Oracle database
344 * Add two-phase commit to all distributed transactions with
345 offline/readonly server status or administrator notification for failure
346 * Allow cross-db queries with transaction semantics
356 * Delay fsync() when other backends are about to commit too
357 o Determine optimal commit_delay value
358 * Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
359 o Allow multiple blocks to be written to WAL with one write()
364 * Shared catalog cache, reduce lseek()'s by caching table size in shared area
365 * Add free-behind capability for large sequential scans [fadvise]
366 * Consider use of open/fcntl(O_DIRECT) to minimize OS caching
367 * Cache last known per-tuple offsets to speed long tuple access
368 * Automatically place fixed-width, NOT NULL columns first in a table
369 * Use a fixed row count and a +/- count with MVCC visibility rules
370 to allow fast COUNT(*) queries with no WHERE clause(?) [count]
376 * Improve speed with indexes (perhaps recreate index instead)
377 * Reduce lock time by moving tuples with read lock, then write
378 lock and truncate table
379 * Provide automatic running of vacuum in the background in backend
380 rather than in /contrib
381 * Allow free space map to be auto-sized or warn when it is too small
382 * Maintain a map of recently-expired of pages so vacuum can reclaim
383 free space without a sequential scan
384 * Have VACUUM FULL use REINDEX rather than index vacuum
390 * Make locking of shared data structures more fine-grained
391 * Add code to detect an SMP machine and handle spinlocks accordingly
392 from distributted.net, http://www1.distributed.net/source,
393 in client/common/cpucheck.cpp
394 * Research use of sched_yield() for spinlock acquisition failure
400 * Experiment with multi-threaded backend [thread]
401 * Add connection pooling [pool]
402 * Allow persistent backends [pool]
403 * Create a transaction processor to aid in persistent connections and
404 connection pooling [pool]
405 * Do listen() in postmaster and accept() in pre-forked backend
406 * Have pre-forked backend pre-connect to last requested database or pass
407 file descriptor to backend pre-forked for matching database
413 * Have after-change WAL write()'s write only modified data to kernel
414 * Reduce number of after-change WAL writes; they exist only to gaurd against
415 partial page writes [wal]
416 * Turn off after-change writes if fsync is disabled (?)
417 * Add WAL index reliability improvement to non-btree indexes
418 * Find proper defaults for postgresql.conf WAL entries
419 * Allow xlog directory location to be specified during initdb, perhaps
421 * Allow WAL information to recover corrupted pg_controldata
422 * Find a way to reduce rotational delay when repeatedly writing
429 * Missing optimizer selectivities for date, r-tree, etc
430 * Allow ORDER BY ... LIMIT to select top values without sort or index
431 using a sequential scan for highest/lowest values (Oleg)
432 * Precompile SQL functions to avoid overhead (Neil)
433 * Add utility to compute accurate random_page_cost value
434 * Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
435 * Use CHECK constraints to improve optimizer decisions
436 * Check GUC geqo_threshold to see if it is still accurate
437 * Allow sorting, temp files, temp tables to use multiple work directories
438 * Improve the planner to use CHECK constraints to prune the plan (for subtables)
439 * Have EXPLAIN ANALYZE highlight poor optimizer estimates
445 * Do async I/O for faster random read-ahead of data
446 * Use mmap() rather than SYSV shared memory or to write WAL files (?) [mmap]
447 * Improve caching of attribute offsets when NULLs exist in the row
448 * Add a script to ask system configuration questions and tune postgresql.conf
449 * Allow partitioning of table into multiple subtables
450 * -Use background process to write dirty shared buffers to disk
456 * Add use of 'const' for variables in source tree
457 * Rename some /contrib modules from pg* to pg_*
458 * Move some things from /contrib into main tree
459 * Remove warnings created by -Wcast-align
460 * Move platform-specific ps status display info from ps_status.c to ports
461 * Improve access-permissions check on data directory in Cygwin (Tom)
462 * Add documentation for perl, including mention of DBI/DBD perl location
463 * Create improved PostgreSQL introductory documentation for the PHP
465 * Add optional CRC checksum to heap and index pages
466 * Change representation of whole-tuple parameters to functions
467 * Clarify use of 'application' and 'command' tags in SGML docs
468 * Better document ability to build only certain interfaces (Marc)
469 * Remove or relicense modules that are not under the BSD license, if possible
470 * Remove memory/file descriptor freeing before ereport(ERROR) (Bruce)
471 * Acquire lock on a relation before building a relcache entry for it
472 * Research interaction of setitimer() and sleep() used by statement_timeout
473 * -Add checks for fclose() failure (Tom)
474 * -Change CVS ID to PostgreSQL
475 * -Exit postmaster if postgresql.conf can not be opened
476 * Rename /scripts directory because they are all C programs now
477 * Allow the regression tests to start postmaster with -i so the tests
478 can be run on systems that don't support unix-domain sockets
479 * Allow creation of a libpq-only tarball
480 * Promote debug_query_string into a server-side function current_query()
481 * Allow the identifier length to be increased via a configure option
482 * Improve CREATE SCHEMA regression test
485 * Wire Protocol Changes
486 o Dynamic character set handling
487 o Add decoded type, length, precision
489 o Update clients to use data types, typmod, schema.table.column names of
490 result sets using new query protocol
492 ---------------------------------------------------------------------------
495 Developers who have claimed items are:
496 --------------------------------------
497 * Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
498 * Barry is Barry Lind <barry@xythos.com>
499 * Billy is Billy G. Allie <Bill.Allie@mug.org>
500 * Bruce is Bruce Momjian <pgman@candle.pha.pa.us> of Software Research Assoc.
501 * Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
502 Family Health Network
503 * D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
504 * Dave is Dave Cramer <dave@fastcrypt.com>
505 * Edmund is Edmund Mergl <E.Mergl@bawue.de>
506 * Fernando is Fernando Nasser <fnasser@redhat.com> of Red Hat
507 * Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
508 * Greg is Greg Sabino Mullane <greg@turnstep.com>
509 * Hiroshi is Hiroshi Inoue <Inoue@tpf.co.jp>
510 * Karel is Karel Zak <zakkr@zf.jcu.cz>
511 * Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
512 * Joe is Joe Conway <mail@joeconway.com>
513 * Liam is Liam Stewart <liams@redhat.com> of Red Hat
514 * Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
515 * Mark is Mark Hollomon <mhh@mindspring.com>
516 * Michael is Michael Meskes <meskes@postgresql.org> of Credativ
517 * Neil is Neil Conway <neilc@samurai.com>
518 * Oleg is Oleg Bartunov <oleg@sai.msu.su>
519 * Peter M is Peter T Mount <peter@retep.org.uk> of Retep Software
520 * Peter E is Peter Eisentraut <peter_e@gmx.net>
521 * Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
522 * Rod is Rod Taylor <pg@rbt.ca>
523 * Ross is Ross J. Reedstrom <reedstrm@wallace.ece.rice.edu>
524 * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
525 * Tatsuo is Tatsuo Ishii <t-ishii@sra.co.jp> of Software Research Assoc.
526 * Thomas is Thomas Lockhart <lockhart@fourpalms.org> of Jet Propulsion Labratory
527 * Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat