1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
2 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
5 <title>PostgreSQL TODO List</title>
6 <meta name="generator" content="HTML::TextToHTML v2.25"/>
8 <body bgcolor="#FFFFFF" text="#000000" link="#FF0000" vlink="#A00000" alink="#0000FF">
9 <h1><a name="section_1">PostgreSQL TODO List</a></h1>
10 <p>Current maintainer: Bruce Momjian (<a href="mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</a>)<br/>
11 Last updated: Sun Oct 2 20:42:45 EDT 2005
13 <p>The most recent version of this document can be viewed at<br/>
14 <a href="http://www.postgresql.org/docs/faqs.TODO.html">http://www.postgresql.org/docs/faqs.TODO.html</a>.
16 <p><strong>A hyphen, "-", marks changes that will appear in the upcoming 8.1 release.</strong><br/>
17 <strong>A percent sign, "%", marks items that are easier to implement.</strong>
19 <p>Bracketed items, "[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?"></a>]", have more detail.
21 <p>This list contains all known PostgreSQL bugs and feature requests. If<br/>
22 you would like to work on an item, please read the Developer's FAQ<br/>
25 <h1><a name="section_2">Administration</a></h1>
28 <li>%Remove behavior of postmaster -o after making postmaster/postgres
30 </li><li>%Allow pooled connections to list all prepared queries
31 <p> This would allow an application inheriting a pooled connection to know
32 the queries prepared in the current session.
34 </li><li>Allow major upgrades without dump/reload, perhaps using pg_upgrade
35 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pg_upgrade">pg_upgrade</a>]
36 </li><li>Check for unreferenced table files created by transactions that were
37 in-progress when the server terminated abruptly
38 </li><li>Allow administrators to safely terminate individual sessions either
39 via an SQL function or SIGTERM
40 <p> Lock table corruption following SIGTERM of an individual backend
41 has been reported in 8.0. A possible cause was fixed in 8.1, but
42 it is unknown whether other problems exist. This item mostly
43 requires additional testing rather than of writing any new code.
45 </li><li>%Set proper permissions on non-system schemas during db creation
46 <p> Currently all schemas are owned by the super-user because they are
47 copied from the template1 database.
49 </li><li>Support table partitioning that allows a single table to be stored
50 in subtables that are partitioned based on the primary key or a WHERE
52 </li><li>Add function to report the time of the most recent server reload
53 </li><li>Improve replication solutions
56 <p> You can use any of the master/slave replication servers to use a
57 standby server for data warehousing. To allow read/write queries to
58 multiple servers, you need multi-master replication like pgcluster.
60 </li><li>Allow replication over unreliable or non-persistent links
62 </li><li>Configuration files
64 <li>%Add "include file" functionality in postgresql.conf
65 </li><li>%Allow commenting of variables in postgresql.conf to restore them
67 <p> Currently, if a variable is commented out, it keeps the
68 previous uncommented value until a server restarted.
70 </li><li>%Allow pg_hba.conf settings to be controlled via SQL
71 <p> This would add a function to load the SQL table from
72 pg_hba.conf, and one to writes its contents to the flat file.
73 The table should have a line number that is a float so rows
74 can be inserted between existing rows, e.g. row 2.5 goes
75 between row 2 and row 3.
77 </li><li>%Allow postgresql.conf file values to be changed via an SQL
78 API, perhaps using SET GLOBAL
79 </li><li>Allow the server to be stopped/restarted via an SQL API
80 </li><li>Issue a warning if a change-on-restart-only postgresql.conf value
81 is modified and the server config files are reloaded
82 </li><li>Mark change-on-restart-only values in postgresql.conf
86 <li>Allow a database in tablespace t1 with tables created in
87 tablespace t2 to be used as a template for a new database created
88 with default tablespace t2
89 <p> All objects in the default database tablespace must have default
90 tablespace specifications. This is because new databases are
91 created by copying directories. If you mix default tablespace
92 tables and tablespace-specified tables in the same directory,
93 creating a new database from such a mixed directory would create a
94 new database with tables that had incorrect explicit tablespaces.
95 To fix this would require modifying pg_class in the newly copied
96 database, which we don't currently do.
98 </li><li>Allow reporting of which objects are in which tablespaces
99 <p> This item is difficult because a tablespace can contain objects
100 from multiple databases. There is a server-side function that
101 returns the databases which use a specific tablespace, so this
102 requires a tool that will call that function and connect to each
103 database to find the objects in each database for that tablespace.
106 <li>%Add a GUC variable to control the tablespace for temporary objects
108 <p> It could start with a random tablespace from a supplied list and
109 cycle through the list.
111 </li><li>Allow WAL replay of CREATE TABLESPACE to work when the directory
112 structure on the recovery computer is different from the original
113 </li><li>Allow per-tablespace quotas
116 </li><li>Point-In-Time Recovery (PITR)
118 <li>Allow point-in-time recovery to archive partially filled
119 write-ahead logs [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
120 <p> Currently only full WAL files are archived. This means that the
121 most recent transactions aren't available for recovery in case
122 of a disk failure. This could be triggered by a user command or
125 </li><li>Automatically force archiving of partially-filled WAL files when
126 pg_stop_backup() is called or the server is stopped
127 <p> Doing this will allow administrators to know more easily when
128 the archive contins all the files needed for point-in-time
131 </li><li>%Create dump tool for write-ahead logs for use in determining
132 transaction id for point-in-time recovery
133 </li><li>Allow a warm standby system to also allow read-only queries
134 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
135 <p> This is useful for checking PITR recovery.
