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: Mon Mar 6 10:13:39 EST 2006
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 8.2 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
29 </li><li>-*%Allow pooled connections to list all prepared statements*
30 <p> This would allow an application inheriting a pooled connection to know
31 the statements prepared in the current session.
33 </li><li>Allow major upgrades without dump/reload, perhaps using pg_upgrade
34 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pg_upgrade">pg_upgrade</a>]
35 </li><li>Check for unreferenced table files created by transactions that were
36 in-progress when the server terminated abruptly
37 </li><li>Allow administrators to safely terminate individual sessions either
38 via an SQL function or SIGTERM
39 <p> Lock table corruption following SIGTERM of an individual backend
40 has been reported in 8.0. A possible cause was fixed in 8.1, but
41 it is unknown whether other problems exist. This item mostly
42 requires additional testing rather than of writing any new code.
44 </li><li>%Set proper permissions on non-system schemas during db creation
45 <p> Currently all schemas are owned by the super-user because they are
46 copied from the template1 database.
48 </li><li>Support table partitioning that allows a single table to be stored
49 in subtables that are partitioned based on the primary key or a WHERE
51 </li><li>Add function to report the time of the most recent server reload
52 </li><li>Allow statistics collector information to be pulled from the collector
53 process directly, rather than requiring the collector to write a
54 filesystem file twice a second?
55 </li><li>Improve replication solutions
58 <p> You can use any of the master/slave replication servers to use a
59 standby server for data warehousing. To allow read/write queries to
60 multiple servers, you need multi-master replication like pgcluster.
62 </li><li>Allow replication over unreliable or non-persistent links
64 </li><li>Configuration files
66 <li>-<em>Add "include file" functionality in postgresql.conf</em>
67 </li><li>%Allow commenting of variables in postgresql.conf to restore them
69 <p> Currently, if a variable is commented out, it keeps the
70 previous uncommented value until a server restarted.
72 </li><li>%Allow pg_hba.conf settings to be controlled via SQL
73 <p> This would add a function to load the SQL table from
74 pg_hba.conf, and one to writes its contents to the flat file.
75 The table should have a line number that is a float so rows
76 can be inserted between existing rows, e.g. row 2.5 goes
77 between row 2 and row 3.
79 </li><li>Allow pg_hba.conf to specify host names along with IP addresses
80 <p> Host name lookup could occur when the postmaster reads the
81 pg_hba.conf file, or when the backend starts. Another
82 solution would be to reverse lookup the connection IP and
83 check that hostname against the host names in pg_hba.conf.
84 We could also then check that the host name maps to the IP
87 </li><li>%Allow postgresql.conf file values to be changed via an SQL
88 API, perhaps using SET GLOBAL
89 </li><li>Allow the server to be stopped/restarted via an SQL API
90 </li><li>-<em>Issue a warning if a change-on-restart-only postgresql.conf value</em>
91 is modified and the server config files are reloaded
92 </li><li>Mark change-on-restart-only values in postgresql.conf
96 <li>Allow a database in tablespace t1 with tables created in
97 tablespace t2 to be used as a template for a new database created
98 with default tablespace t2
99 <p> All objects in the default database tablespace must have default
100 tablespace specifications. This is because new databases are
101 created by copying directories. If you mix default tablespace
102 tables and tablespace-specified tables in the same directory,
103 creating a new database from such a mixed directory would create a
104 new database with tables that had incorrect explicit tablespaces.
105 To fix this would require modifying pg_class in the newly copied
106 database, which we don't currently do.
108 </li><li>Allow reporting of which objects are in which tablespaces
109 <p> This item is difficult because a tablespace can contain objects
110 from multiple databases. There is a server-side function that
111 returns the databases which use a specific tablespace, so this
112 requires a tool that will call that function and connect to each
113 database to find the objects in each database for that tablespace.
116 <li>%Add a GUC variable to control the tablespace for temporary objects
118 <p> It could start with a random tablespace from a supplied list and
119 cycle through the list.
121 </li><li>Allow WAL replay of CREATE TABLESPACE to work when the directory
122 structure on the recovery computer is different from the original
123 </li><li>Allow per-tablespace quotas
126 </li><li>Point-In-Time Recovery (PITR)
128 <li>Allow point-in-time recovery to archive partially filled
129 write-ahead logs [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
130 <p> Currently only full WAL files are archived. This means that the
131 most recent transactions aren't available for recovery in case
132 of a disk failure. This could be triggered by a user command or
135 </li><li>Automatically force archiving of partially-filled WAL files when
136 pg_stop_backup() is called or the server is stopped
137 <p> Doing this will allow administrators to know more easily when
138 the archive contains all the files needed for point-in-time
141 </li><li>%Create dump tool for write-ahead logs for use in determining
142 transaction id for point-in-time recovery
143 </li><li>Allow a warm standby system to also allow read-only statements
144 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
145 <p> This is useful for checking PITR recovery.
147 </li><li>Allow the PITR process to be debugged and data examined
150 <h1><a name="section_3">Monitoring</a></h1>
153 <li>Allow server log information to be output as INSERT statements
154 <p> This would allow server log information to be easily loaded into
155 a database for analysis.
157 </li><li>%Add ability to monitor the use of temporary sort files
158 </li><li>Allow server logs to be remotely read and removed using SQL commands
159 </li><li>Allow protocol-level BIND parameter values to be logged
161 <h1><a name="section_4">Data Types</a></h1>
164 <li>Improve the MONEY data type
165 <p> Change the MONEY data type to use DECIMAL internally, with special
166 locale-aware output formatting.
168 </li><li>Change NUMERIC to enforce the maximum precision
169 </li><li>Add NUMERIC division operator that doesn't round?
170 <p> Currently NUMERIC _rounds_ the result to the specified precision.
171 This means division can return a result that multiplied by the
172 divisor is greater than the dividend, e.g. this returns a value > 10:
175 <p> SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
177 <p> The positive modulus result returned by NUMERICs might be considered<br/>
178 inaccurate, in one sense.
181 <li>%Disallow changing default expression of a SERIAL column
182 </li><li>%Disallow ALTER SEQUENCE changes for SERIAL sequences because pg_dump
183 does not dump the changes
184 </li><li>Fix data types where equality comparison isn't intuitive, e.g. box
185 </li><li>-<em>Zero umasked bits in conversion from INET cast to CIDR</em>
186 </li><li>-<em>Prevent INET cast to CIDR from dropping netmask, SELECT '<a href="telnet://1.1.1.1">1.1.1.1</a>'::inet::cidr</em>
187 </li><li>-<em>Allow INET + INT8 to increment the host part of the address or</em>
188 throw an error on overflow
189 </li><li>-<em>Add 'tid != tid ' operator for use in corruption recovery</em>
190 </li><li>Allow user-defined types to specify a type modifier at table creation
192 </li><li>Dates and Times
194 <li>Allow infinite dates and intervals just like infinite timestamps
195 </li><li>Merge hardwired timezone names with the TZ database; allow either
196 kind everywhere a TZ name is currently taken
197 </li><li>Allow customization of the known set of TZ names (generalize the
198 present australian_timezones hack)
199 </li><li>Allow TIMESTAMP WITH TIME ZONE to store the original timezone
200 information, either zone name or offset from UTC [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?timezone">timezone</a>]
201 <p> If the TIMESTAMP value is stored with a time zone name, interval
202 computations should adjust based on the time zone rules.
204 </li><li>Fix SELECT '0.01 years'::interval, '0.01 months'::interval
205 </li><li>Fix SELECT INTERVAL '1' MONTH
206 </li><li>Add a GUC variable to allow output of interval values in ISO8601
208 </li><li>Improve timestamptz subtraction to be DST-aware
209 <p> Currently, subtracting one date from another that crosses a
210 daylight savings time adjustment can return '1 day 1 hour', but
211 adding that back to the first date returns a time one hour in
212 the future. This is caused by the adjustment of '25 hours' to
213 '1 day 1 hour', and '1 day' is the same time the next day, even
214 if daylight savings adjustments are involved.
216 </li><li>Fix interval display to support values exceeding 2^31 hours
217 </li><li>Add overflow checking to timestamp and interval arithmetic
218 </li><li>Add ISO INTERVAL handling
220 <li>Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
222 </li><li>Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
223 </li><li>For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
224 '1', treat as ISO if there is a range specification clause,
225 and as PG if there no clause is present, e.g. interpret
226 <p> '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
227 interpret '1:30' as '1 hour, 30 minutes'
228 <li>Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
229 INTERVAL MONTH), and this should return '12 months'
230 </li><li>Round or truncate values to the requested precision, e.g.
231 INTERVAL '11 months' AS YEAR should return one or zero
232 </li><li>Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
237 <li>-<em>Allow NULLs in arrays</em>
238 </li><li>Delay resolution of array expression's data type so assignment
239 coercion can be performed on empty array expressions
243 <li>Improve vacuum of large objects, like /contrib/vacuumlo?
244 </li><li>Add security checking for large objects
245 </li><li>Auto-delete large objects when referencing row is deleted
246 <p> /contrib/lo offers this functionality.
248 </li><li>Allow read/write into TOAST values like large objects
249 <p> This requires the TOAST column to be stored EXTERNAL.
253 <h1><a name="section_5">Functions</a></h1>
256 <li>Allow INET subnet tests using non-constants to be indexed
257 </li><li>Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
259 <p> Current CURRENT_TIMESTAMP returns the start time of the current
260 transaction, and gettimeofday() returns the wallclock time. This will
261 make time reporting more consistent and will allow reporting of
262 the statement start time.
264 </li><li>%Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
265 pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
266 </li><li>-<em>Allow to_char() to print localized month names</em>
267 </li><li>Allow to_date() and to_timestamp() accept localized month names
268 </li><li>Add missing parameter handling in to_char()
270 <p> <a href="http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php">http://archives.postgresql.org/pgsql-hackers/2005-12/msg00948.php</a>
273 <li>Allow functions to have a schema search path specified at creation time
274 </li><li>Allow substring/replace() to get/set bit values
275 </li><li>Allow to_char() on interval values to accumulate the highest unit
277 <p> Some special format flag would be required to request such
278 accumulation. Such functionality could also be added to EXTRACT.
279 Prevent accumulation that crosses the month/day boundary because of
280 the uneven number of days in a month.
283 <li>to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
284 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
285 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
286 </li><li>to_char(INTERVAL '3 years 5 months','MM') => 41
288 </li><li>-<em>Add sleep() function, remove from regress.c</em>
289 </li><li>Allow user-defined functions retuning a domain value to enforce domain
291 </li><li>Add SPI_gettypmod() to return the typemod for a TupleDesc
293 <h1><a name="section_6">Multi-Language Support</a></h1>
296 <li>Add NCHAR (as distinguished from ordinary varchar),
297 </li><li>Allow locale to be set at database creation
298 <p> Currently locale can only be set during initdb. No global tables have
299 locale-aware columns. However, the database template used during
300 database creation might have locale-aware indexes. The indexes would
301 need to be reindexed to match the new locale.