137 </li><li>Allow the PITR process to be debugged and data examined
140 <h1><a name="section_3">Monitoring</a></h1>
143 <li>Allow server log information to be output as INSERT statements
144 <p> This would allow server log information to be easily loaded into
145 a database for analysis.
147 </li><li>%Add ability to monitor the use of temporary sort files
148 </li><li>Allow server logs to be remotely read and removed using SQL commands
149 </li><li>Allow protocol-level BIND parameter values to be logged
151 <h1><a name="section_4">Data Types</a></h1>
154 <li>Improve the MONEY data type
155 <p> Change the MONEY data type to use DECIMAL internally, with special
156 locale-aware output formatting.
158 </li><li>Change NUMERIC to enforce the maximum precision, and increase it
159 </li><li>Add NUMERIC division operator that doesn't round?
160 <p> Currently NUMERIC _rounds_ the result to the specified precision.
161 This means division can return a result that multiplied by the
162 divisor is greater than the dividend, e.g. this returns a value > 10:
165 <p> SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
167 <p> The positive modulus result returned by NUMERICs might be considered<br/>
168 inaccurate, in one sense.
171 <li>%Disallow changing default expression of a SERIAL column?
172 </li><li>Fix data types where equality comparison isn't intuitive, e.g. box
173 </li><li>%Prevent INET cast to CIDR if the unmasked bits are not zero, or
175 </li><li>%Prevent INET cast to CIDR from droping netmask, SELECT '<a href="telnet://1.1.1.1">1.1.1.1</a>'::inet::cidr
176 </li><li>Allow INET + INT4 to increment the host part of the address, or
177 throw an error on overflow
178 </li><li>%Add 'tid != tid ' operator for use in corruption recovery
179 </li><li>Allow user-defined types to specify a type modifier at table creation
181 </li><li>Dates and Times
183 <li>Allow infinite dates just like infinite timestamps
184 </li><li>Add a GUC variable to allow output of interval values in ISO8601
186 </li><li>Merge hardwired timezone names with the TZ database; allow either
187 kind everywhere a TZ name is currently taken
188 </li><li>Allow customization of the known set of TZ names (generalize the
189 present australian_timezones hack)
190 </li><li>Allow TIMESTAMP WITH TIME ZONE to store the original timezone
191 information, either zone name or offset from UTC [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?timezone">timezone</a>]
192 <p> If the TIMESTAMP value is stored with a time zone name, interval
193 computations should adjust based on the time zone rules.
195 </li><li>Fix SELECT '0.01 years'::interval, '0.01 months'::interval
196 </li><li>Fix SELECT INTERVAL '1' MONTH
197 </li><li>Add ISO INTERVAL handling
199 <li>Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
201 </li><li>Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
202 </li><li>For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
203 '1', treat as ISO if there is a range specification clause,
204 and as PG if there no clause is present, e.g. interpret
205 <p> '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
206 interpret '1:30' as '1 hour, 30 minutes'
207 <li>Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
208 INTERVAL MONTH), and this should return '12 months'
209 </li><li>Round or truncate values to the requested precision, e.g.
210 INTERVAL '11 months' AS YEAR should return one or zero
211 </li><li>Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
216 <li>Allow NULLs in arrays
217 </li><li>Delay resolution of array expression's data type so assignment
218 coercion can be performed on empty array expressions
222 <li>Improve vacuum of large objects, like /contrib/vacuumlo?
223 </li><li>Add security checking for large objects
224 </li><li>Auto-delete large objects when referencing row is deleted
225 <p> /contrib/lo offers this functionality.
227 </li><li>Allow read/write into TOAST values like large objects
228 <p> This requires the TOAST column to be stored EXTERNAL.
232 <h1><a name="section_5">Functions</a></h1>
235 <li>Allow INET subnet tests using non-constants to be indexed
236 </li><li>Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
238 <p> Current CURRENT_TIMESTAMP returns the start time of the current
239 transaction, and gettimeofday() returns the wallclock time. This will
240 make time reporting more consistent and will allow reporting of
241 the statement start time.
243 </li><li>%Add pg_get_acldef(), pg_get_typedefault(), and pg_get_attrdef()
244 </li><li>Allow to_char() to print localized month names
245 </li><li>Allow functions to have a schema search path specified at creation time
246 </li><li>Allow substring/replace() to get/set bit values
247 </li><li>Allow to_char() on interval values to accumulate the highest unit
249 <p> Some special format flag would be required to request such
250 accumulation. Such functionality could also be added to EXTRACT.
251 Prevent accumulation that crosses the month/day boundary because of
252 the uneven number of days in a month.
255 <li>to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
256 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
257 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
258 </li><li>to_char(INTERVAL '3 years 5 months','MM') => 41
260 </li><li>Add sleep() function, remove from regress.c
262 <h1><a name="section_6">Multi-Language Support</a></h1>
265 <li>Add NCHAR (as distinguished from ordinary varchar),
266 </li><li>Allow locale to be set at database creation
267 <p> Currently locale can only be set during initdb. No global tables have
268 locale-aware columns. However, the database template used during
269 database creation might have locale-aware indexes. The indexes would
270 need to be reindexed to match the new locale.
272 </li><li>Allow encoding on a per-column basis
273 <p> Right now only one encoding is allowed per database.
275 </li><li>Support multiple simultaneous character sets, per SQL92
276 </li><li>Improve UTF8 combined character handling?
277 </li><li>Add octet_length_server() and octet_length_client()
278 </li><li>Make octet_length_client() the same as octet_length()?
279 </li><li>Fix problems with wrong runtime encoding conversion for NLS message files
281 <h1><a name="section_7">Views / Rules</a></h1>
284 <li>%Automatically create rules on views so they are updateable, per SQL99
285 <p> We can only auto-create rules for simple views. For more complex
286 cases users will still have to write rules.
288 </li><li>Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
289 </li><li>Allow NOTIFY in rules involving conditionals
290 </li><li>Allow VIEW/RULE recompilation when the underlying tables change
291 <p> Another issue is whether underlying table changes should be reflected
292 in the view, e.g. should SELECT * show additional columns if they
293 are added after the view is created.
296 <h1><a name="section_8">SQL Commands</a></h1>
299 <li>Change LIMIT/OFFSET and FETCH/MOVE to use int8
300 </li><li>Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
301 </li><li>Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
302 </li><li>%Allow SET CONSTRAINTS to be qualified by schema/table name
303 </li><li>%Allow TRUNCATE ... CASCADE/RESTRICT
304 <p> This is like DELETE CASCADE, but truncates.
306 </li><li>%Add a separate TRUNCATE permission
307 <p> Currently only the owner can TRUNCATE a table because triggers are not
308 called, and the table is locked in exclusive mode.
310 </li><li>Allow PREPARE of cursors
311 </li><li>Allow PREPARE to automatically determine parameter types based on the SQL
313 </li><li>Allow finer control over the caching of prepared query plans
314 <p> Currently, queries prepared via the libpq API are planned on first
315 execute using the supplied parameters --- allow SQL PREPARE to do the
316 same. Also, allow control over replanning prepared queries either
317 manually or automatically when statistics for execute parameters
318 differ dramatically from those used during planning.
320 </li><li>Allow LISTEN/NOTIFY to store info in memory rather than tables?
321 <p> Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
322 such information in memory would improve performance.
324 </li><li>Add optional textual message to NOTIFY
325 <p> This would allow an informational message to be added to the notify
326 message, perhaps indicating the row modified or other custom
329 </li><li>Add a GUC variable to warn about non-standard SQL usage in queries
330 </li><li>Add MERGE command that does UPDATE/DELETE, or on failure, INSERT (rules,
332 </li><li>Add NOVICE output level for helpful messages like automatic sequence/index
334 </li><li>%Add COMMENT ON for all cluster global objects (roles, databases
336 </li><li>%Make row-wise comparisons work per SQL spec
337 </li><li>Add RESET CONNECTION command to reset all session state
338 <p> This would include resetting of all variables (RESET ALL), dropping of
339 temporary tables, removing any NOTIFYs, cursors, open transactions,
340 prepared queries, currval()s, etc. This could be used for connection
341 pooling. We could also change RESET ALL to have this functionality.
342 The difficult of this features is allowing RESET ALL to not affect
343 changes made by the interface driver for its internal use. One idea
344 is for this to be a protocol-only feature. Another approach is to
345 notify the protocol when a RESET CONNECTION command is used.
347 </li><li>Add GUC to issue notice about queries that use unjoined tables
348 </li><li>Allow EXPLAIN to identify tables that were skipped because of
350 </li><li>Allow EXPLAIN output to be more easily processed by scripts
351 </li><li>Eventually enable escape_string<u>warning and standard</u>conforming_strings
352 </li><li>Simplify dropping roles that have objects in several databases
355 <li>Allow CREATE TABLE AS to determine column lengths for complex
356 expressions like SELECT col1 || col2
357 </li><li>Use more reliable method for CREATE DATABASE to get a consistent
359 </li><li>Add ON COMMIT capability to CREATE TABLE AS ... SELECT
363 <li>Allow UPDATE to handle complex aggregates [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?update">update</a>]?
364 </li><li>Allow an alias to be provided for the target table in
366 <p> This is not SQL-spec but many DBMSs allow it.
368 </li><li>Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
373 <li>%Have ALTER TABLE RENAME rename SERIAL sequence names
374 </li><li>Add ALTER DOMAIN to modify the underlying data type
375 </li><li>%Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
376 </li><li>%Allow ALTER TABLE to change constraint deferrability and actions
377 </li><li>Add missing object types for ALTER ... SET SCHEMA
378 </li><li>Allow ALTER TABLESPACE to move to different directories
379 </li><li>Allow databases to be moved to different tablespaces
380 </li><li>Allow moving system tables to other tablespaces, where possible
381 <p> Currently non-global system tables must be in the default database
382 tablespace. Global system tables can never be moved.
384 </li><li>%Disallow dropping of an inherited constraint
385 </li><li>%Prevent child tables from altering or dropping constraints
386 like CHECK that were inherited from the parent table
390 <li>Automatically maintain clustering on a table
391 <p> This might require some background daemon to maintain clustering
392 during periods of low usage. It might also require tables to be only
393 paritally filled for easier reorganization. Another idea would
394 be to create a merged heap/index data file so an index lookup would
395 automatically access the heap data too. A third idea would be to
396 store heap rows in hashed groups, perhaps using a user-supplied
399 </li><li>%Add default clustering to system tables
400 <p> To do this, determine the ideal cluster index for each system
401 table and set the cluster setting during initdb.