303 </li><li>Allow encoding on a per-column basis
304 <p> Right now only one encoding is allowed per database.
306 </li><li>Support multiple simultaneous character sets, per SQL92
307 </li><li>Improve UTF8 combined character handling?
308 </li><li>Add octet_length_server() and octet_length_client()
309 </li><li>Make octet_length_client() the same as octet_length()?
310 </li><li>Fix problems with wrong runtime encoding conversion for NLS message files
312 <h1><a name="section_7">Views / Rules</a></h1>
315 <li>Automatically create rules on views so they are updateable, per SQL99
316 <p> We can only auto-create rules for simple views. For more complex
317 cases users will still have to write rules manually.
319 </li><li>Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
320 </li><li>Allow NOTIFY in rules involving conditionals
321 </li><li>Allow VIEW/RULE recompilation when the underlying tables change
322 <p> Another issue is whether underlying table changes should be reflected
323 in the view, e.g. should SELECT * show additional columns if they
324 are added after the view is created.
327 <h1><a name="section_8">SQL Commands</a></h1>
330 <li>Change LIMIT/OFFSET and FETCH/MOVE to use int8
331 </li><li>Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
332 </li><li>Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
333 </li><li>%Allow SET CONSTRAINTS to be qualified by schema/table name
334 </li><li>-<em>Allow TRUNCATE ... CASCADE/RESTRICT</em>
335 <p> This is like DELETE CASCADE, but truncates.
337 </li><li>%Add a separate TRUNCATE permission
338 <p> Currently only the owner can TRUNCATE a table because triggers are not
339 called, and the table is locked in exclusive mode.
341 </li><li>Allow PREPARE of cursors
342 </li><li>-<em>Allow PREPARE to automatically determine parameter types based on the SQL</em>
344 </li><li>Allow finer control over the caching of prepared query plans
345 <p> Currently, queries prepared via the libpq API are planned on first
346 execute using the supplied parameters --- allow SQL PREPARE to do the
347 same. Also, allow control over replanning prepared queries either
348 manually or automatically when statistics for execute parameters
349 differ dramatically from those used during planning.
351 </li><li>Allow LISTEN/NOTIFY to store info in memory rather than tables?
352 <p> Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
353 such information in memory would improve performance.
355 </li><li>Add optional textual message to NOTIFY
356 <p> This would allow an informational message to be added to the notify
357 message, perhaps indicating the row modified or other custom
360 </li><li>Add a GUC variable to warn about non-standard SQL usage in queries
361 </li><li>Add SQL-standard MERGE command, typically used to merge two tables
362 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?merge">merge</a>]
363 <p> This is similar to UPDATE, then for unmatched rows, INSERT.
364 Whether concurrent access allows modifications which could cause
365 row loss is implementation independent.
367 </li><li>Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
368 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?merge">merge</a>]
369 <p> To implement this cleanly requires that the table have a unique index
370 so duplicate checking can be easily performed. It is possible to
371 do it without a unique index if we require the user to LOCK the table
374 </li><li>Add NOVICE output level for helpful messages like automatic sequence/index
376 </li><li>-<em>Add COMMENT ON for all cluster global objects (roles, databases</em>
378 </li><li>-<em>Make row-wise comparisons work per SQL spec</em>
379 <p> Right now, '(a, b) < (1, 2)' is processed as 'a < 1 and b < 2', but
380 the SQL standard requires it to be processed as a column-by-column
381 comparison, so the proper comparison is '(a < 1) OR (a = 1 AND b < 2)'.
383 </li><li>Add RESET CONNECTION command to reset all session state
384 <p> This would include resetting of all variables (RESET ALL), dropping of
385 temporary tables, removing any NOTIFYs, cursors, open transactions,
386 prepared queries, currval()s, etc. This could be used for connection
387 pooling. We could also change RESET ALL to have this functionality.
388 The difficult of this features is allowing RESET ALL to not affect
389 changes made by the interface driver for its internal use. One idea
390 is for this to be a protocol-only feature. Another approach is to
391 notify the protocol when a RESET CONNECTION command is used.
393 </li><li>Add GUC to issue notice about statements that use unjoined tables
394 </li><li>Allow EXPLAIN to identify tables that were skipped because of
396 </li><li>Allow EXPLAIN output to be more easily processed by scripts
397 </li><li>Eventually enable escape_string<u>warning and standard</u>conforming_strings
398 </li><li>Simplify dropping roles that have objects in several databases
399 </li><li>Allow COMMENT ON to accept an expression rather than just a string
400 </li><li>Allow the count returned by SELECT, etc to be to represent as an int64
401 to allow a higher range of values
402 </li><li>Make CLUSTER preserve recently-dead tuples per MVCC requirements
403 </li><li>Add SQL99 WITH clause to SELECT
404 </li><li>Add SQL99 WITH RECURSIVE to SELECT
407 <li>Allow CREATE TABLE AS to determine column lengths for complex
408 expressions like SELECT col1 || col2
409 </li><li>Use more reliable method for CREATE DATABASE to get a consistent
411 </li><li>-<em>Add ON COMMIT capability to CREATE TABLE AS ... SELECT</em>
415 <li>Allow UPDATE to handle complex aggregates [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?update">update</a>]?