406 <li>Allow COPY to report error lines and continue
407 <p> This requires the use of a savepoint before each COPY line is
408 processed, with ROLLBACK on COPY failure.
410 </li><li>%Have COPY return the number of rows loaded/unloaded?
411 </li><li>Allow COPY on a newly-created table to skip WAL logging
412 <p> On crash recovery, the table involved in the COPY would
413 be removed or have its heap and index files truncated. One
414 issue is that no other backend should be able to add to
415 the table at the same time, which is something that is
418 </li><li>Allow COPY to output from views
419 <p> Another idea would be to allow actual SELECT queries in a COPY.
422 </li><li>GRANT/REVOKE
424 <li>Allow column-level privileges
425 </li><li>%Allow GRANT/REVOKE permissions to be applied to all schema objects
427 <p> The proposed syntax is:
428 </p><p> GRANT SELECT ON ALL TABLES IN public TO phpuser;
429 GRANT SELECT ON NEW TABLES IN public TO phpuser;
432 <li>Allow GRANT/REVOKE permissions to be inherited by objects based on
438 <li>Allow UPDATE/DELETE WHERE CURRENT OF cursor
439 <p> This requires using the row ctid to map cursor rows back to the
440 original heap row. This become more complicated if WITH HOLD cursors
441 are to be supported because WITH HOLD cursors have a copy of the row
442 and no FOR UPDATE lock.
444 </li><li>Prevent DROP TABLE from dropping a row referenced by its own open
446 </li><li>%Allow pooled connections to list all open WITH HOLD cursors
447 <p> Because WITH HOLD cursors exist outside transactions, this allows
448 them to be listed so they can be closed.
453 <li>Allow INSERT/UPDATE of the system-generated oid value for a row
454 </li><li>Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
455 </li><li>Allow INSERT/UPDATE ... RETURNING new.col or old.col
456 <p> This is useful for returning the auto-generated key for an INSERT.
457 One complication is how to handle rules that run as part of
463 <li>Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
465 </li><li>Add SET PATH for schemas?
466 <p> This is basically the same as SET search_path.
469 </li><li>Server-Side Languages
471 <li>Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
472 </li><li>Allow function parameters to be passed by name,
473 get_employee_salary(emp_id => 12345, tax_year => 2001)
474 </li><li>Add Oracle-style packages
475 </li><li>Add table function support to pltcl, plpython
476 </li><li>Add capability to create and call PROCEDURES
477 </li><li>Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?"></a>]
478 </li><li>Allow function argument names to be queries from PL/PgSQL
479 </li><li>Add MOVE to PL/pgSQL
480 </li><li>Add support for polymorphic arguments and return types to
481 languages other than PL/PgSQL
482 </li><li>Add support for OUT and INOUT parameters to languages other
484 </li><li>Add single-step debugging of PL/PgSQL functions
485 </li><li>Allow PL/PgSQL to support WITH HOLD cursors
488 <h1><a name="section_9">Clients</a></h1>
491 <li>Add a libpq function to support Parse/DescribeStatement capability
492 </li><li>Prevent libpq's PQfnumber() from lowercasing the column name?
493 </li><li>Add PQescapeIdentifier() to libpq
494 </li><li>Have initdb set the input DateStyle (MDY or DMY) based on locale?
495 </li><li>Have pg_ctl look at PGHOST in case it is a socket directory?
496 </li><li>Allow pg_ctl to work properly with configuration files located outside
498 <p> pg_ctl can not read the pid file because it isn't located in the
499 config directory but in the PGDATA directory. The solution is to
500 allow pg_ctl to read and understand postgresql.conf to find the
501 data_directory value.
505 <li>Have psql show current values for a sequence
506 </li><li>Move psql backslash database information into the backend, use
507 mnemonic commands? [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?psql">psql</a>]
508 <p> This would allow non-psql clients to pull the same information out
509 of the database as psql.
511 </li><li>Fix psql's display of schema information (Neil)
512 </li><li>Allow psql \pset boolean variables to set to fixed values, rather
514 </li><li>Consistently display privilege information for all objects in psql
515 </li><li>Improve psql's handling of multi-line queries
516 <p> Currently, while \e saves a single query as one entry, interactive
517 queries are saved one line at a time. Ideally all queries
518 whould be saved like \e does.
520 </li><li>Allow multi-line column values to align in the proper columns
521 <p> If the second output column value is 'a\nb', the 'b' should appear
522 in the second display column, rather than the first column as it
525 </li><li>Display IN, INOUT, and OUT parameters in \df+
526 <p> It probably requires psql to output newlines in the proper
527 column, which is already on the TODO list.
532 <li>%Have pg_dump use multi-statement transactions for INSERT dumps
533 </li><li>%Allow pg_dump to use multiple -t and -n switches [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pg_dump">pg_dump</a>]
534 </li><li>%Add dumping of comments on composite type columns
535 </li><li>%Add dumping of comments on index columns
536 </li><li>%Replace crude DELETE FROM method of pg_dumpall --clean for
537 cleaning of roles with separate DROP commands
538 </li><li>Stop dumping CASCADE on DROP TYPE commands in clean mode
539 </li><li>%Add full object name to the tag field. eg. for operators we need
540 '=(integer, integer)', instead of just '='.
541 </li><li>Add pg_dumpall custom format dumps?
542 </li><li>%Add CSV output format
543 </li><li>Update pg_dump and psql to use the new COPY libpq API (Christopher)
544 </li><li>Remove unnecessary function pointer abstractions in pg_dump source
550 <p> Document differences between ecpg and the SQL standard and
551 information about the Informix-compatibility module.
553 </li><li>Solve cardinality > 1 for input descriptors / variables?
554 </li><li>Add a semantic check level, e.g. check if a table really exists
555 </li><li>fix handling of DB attributes that are arrays
556 </li><li>Use backend PREPARE/EXECUTE facility for ecpg where possible
557 </li><li>Implement SQLDA
558 </li><li>Fix nested C comments
559 </li><li>%sqlwarn[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?6">6</a>] should be 'W' if the PRECISION or SCALE value specified
560 </li><li>Make SET CONNECTION thread-aware, non-standard?
561 </li><li>Allow multidimensional arrays
562 </li><li>Add internationalized message strings
565 <h1><a name="section_10">Referential Integrity</a></h1>
568 <li>Add MATCH PARTIAL referential integrity
569 </li><li>Add deferred trigger queue file
570 <p> Right now all deferred trigger information is stored in backend
571 memory. This could exhaust memory for very large trigger queues.
572 This item involves dumping large queues into files.
574 </li><li>Change foreign key constraint for array -> element to mean element
576 </li><li>Allow DEFERRABLE UNIQUE constraints?
577 </li><li>Allow triggers to be disabled in only the current session.
578 <p> This is currently possible by starting a multi-statement transaction,
579 modifying the system tables, performing the desired SQL, restoring the
580 system tables, and committing the transaction. ALTER TABLE ...
581 TRIGGER requires a table lock so it is not ideal for this usage.
583 </li><li>With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
584 <p> If the dump is known to be valid, allow foreign keys to be added
585 without revalidating the data.
587 </li><li>Allow statement-level triggers to access modified rows
588 </li><li>Support triggers on columns (Greg Sabino Mullane)
589 </li><li>Enforce referential integrity for system tables
590 </li><li>Allow AFTER triggers on system tables
591 <p> System tables are modified in many places in the backend without going
592 through the executor and therefore not causing triggers to fire. To
593 complete this item, the functions that modify system tables will have
597 <h1><a name="section_11">Dependency Checking</a></h1>
600 <li>Flush cached query plans when the dependent objects change
601 </li><li>Track dependencies in function bodies and recompile/invalidate
602 <p> This is particularly important for references to temporary tables
603 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
604 in PL/PgSQL is to use EXECUTE. One complexity is that a function
605 might itself drop and recreate dependent tables, causing it to
606 invalidate its own query plan.
609 <h1><a name="section_12">Exotic Features</a></h1>
612 <li>Add SQL99 WITH clause to SELECT
613 </li><li>Add SQL99 WITH RECURSIVE to SELECT
614 </li><li>Add pre-parsing phase that converts non-ISO syntax to supported
616 <p> This could allow SQL written for other databases to run without
619 </li><li>Allow plug-in modules to emulate features from other databases
620 </li><li>SQL*Net listener that makes PostgreSQL appear as an Oracle database
622 </li><li>Allow queries across databases or servers with transaction
624 <p> This can be done using dblink and two-phase commit.
626 </li><li>Add the features of packages
628 <li>Make private objects accessable only to objects in the same schema
629 </li><li>Allow current_schema.objname to access current schema objects
630 </li><li>Add session variables
631 </li><li>Allow nested schemas
634 <h1><a name="section_13">Indexes</a></h1>
637 <li>Allow inherited tables to inherit index, UNIQUE constraint, and primary
639 </li><li>UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
640 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
642 <p> The main difficulty with this item is the problem of creating an index
643 that can span more than one table.
645 </li><li>Allow SELECT ... FOR UPDATE on inherited tables
646 </li><li>Add UNIQUE capability to non-btree indexes
647 </li><li>Prevent index uniqueness checks when UPDATE does not modify the column
648 <p> Uniqueness (index) checks are done when updating a column even if the
649 column is not modified by the UPDATE.
651 </li><li>Allow the creation of on-disk bitmap indexes which can be quickly
652 combined with other bitmap indexes
653 <p> Such indexes could be more compact if there are only a few distinct values.
654 Such indexes can also be compressed. Keeping such indexes updated can be
657 </li><li>Allow use of indexes to search for NULLs
658 <p> One solution is to create a partial index on an IS NULL expression.
660 </li><li>Allow accurate statistics to be collected on indexes with more than
661 one column or expression indexes, perhaps using per-index statistics
662 </li><li>Add fillfactor to control reserved free space during index creation
663 </li><li>Allow the creation of indexes with mixed ascending/descending specifiers
664 </li><li>Allow constraint_exclusion to work for UNIONs like it does for
665 inheritance, allow it to work for UPDATE and DELETE queries, and allow
666 it to be used for all queries with little performance impact
667 </li><li>Allow CREATE INDEX to take an additional parameter for use with
671 <li>Add more GIST index support for geometric data types
672 </li><li>Allow GIST indexes to create certain complex index types, like
673 digital trees (see Aoki)
677 <li>Pack hash index buckets onto disk pages more efficiently
678 <p> Currently only one hash bucket can be stored on a page. Ideally
679 several hash buckets could be stored on a single page and greater
680 granularity used for the hash algorithm.