416 </li><li>-<em>Allow an alias to be provided for the target table in</em>
418 </li><li>Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
423 <li>%Have ALTER TABLE RENAME rename SERIAL sequence names
424 </li><li>Add ALTER DOMAIN to modify the underlying data type
425 </li><li>%Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
426 </li><li>%Allow ALTER TABLE to change constraint deferrability and actions
427 </li><li>Add missing object types for ALTER ... SET SCHEMA
428 </li><li>Allow ALTER TABLESPACE to move to different directories
429 </li><li>Allow databases to be moved to different tablespaces
430 </li><li>Allow moving system tables to other tablespaces, where possible
431 <p> Currently non-global system tables must be in the default database
432 tablespace. Global system tables can never be moved.
434 </li><li>%Prevent child tables from altering or dropping constraints
435 like CHECK that were inherited from the parent table
436 </li><li>Have ALTER INDEX update the name of a constraint using that index
437 </li><li>Add ALTER TABLE RENAME CONSTRAINT, update index name also
441 <li>Automatically maintain clustering on a table
442 <p> This might require some background daemon to maintain clustering
443 during periods of low usage. It might also require tables to be only
444 partially filled for easier reorganization. Another idea would
445 be to create a merged heap/index data file so an index lookup would
446 automatically access the heap data too. A third idea would be to
447 store heap rows in hashed groups, perhaps using a user-supplied
450 </li><li>%Add default clustering to system tables
451 <p> To do this, determine the ideal cluster index for each system
452 table and set the cluster setting during initdb.
457 <li>Allow COPY to report error lines and continue
458 <p> This requires the use of a savepoint before each COPY line is
459 processed, with ROLLBACK on COPY failure.
461 </li><li>-<em>Have COPY return the number of rows loaded/unloaded?</em>
462 </li><li>Allow COPY on a newly-created table to skip WAL logging
463 <p> On crash recovery, the table involved in the COPY would
464 be removed or have its heap and index files truncated. One
465 issue is that no other backend should be able to add to
466 the table at the same time, which is something that is
469 </li><li>Allow COPY to output from views
470 <p> Another idea would be to allow actual SELECT statements in a COPY.
473 </li><li>GRANT/REVOKE
475 <li>Allow column-level privileges
476 </li><li>%Allow GRANT/REVOKE permissions to be applied to all schema objects
478 <p> The proposed syntax is:
479 </p><p> GRANT SELECT ON ALL TABLES IN public TO phpuser;
480 GRANT SELECT ON NEW TABLES IN public TO phpuser;
483 <li>Allow GRANT/REVOKE permissions to be inherited by objects based on
485 </li><li>Allow SERIAL sequences to inherit permissions from the base table?
490 <li>Allow UPDATE/DELETE WHERE CURRENT OF cursor
491 <p> This requires using the row ctid to map cursor rows back to the
492 original heap row. This become more complicated if WITH HOLD cursors
493 are to be supported because WITH HOLD cursors have a copy of the row
494 and no FOR UPDATE lock.
496 </li><li>Prevent DROP TABLE from dropping a row referenced by its own open
498 </li><li>-<em>Allow pooled connections to list all open WITH HOLD cursors</em>
499 <p> Because WITH HOLD cursors exist outside transactions, this allows
500 them to be listed so they can be closed.
505 <li>Allow INSERT/UPDATE of the system-generated oid value for a row
506 </li><li>Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
507 </li><li>Allow INSERT/UPDATE ... RETURNING new.col or old.col
508 <p> This is useful for returning the auto-generated key for an INSERT.
509 One complication is how to handle rules that run as part of
515 <li>Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
517 </li><li>Add SET PATH for schemas?
518 <p> This is basically the same as SET search_path.
521 </li><li>Server-Side Languages
523 <li>Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
524 </li><li>Allow function parameters to be passed by name,
525 get_employee_salary(emp_id => 12345, tax_year => 2001)
526 </li><li>Add Oracle-style packages
527 </li><li>Add table function support to pltcl, plpython
528 </li><li>Add capability to create and call PROCEDURES
529 </li><li>Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?"></a>]
530 </li><li>Allow function argument names to be statements from PL/PgSQL
531 </li><li>Add MOVE to PL/pgSQL
532 </li><li>Add support for polymorphic arguments and return types to
533 languages other than PL/PgSQL
534 </li><li>Add support for OUT and INOUT parameters to languages other
536 </li><li>Add single-step debugging of PL/PgSQL functions
537 </li><li>Allow PL/PgSQL to support WITH HOLD cursors
540 <h1><a name="section_9">Clients</a></h1>
543 <li>-<em>Have initdb set the input DateStyle (MDY or DMY) based on locale</em>
544 </li><li>Have pg_ctl look at PGHOST in case it is a socket directory?
545 </li><li>Allow pg_ctl to work properly with configuration files located outside
547 <p> pg_ctl can not read the pid file because it isn't located in the
548 config directory but in the PGDATA directory. The solution is to
549 allow pg_ctl to read and understand postgresql.conf to find the
550 data_directory value.
554 <li>Have psql show current values for a sequence
555 </li><li>Move psql backslash database information into the backend, use
556 mnemonic commands? [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?psql">psql</a>]
557 <p> This would allow non-psql clients to pull the same information out
558 of the database as psql.
560 </li><li>Fix psql's display of schema information (Neil)
561 </li><li>Allow psql \pset boolean variables to set to fixed values, rather
563 </li><li>Consistently display privilege information for all objects in psql
564 </li><li>-<em>Improve psql's handling of multi-line statements</em>
565 <p> Currently, while \e saves a single statement as one entry, interactive
566 statements are saved one line at a time. Ideally all statements
567 would be saved like \e does.
569 </li><li>-<em>Allow multi-line column values to align in the proper columns</em>
570 <p> If the second output column value is 'a\nb', the 'b' should appear
571 in the second display column, rather than the first column as it
574 </li><li>Display IN, INOUT, and OUT parameters in \df+
575 <p> It probably requires psql to output newlines in the proper
576 column, which is already on the TODO list.