682 </li><li>Consider sorting hash buckets so entries can be found using a
683 binary search, rather than a linear scan
684 </li><li>In hash indexes, consider storing the hash value with or instead
686 </li><li>Add WAL logging for crash recovery
687 </li><li>Allow multi-column hash indexes
690 <h1><a name="section_14">Fsync</a></h1>
693 <li>Improve commit_delay handling to reduce fsync()
694 </li><li>Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
695 <p> Ideally this requires a separate test program that can be run
696 at initdb time or optionally later.
698 </li><li>%Add an option to sync() before fsync()'ing checkpoint files
699 </li><li>Add program to test if fsync has a delay compared to non-fsync
701 <h1><a name="section_15">Cache Usage</a></h1>
704 <li>Allow free-behind capability for large sequential scans, perhaps using
706 <p> Posix_fadvise() can control both sequential/random file caching and
707 free-behind behavior, but it is unclear how the setting affects other
708 backends that also have the file open, and the feature is not supported
709 on all operating systems.
711 </li><li>Speed up COUNT(*)
712 <p> We could use a fixed row count and a +/- count to follow MVCC
713 visibility rules, or a single cached value could be used and
714 invalidated if anyone modifies the table. Another idea is to
715 get a count directly from a unique index, but for this to be
716 faster than a sequential scan it must avoid access to the heap
717 to obtain tuple visibility information.
719 </li><li>Allow data to be pulled directly from indexes
720 <p> Currently indexes do not have enough tuple visibility information
721 to allow data to be pulled from the index without also accessing
722 the heap. One way to allow this is to set a bit to index tuples
723 to indicate if a tuple is currently visible to all transactions
724 when the first valid heap lookup happens. This bit would have to
725 be cleared when a heap tuple is expired.
727 </li><li>Consider automatic caching of queries at various levels:
729 <li>Parsed query tree
730 </li><li>Query execute plan
731 </li><li>Query results
733 </li><li>Allow sequential scans to take advantage of other concurrent
734 sequentiqal scans, also called "Synchronised Scanning"
735 <p> One possible implementation is to start sequential scans from the lowest
736 numbered buffer in the shared cache, and when reaching the end wrap
737 around to the beginning, rather than always starting sequential scans
738 at the start of the table.
741 <h1><a name="section_16">Vacuum</a></h1>
744 <li>Improve speed with indexes
745 <p> For large table adjustements during VACUUM FULL, it is faster to
746 reindex rather than update the index.
748 </li><li>Reduce lock time during VACUUM FULL by moving tuples with read lock,
749 then write lock and truncate table
750 <p> Moved tuples are invisible to other backends so they don't require a
751 write lock. However, the read lock promotion to write lock could lead
752 to deadlock situations.
754 </li><li>Auto-fill the free space map by scanning the buffer cache or by
755 checking pages written by the background writer
756 </li><li>Create a bitmap of pages that need vacuuming
757 <p> Instead of sequentially scanning the entire table, have the background
758 writer or some other process record pages that have expired rows, then
759 VACUUM can look at just those pages rather than the entire table. In
760 the event of a system crash, the bitmap would probably be invalidated.
761 One complexity is that index entries still have to be vacuumed, and
762 doing this without an index scan (by using the heap values to find the
763 index entry) might be slow and unreliable, especially for user-defined
766 </li><li>%Add system view to show free space map contents
769 <li>Use free-space map information to guide refilling
770 </li><li>%Issue log message to suggest VACUUM FULL if a table is nearly
772 </li><li>Improve xid wraparound detection by recording per-table rather
776 <h1><a name="section_17">Locking</a></h1>
779 <li>Add code to detect an SMP machine and handle spinlocks accordingly
780 from distributted.net, <a href="http://www1.distributed.net/source">http://www1.distributed.net/source</a>,
781 in client/common/cpucheck.cpp
782 <p> On SMP machines, it is possible that locks might be released shortly,
783 while on non-SMP machines, the backend should sleep so the process
784 holding the lock can complete and release it.
786 </li><li>Fix priority ordering of read and write light-weight locks (Neil)
788 <h1><a name="section_18">Startup Time Improvements</a></h1>
791 <li>Experiment with multi-threaded backend [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?thread">thread</a>]
792 <p> This would prevent the overhead associated with process creation. Most
793 operating systems have trivial process creation time compared to
794 database startup overhead, but a few operating systems (WIn32,
795 Solaris) might benefit from threading. Also explore the idea of
796 a single session using multiple threads to execute a query faster.
798 </li><li>Add connection pooling
799 <p> It is unclear if this should be done inside the backend code or done
800 by something external like pgpool. The passing of file descriptors to
801 existing backends is one of the difficulties with a backend approach.
804 <h1><a name="section_19">Write-Ahead Log</a></h1>
807 <li>Eliminate need to write full pages to WAL before page modification [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?wal">wal</a>]
808 <p> Currently, to protect against partial disk page writes, we write
809 full page images to WAL before they are modified so we can correct any
810 partial page writes during recovery. These pages can also be
811 eliminated from point-in-time archive files.
814 <li>When off, write CRC to WAL and check file system blocks
816 <p> If CRC check fails during recovery, remember the page in case
817 a later CRC for that page properly matches.
819 </li><li>Write full pages during file system write and not when
820 the page is modified in the buffer cache
821 <p> This allows most full page writes to happen in the background
822 writer. It might cause problems for applying WAL on recovery
823 into a partially-written page, but later the full page will be
827 </li><li>Allow WAL traffic to be streamed to another server for stand-by
829 </li><li>Reduce WAL traffic so only modified values are written rather than
831 </li><li>Allow the pg_xlog directory location to be specified during initdb
832 with a symlink back to the /data location
833 </li><li>Allow WAL information to recover corrupted pg_controldata
834 </li><li>Find a way to reduce rotational delay when repeatedly writing
836 <p> Currently fsync of WAL requires the disk platter to perform a full
837 rotation to fsync again. One idea is to write the WAL to different
838 offsets that might reduce the rotational delay.
840 </li><li>Allow buffered WAL writes and fsync
841 <p> Instead of guaranteeing recovery of all committed transactions, this
842 would provide improved performance by delaying WAL writes and fsync
843 so an abrupt operating system restart might lose a few seconds of
844 committed transactions but still be consistent. We could perhaps
845 remove the 'fsync' parameter (which results in an an inconsistent
846 database) in favor of this capability.
849 <h1><a name="section_20">Optimizer / Executor</a></h1>
852 <li>Add missing optimizer selectivities for date, r-tree, etc
853 </li><li>Allow ORDER BY ... LIMIT # to select high/low value without sort or
854 index using a sequential scan for highest/lowest values
855 <p> Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
856 all values to return the high/low value. Instead The idea is to do a
857 sequential scan to find the high/low value, thus avoiding the sort.
858 MIN/MAX already does this, but not for LIMIT > 1.
860 </li><li>Precompile SQL functions to avoid overhead
861 </li><li>Create utility to compute accurate random_page_cost value
862 </li><li>Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
863 </li><li>Have EXPLAIN ANALYZE highlight poor optimizer estimates
864 </li><li>Consider using hash buckets to do DISTINCT, rather than sorting
865 <p> This would be beneficial when there are few distinct values. This is
866 already used by GROUP BY.
868 </li><li>Log queries where the optimizer row estimates were dramatically
869 different from the number of rows actually found?
871 <h1><a name="section_21">Miscellaneous Performance</a></h1>
874 <li>Do async I/O for faster random read-ahead of data
875 <p> Async I/O allows multiple I/O requests to be sent to the disk with
876 results coming back asynchronously.
878 </li><li>Use mmap() rather than SYSV shared memory or to write WAL files?
879 <p> This would remove the requirement for SYSV SHM but would introduce
880 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
881 to prevent I/O overhead.
883 </li><li>Consider mmap()'ing files into a backend?
884 <p> Doing I/O to large tables would consume a lot of address space or
885 require frequent mapping/unmapping. Extending the file also causes
886 mapping problems that might require mapping only individual pages,
887 leading to thousands of mappings. Another problem is that there is no
888 way to _prevent_ I/O to disk from the dirty shared buffers so changes
889 could hit disk before WAL is written.
891 </li><li>Add a script to ask system configuration questions and tune postgresql.conf
892 </li><li>Merge xmin/xmax/cmin/cmax back into three header fields
893 <p> Before subtransactions, there used to be only three fields needed to
894 store these four values. This was possible because only the current
895 transaction looks at the cmin/cmax values. If the current transaction
896 created and expired the row the fields stored where xmin (same as
897 xmax), cmin, cmax, and if the transaction was expiring a row from a
898 another transaction, the fields stored were xmin (cmin was not
899 needed), xmax, and cmax. Such a system worked because a transaction
900 could only see rows from another completed transaction. However,
901 subtransactions can see rows from outer transactions, and once the
902 subtransaction completes, the outer transaction continues, requiring
903 the storage of all four fields. With subtransactions, an outer
904 transaction can create a row, a subtransaction expire it, and when the
905 subtransaction completes, the outer transaction still has to have
906 proper visibility of the row's cmin, for example, for cursors.
908 <p> One possible solution is to create a phantom cid which represents a
909 cmin/cmax pair and is stored in local memory. Another idea is to
910 store both cmin and cmax only in local memory.
912 </li><li>Research storing disk pages with no alignment/padding
914 <h1><a name="section_22">Source Code</a></h1>
917 <li>Add use of 'const' for variables in source tree
918 </li><li>Rename some /contrib modules from pg* to pg_*
919 </li><li>Move some things from /contrib into main tree
920 </li><li>Move some /contrib modules out to their own project sites
921 </li><li>%Remove warnings created by -Wcast-align
922 </li><li>Move platform-specific ps status display info from ps_status.c to ports
923 </li><li>Add optional CRC checksum to heap and index pages
924 </li><li>Improve documentation to build only interfaces (Marc)
925 </li><li>Remove or relicense modules that are not under the BSD license, if possible
926 </li><li>%Remove memory/file descriptor freeing before ereport(ERROR)
927 </li><li>Acquire lock on a relation before building a relcache entry for it
928 </li><li>%Promote debug_query_string into a server-side function current_query()
929 </li><li>%Allow the identifier length to be increased via a configure option
930 </li><li>Remove Win32 rename/unlink looping if unnecessary
931 </li><li>Allow cross-compiling by generating the zic database on the target system
932 </li><li>Improve NLS maintenace of libpgport messages linked onto applications
933 </li><li>Allow ecpg to work with MSVC and BCC
934 </li><li>Add xpath_array() to /contrib/xml2 to return results as an array
935 </li><li>Allow building in directories containing spaces
936 <p> This is probably not possible because 'gmake' and other compiler tools
937 do not fully support quoting of paths with spaces.
939 </li><li>Allow installing to directories containing spaces
940 <p> This is possible if proper quoting is added to the makefiles for the
941 install targets. Because PostgreSQL supports relocatable installs, it
942 is already possible to install into a directory that doesn't contain
943 spaces and then copy the install to a directory with spaces.