578 </li><li>Add auto-expanded mode so expanded output is used if the row
579 length is wider than the screen width.
580 <p> Consider using auto-expanded mode for backslash commands like \df+.
582 </li><li>Prevent tab completion of SET TRANSACTION from querying the
583 database and therefore preventing the transaction isolation
584 level from being set.
585 <p> Currently, SET <tab> causes a database lookup to check all
586 supported session variables. This query causes problems
587 because setting the transaction isolation level must be the
588 first statement of a transaction.
593 <li>%Have pg_dump use multi-statement transactions for INSERT dumps
594 </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>]
595 </li><li>%Add dumping of comments on index columns and composite type columns
596 </li><li>%Add full object name to the tag field. eg. for operators we need
597 '=(integer, integer)', instead of just '='.
598 </li><li>Add pg_dumpall custom format dumps?
599 </li><li>%Add CSV output format
600 </li><li>Update pg_dump and psql to use the new COPY libpq API (Christopher)
601 </li><li>Remove unnecessary function pointer abstractions in pg_dump source
603 </li><li>Allow selection of individual object(s) of all types, not just
605 </li><li>In a selective dump, allow dumping of an object and all its
607 </li><li>Add options like pg_restore -l and -L to pg_dump
608 </li><li>Stop dumping CASCADE on DROP TYPE commands in clean mode
609 </li><li>Allow pg_dump --clean to drop roles that own objects or have
611 </li><li>Add -f to pg_dumpall
616 <p> Document differences between ecpg and the SQL standard and
617 information about the Informix-compatibility module.
619 </li><li>Solve cardinality > 1 for input descriptors / variables?
620 </li><li>Add a semantic check level, e.g. check if a table really exists
621 </li><li>fix handling of DB attributes that are arrays
622 </li><li>Use backend PREPARE/EXECUTE facility for ecpg where possible
623 </li><li>Implement SQLDA
624 </li><li>Fix nested C comments
625 </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
626 </li><li>Make SET CONNECTION thread-aware, non-standard?
627 </li><li>Allow multidimensional arrays
628 </li><li>Add internationalized message strings
629 </li><li>Add COPY TO STDIN / STDOUT handling
635 <li>Add a function to support Parse/DescribeStatement capability
636 </li><li>Add PQescapeIdentifier()
637 </li><li>Prevent PQfnumber() from lowercasing unquoted the column name
638 <p> PQfnumber() should never have been doing lowercasing, but
639 historically it has so we need a way to prevent it
641 </li><li>Allow statement results to be automatically batched to the client
642 <p> Currently, all statement results are transferred to the libpq
643 client before libpq makes the results available to the
644 application. This feature would allow the application to make
645 use of the first result rows while the rest are transferred, or
646 held on the server waiting for them to be requested by libpq.
647 One complexity is that a statement like SELECT 1/col could error
648 out mid-way through the result set.
651 <h1><a name="section_10">Referential Integrity</a></h1>
654 <li>Add MATCH PARTIAL referential integrity
655 </li><li>Add deferred trigger queue file
656 <p> Right now all deferred trigger information is stored in backend
657 memory. This could exhaust memory for very large trigger queues.
658 This item involves dumping large queues into files.
660 </li><li>Change foreign key constraint for array -> element to mean element
662 </li><li>Allow DEFERRABLE UNIQUE constraints?
663 </li><li>Allow triggers to be disabled in only the current session.
664 <p> This is currently possible by starting a multi-statement transaction,
665 modifying the system tables, performing the desired SQL, restoring the
666 system tables, and committing the transaction. ALTER TABLE ...
667 TRIGGER requires a table lock so it is not ideal for this usage.
669 </li><li>With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
670 <p> If the dump is known to be valid, allow foreign keys to be added
671 without revalidating the data.
673 </li><li>Allow statement-level triggers to access modified rows
674 </li><li>Support triggers on columns (Greg Sabino Mullane)
675 </li><li>Enforce referential integrity for system tables
676 </li><li>Allow AFTER triggers on system tables
677 <p> System tables are modified in many places in the backend without going
678 through the executor and therefore not causing triggers to fire. To
679 complete this item, the functions that modify system tables will have
683 <h1><a name="section_11">Dependency Checking</a></h1>
686 <li>Flush cached query plans when the dependent objects change,
687 when the cardinality of parameters changes dramatically, or
688 when new ANALYZE statistics are available
689 <p> A more complex solution would be to save multiple plans for different
690 cardinality and use the appropriate plan based on the EXECUTE values.
692 </li><li>Track dependencies in function bodies and recompile/invalidate
693 <p> This is particularly important for references to temporary tables
694 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
695 in PL/PgSQL is to use EXECUTE. One complexity is that a function
696 might itself drop and recreate dependent tables, causing it to
697 invalidate its own query plan.
700 <h1><a name="section_12">Exotic Features</a></h1>
703 <li>Add pre-parsing phase that converts non-ISO syntax to supported
705 <p> This could allow SQL written for other databases to run without
708 </li><li>Allow plug-in modules to emulate features from other databases
709 </li><li>SQL*Net listener that makes PostgreSQL appear as an Oracle database
711 </li><li>Allow statements across databases or servers with transaction
713 <p> This can be done using dblink and two-phase commit.