945 </li><li>Fix sgmltools so PDFs can be generated with bookmarks
946 </li><li>%Clean up compiler warnings (especially with gcc version 4)
947 </li><li>Add function to return the thread safety status of libpq and ecpg
948 </li><li>Use UTF8 encoding for NLS messages so all server encodings can
950 </li><li>Update Bonjour to work with newer cross-platform SDK
951 </li><li>Remove BeOS and QNX-specific code
954 <li>Remove configure.in check for link failure when cause is found
955 </li><li>Remove readdir() errno patch when runtime/mingwex/dirent.c rev
957 </li><li>Remove psql newline patch when we find out why mingw outputs an
959 </li><li>Allow psql to use readline once non-US code pages work with
961 </li><li>Re-enable timezone output on log_line_prefix '%t' when a
962 shorter timezone string is available
963 </li><li>Fix problem with shared memory on the Win32 Terminal Server
964 </li><li>Improve signal handling,
965 <a href="http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php">http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php</a>
967 </li><li>Wire Protocol Changes
969 <li>Allow dynamic character set handling
970 </li><li>Add decoded type, length, precision
971 </li><li>Use compression?
972 </li><li>Update clients to use data types, typmod, <a href="http://schema.table.column">schema.table.column</a>/ names
973 of result sets using new query protocol
978 <h2><a name="section_22_1">Developers who have claimed items are:</a></h2>
980 <li>Alvaro is Alvaro Herrera <<a href="mailto:alvherre@dcc.uchile.cl">alvherre@dcc.uchile.cl</a>>
981 </li><li>Andrew is Andrew Dunstan <<a href="mailto:andrew@dunslane.net">andrew@dunslane.net</a>>
982 </li><li>Bruce is Bruce Momjian <<a href="mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</a>> of Software Research Assoc.
983 </li><li>Christopher is Christopher Kings-Lynne <<a href="mailto:chriskl@familyhealth.com.au">chriskl@familyhealth.com.au</a>> of
984 Family Health Network
985 </li><li>Claudio is Claudio Natoli <<a href="mailto:claudio.natoli@memetrics.com">claudio.natoli@memetrics.com</a>>
986 </li><li>D'Arcy is D'Arcy J.M. Cain <<a href="mailto:darcy@druid.net">darcy@druid.net</a>> of The Cain Gang Ltd.
987 </li><li>Fabien is Fabien Coelho <<a href="mailto:coelho@cri.ensmp.fr">coelho@cri.ensmp.fr</a>>
988 </li><li>Gavin is Gavin Sherry <<a href="mailto:swm@linuxworld.com.au">swm@linuxworld.com.au</a>> of Alcove Systems Engineering
989 </li><li>Greg is Greg Sabino Mullane <<a href="mailto:greg@turnstep.com">greg@turnstep.com</a>>
990 </li><li>Hiroshi is Hiroshi Inoue <<a href="mailto:Inoue@tpf.co.jp">Inoue@tpf.co.jp</a>>
991 </li><li>Jan is Jan Wieck <<a href="mailto:JanWieck@Yahoo.com">JanWieck@Yahoo.com</a>> of Afilias, Inc.
992 </li><li>Joe is Joe Conway <<a href="mailto:mail@joeconway.com">mail@joeconway.com</a>>
993 </li><li>Karel is Karel Zak <<a href="mailto:zakkr@zf.jcu.cz">zakkr@zf.jcu.cz</a>>
994 </li><li>Magnus is Magnus Hagander <<a href="mailto:mha@sollentuna.net">mha@sollentuna.net</a>>
995 </li><li>Marc is Marc Fournier <<a href="mailto:scrappy@hub.org">scrappy@hub.org</a>> of PostgreSQL, Inc.
996 </li><li>Matthew T. O'Connor <<a href="mailto:matthew@zeut.net">matthew@zeut.net</a>>
997 </li><li>Michael is Michael Meskes <<a href="mailto:meskes@postgresql.org">meskes@postgresql.org</a>> of Credativ
998 </li><li>Neil is Neil Conway <<a href="mailto:neilc@samurai.com">neilc@samurai.com</a>>
999 </li><li>Oleg is Oleg Bartunov <<a href="mailto:oleg@sai.msu.su">oleg@sai.msu.su</a>>
1000 </li><li>Peter is Peter Eisentraut <<a href="mailto:peter_e@gmx.net">peter_e@gmx.net</a>>
1001 </li><li>Philip is Philip Warner <<a href="mailto:pjw@rhyme.com.au">pjw@rhyme.com.au</a>> of Albatross Consulting Pty. Ltd.
1002 </li><li>Rod is Rod Taylor <<a href="mailto:pg@rbt.ca">pg@rbt.ca</a>>
1003 </li><li>Simon is Simon Riggs <<a href="mailto:simon@2ndquadrant.com">simon@2ndquadrant.com</a>>
1004 </li><li>Stephan is Stephan Szabo <<a href="mailto:sszabo@megazone23.bigpanda.com">sszabo@megazone23.bigpanda.com</a>>
1005 </li><li>Tatsuo is Tatsuo Ishii <<a href="mailto:t-ishii@sra.co.jp">t-ishii@sra.co.jp</a>> of Software Research Assoc.
1006 </li><li>Tom is Tom Lane <<a href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>> of Red Hat
1008 </li></ul></li></ul>