715 </li><li>Add the features of packages
717 <li>Make private objects accessible only to objects in the same schema
718 </li><li>Allow current_schema.objname to access current schema objects
719 </li><li>Add session variables
720 </li><li>Allow nested schemas
723 <h1><a name="section_13">Indexes</a></h1>
726 <li>Allow inherited tables to inherit index, UNIQUE constraint, and primary
728 </li><li>UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
729 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
731 <p> The main difficulty with this item is the problem of creating an index
732 that can span more than one table.
734 </li><li>Allow SELECT ... FOR UPDATE on inherited tables
735 </li><li>Add UNIQUE capability to non-btree indexes
736 </li><li>Prevent index uniqueness checks when UPDATE does not modify the column
737 <p> Uniqueness (index) checks are done when updating a column even if the
738 column is not modified by the UPDATE.
740 </li><li>Allow the creation of on-disk bitmap indexes which can be quickly
741 combined with other bitmap indexes
742 <p> Such indexes could be more compact if there are only a few distinct values.
743 Such indexes can also be compressed. Keeping such indexes updated can be
746 </li><li>Allow use of indexes to search for NULLs
747 <p> One solution is to create a partial index on an IS NULL expression.
749 </li><li>Allow accurate statistics to be collected on indexes with more than
750 one column or expression indexes, perhaps using per-index statistics
751 </li><li>Add fillfactor to control reserved free space during index creation
752 </li><li>Allow the creation of indexes with mixed ascending/descending specifiers
753 </li><li>Allow constraint_exclusion to work for UNIONs like it does for
754 inheritance, allow it to work for UPDATE and DELETE statements, and allow
755 it to be used for all statements with little performance impact
756 </li><li>Allow CREATE INDEX to take an additional parameter for use with
758 </li><li>Consider compressing indexes by storing key values duplicated in
759 several rows as a single index entry
760 <p> This is difficult because it requires datatype-specific knowledge.
764 <li>Add more GIST index support for geometric data types
765 </li><li>Allow GIST indexes to create certain complex index types, like
766 digital trees (see Aoki)
770 <li>Pack hash index buckets onto disk pages more efficiently
771 <p> Currently only one hash bucket can be stored on a page. Ideally
772 several hash buckets could be stored on a single page and greater
773 granularity used for the hash algorithm.
775 </li><li>Consider sorting hash buckets so entries can be found using a
776 binary search, rather than a linear scan
777 </li><li>In hash indexes, consider storing the hash value with or instead
779 </li><li>Add WAL logging for crash recovery
780 </li><li>Allow multi-column hash indexes
783 <h1><a name="section_14">Fsync</a></h1>
786 <li>Improve commit_delay handling to reduce fsync()
787 </li><li>Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
788 <p> Ideally this requires a separate test program that can be run
789 at initdb time or optionally later. Consider O_SYNC when
792 </li><li>%Add an option to sync() before fsync()'ing checkpoint files
793 </li><li>Add program to test if fsync has a delay compared to non-fsync
795 <h1><a name="section_15">Cache Usage</a></h1>
798 <li>Allow free-behind capability for large sequential scans, perhaps using
800 <p> Posix_fadvise() can control both sequential/random file caching and
801 free-behind behavior, but it is unclear how the setting affects other
802 backends that also have the file open, and the feature is not supported
803 on all operating systems.
805 </li><li>Speed up COUNT(*)
806 <p> We could use a fixed row count and a +/- count to follow MVCC
807 visibility rules, or a single cached value could be used and
808 invalidated if anyone modifies the table. Another idea is to
809 get a count directly from a unique index, but for this to be
810 faster than a sequential scan it must avoid access to the heap
811 to obtain tuple visibility information.
813 </li><li>Add estimated_count(*) to return an estimate of COUNT(*)
814 <p> This would use the planner ANALYZE statistics to return an estimated
817 </li><li>Allow data to be pulled directly from indexes
818 <p> Currently indexes do not have enough tuple visibility information
819 to allow data to be pulled from the index without also accessing
820 the heap. One way to allow this is to set a bit on index tuples
821 to indicate if a tuple is currently visible to all transactions
822 when the first valid heap lookup happens. This bit would have to
823 be cleared when a heap tuple is expired.
825 <p> Another idea is to maintain a bitmap of heap pages where all rows
826 are visible to all backends, and allow index lookups to reference
827 that bitmap to avoid heap lookups, perhaps the same bitmap we might
828 add someday to determine which heap pages need vacuuming. Frequently
829 accessed bitmaps would have to be stored in shared memory. One 8k
830 page of bitmaps could track 512MB of heap pages.
832 </li><li>Consider automatic caching of statements at various levels:
834 <li>Parsed query tree
835 </li><li>Query execute plan
836 </li><li>Query results
838 </li><li>Allow sequential scans to take advantage of other concurrent
839 sequential scans, also called "Synchronised Scanning"
840 <p> One possible implementation is to start sequential scans from the lowest
841 numbered buffer in the shared cache, and when reaching the end wrap
842 around to the beginning, rather than always starting sequential scans
843 at the start of the table.
846 <h1><a name="section_16">Vacuum</a></h1>
849 <li>Improve speed with indexes
850 <p> For large table adjustments during VACUUM FULL, it is faster to
851 reindex rather than update the index.
853 </li><li>Reduce lock time during VACUUM FULL by moving tuples with read lock,
854 then write lock and truncate table
855 <p> Moved tuples are invisible to other backends so they don't require a
856 write lock. However, the read lock promotion to write lock could lead
857 to deadlock situations.
859 </li><li>Auto-fill the free space map by scanning the buffer cache or by
860 checking pages written by the background writer
861 </li><li>Create a bitmap of pages that need vacuuming
862 <p> Instead of sequentially scanning the entire table, have the background
863 writer or some other process record pages that have expired rows, then
864 VACUUM can look at just those pages rather than the entire table. In
865 the event of a system crash, the bitmap would probably be invalidated.
866 One complexity is that index entries still have to be vacuumed, and
867 doing this without an index scan (by using the heap values to find the
868 index entry) might be slow and unreliable, especially for user-defined
871 </li><li>-<em>Add system view to show free space map contents</em>
872 </li><li>Allow FSM to return free space toward the beginning of the heap file,
873 in hopes that empty pages at the end can be truncated by VACUUM
874 </li><li>Allow FSM page return free space based on table clustering, to assist
875 in maintaining clustering?
878 <li>Use free-space map information to guide refilling
879 </li><li>%Issue log message to suggest VACUUM FULL if a table is nearly
881 </li><li>Improve xid wraparound detection by recording per-table rather
883 </li><li>Consider logging activity either to the logs or a system view
886 <h1><a name="section_17">Locking</a></h1>
889 <li>Fix priority ordering of read and write light-weight locks (Neil)
891 <h1><a name="section_18">Startup Time Improvements</a></h1>
894 <li>Experiment with multi-threaded backend [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?thread">thread</a>]
895 <p> This would prevent the overhead associated with process creation. Most
896 operating systems have trivial process creation time compared to
897 database startup overhead, but a few operating systems (Win32,
898 Solaris) might benefit from threading. Also explore the idea of
899 a single session using multiple threads to execute a statement faster.
901 </li><li>Add connection pooling
902 <p> It is unclear if this should be done inside the backend code or done
903 by something external like pgpool. The passing of file descriptors to
904 existing backends is one of the difficulties with a backend approach.
907 <h1><a name="section_19">Write-Ahead Log</a></h1>
910 <li>Eliminate need to write full pages to WAL before page modification [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?wal">wal</a>]
911 <p> Currently, to protect against partial disk page writes, we write
912 full page images to WAL before they are modified so we can correct any
913 partial page writes during recovery. These pages can also be
914 eliminated from point-in-time archive files.
917 <li>When off, write CRC to WAL and check file system blocks
919 <p> If CRC check fails during recovery, remember the page in case
920 a later CRC for that page properly matches.
922 </li><li>Write full pages during file system write and not when
923 the page is modified in the buffer cache
924 <p> This allows most full page writes to happen in the background
925 writer. It might cause problems for applying WAL on recovery
926 into a partially-written page, but later the full page will be
930 </li><li>Allow WAL traffic to be streamed to another server for stand-by
932 </li><li>Reduce WAL traffic so only modified values are written rather than
934 </li><li>Allow the pg_xlog directory location to be specified during initdb
935 with a symlink back to the /data location
936 </li><li>Allow WAL information to recover corrupted pg_controldata
937 </li><li>Find a way to reduce rotational delay when repeatedly writing
939 <p> Currently fsync of WAL requires the disk platter to perform a full
940 rotation to fsync again. One idea is to write the WAL to different
941 offsets that might reduce the rotational delay.
943 </li><li>Allow buffered WAL writes and fsync
944 <p> Instead of guaranteeing recovery of all committed transactions, this
945 would provide improved performance by delaying WAL writes and fsync
946 so an abrupt operating system restart might lose a few seconds of
947 committed transactions but still be consistent. We could perhaps
948 remove the 'fsync' parameter (which results in an an inconsistent
949 database) in favor of this capability.
951 </li><li>Allow WAL logging to be turned off for a table, but the table
952 might be dropped or truncated during crash recovery [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?walcontrol">walcontrol</a>]
953 <p> Allow tables to bypass WAL writes and just fsync() dirty pages on
954 commit. This should be implemented using ALTER TABLE, e.g. ALTER
955 TABLE PERSISTENCE [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo? DROP | TRUNCATE | DEFAULT "> DROP | TRUNCATE | DEFAULT </a>]. Tables using
956 non-default logging should not use referential integrity with
957 default-logging tables. A table without dirty buffers during a
958 crash could perhaps avoid the drop/truncate.
960 </li><li>Allow WAL logging to be turned off for a table, but the table would
961 avoid being truncated/dropped [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?walcontrol">walcontrol</a>]
962 <p> To do this, only a single writer can modify the table, and writes
963 must happen only on new pages so the new pages can be removed during
964 crash recovery. Readers can continue accessing the table. Such
965 tables probably cannot have indexes. One complexity is the handling
966 of indexes on TOAST tables.
969 <h1><a name="section_20">Optimizer / Executor</a></h1>
972 <li>Improve selectivity functions for geometric operators
973 </li><li>Allow ORDER BY ... LIMIT # to select high/low value without sort or
974 index using a sequential scan for highest/lowest values
975 <p> Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
976 all values to return the high/low value. Instead The idea is to do a
977 sequential scan to find the high/low value, thus avoiding the sort.
978 MIN/MAX already does this, but not for LIMIT > 1.
980 </li><li>Precompile SQL functions to avoid overhead
981 </li><li>Create utility to compute accurate random_page_cost value
982 </li><li>Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
983 </li><li>Have EXPLAIN ANALYZE highlight poor optimizer estimates
984 </li><li>Consider using hash buckets to do DISTINCT, rather than sorting
985 <p> This would be beneficial when there are few distinct values. This is
986 already used by GROUP BY.
988 </li><li>Log statements where the optimizer row estimates were dramatically
989 different from the number of rows actually found?
991 <h1><a name="section_21">Miscellaneous Performance</a></h1>
994 <li>Do async I/O for faster random read-ahead of data
995 <p> Async I/O allows multiple I/O requests to be sent to the disk with
996 results coming back asynchronously.
998 </li><li>Use mmap() rather than SYSV shared memory or to write WAL files?
999 <p> This would remove the requirement for SYSV SHM but would introduce
1000 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
1001 to prevent I/O overhead.
1003 </li><li>Consider mmap()'ing files into a backend?
1004 <p> Doing I/O to large tables would consume a lot of address space or
1005 require frequent mapping/unmapping. Extending the file also causes
1006 mapping problems that might require mapping only individual pages,
1007 leading to thousands of mappings. Another problem is that there is no
1008 way to _prevent_ I/O to disk from the dirty shared buffers so changes
1009 could hit disk before WAL is written.
1011 </li><li>Add a script to ask system configuration questions and tune postgresql.conf
1012 </li><li>Merge xmin/xmax/cmin/cmax back into three header fields
1013 <p> Before subtransactions, there used to be only three fields needed to
1014 store these four values. This was possible because only the current
1015 transaction looks at the cmin/cmax values. If the current transaction
1016 created and expired the row the fields stored where xmin (same as
1017 xmax), cmin, cmax, and if the transaction was expiring a row from a
1018 another transaction, the fields stored were xmin (cmin was not
1019 needed), xmax, and cmax. Such a system worked because a transaction
1020 could only see rows from another completed transaction. However,
1021 subtransactions can see rows from outer transactions, and once the
1022 subtransaction completes, the outer transaction continues, requiring
1023 the storage of all four fields. With subtransactions, an outer
1024 transaction can create a row, a subtransaction expire it, and when the
1025 subtransaction completes, the outer transaction still has to have
1026 proper visibility of the row's cmin, for example, for cursors.
1028 <p> One possible solution is to create a phantom cid which represents a
1029 cmin/cmax pair and is stored in local memory. Another idea is to
1030 store both cmin and cmax only in local memory.
1032 </li><li>Research storing disk pages with no alignment/padding
1034 <h1><a name="section_22">Source Code</a></h1>
1037 <li>Add use of 'const' for variables in source tree
1038 </li><li>Rename some /contrib modules from pg* to pg_*
1039 </li><li>Move some things from /contrib into main tree
1040 </li><li>Move some /contrib modules out to their own project sites
1041 </li><li>%Remove warnings created by -Wcast-align
1042 </li><li>Move platform-specific ps status display info from ps_status.c to ports
1043 </li><li>Add optional CRC checksum to heap and index pages
1044 </li><li>Improve documentation to build only interfaces (Marc)
1045 </li><li>Remove or relicense modules that are not under the BSD license, if possible
1046 </li><li>%Remove memory/file descriptor freeing before ereport(ERROR)
1047 </li><li>Acquire lock on a relation before building a relcache entry for it
1048 </li><li>%Promote debug_query_string into a server-side function current_query()
1049 </li><li>%Allow the identifier length to be increased via a configure option
1050 </li><li>Allow cross-compiling by generating the zic database on the target system
1051 </li><li>Improve NLS maintenance of libpgport messages linked onto applications
1052 </li><li>Allow ecpg to work with MSVC and BCC
1053 </li><li>Add xpath_array() to /contrib/xml2 to return results as an array
1054 </li><li>Allow building in directories containing spaces
1055 <p> This is probably not possible because 'gmake' and other compiler tools
1056 do not fully support quoting of paths with spaces.
1058 </li><li>-<em>Allow installing to directories containing spaces</em>
1059 <p> This is possible if proper quoting is added to the makefiles for the
1060 install targets. Because PostgreSQL supports relocatable installs, it
1061 is already possible to install into a directory that doesn't contain
1062 spaces and then copy the install to a directory with spaces.
1064 </li><li>Fix sgmltools so PDFs can be generated with bookmarks
1065 </li><li>-*%Clean up compiler warnings (especially with gcc version 4)*
1066 </li><li>Use UTF8 encoding for NLS messages so all server encodings can
1068 </li><li>Update Bonjour to work with newer cross-platform SDK
1069 </li><li>-<em>Remove BeOS and QNX-specific code</em>
1070 </li><li>Split out libpq pgpass and environment documentation sections to make
1071 it easier for non-developers to find
1072 </li><li>Improve port/qsort() to handle sorts with 50% unique and 50% duplicate
1073 value [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?qsort">qsort</a>]
1074 <p> This involves choosing better pivot points for the quicksort.
1078 <li>Remove configure.in check for link failure when cause is found
1079 </li><li>Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1081 </li><li>Remove psql newline patch when we find out why mingw outputs an
1083 </li><li>Allow psql to use readline once non-US code pages work with
1085 </li><li>Re-enable timezone output on log_line_prefix '%t' when a
1086 shorter timezone string is available
1087 </li><li>Fix problem with shared memory on the Win32 Terminal Server
1088 </li><li>Improve signal handling,
1089 <a href="http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php">http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php</a>
1090 </li><li>Add long file support for binary pg_dump output
1091 <p> While Win32 supports 64-bit files, the MinGW API does not,
1092 meaning we have to build an fseeko replacement on top of the
1093 Win32 API, and we have to make sure MinGW handles it. Another
1094 option is to wait for the MinGW project to fix it, or use the
1095 code from the LibGW32C project as a guide.
1097 </li><li>Port contrib/xml2
1099 </li><li>Wire Protocol Changes
1101 <li>Allow dynamic character set handling
1102 </li><li>Add decoded type, length, precision
1103 </li><li>Use compression?
1104 </li><li>Update clients to use data types, typmod, <a href="http://schema.table.column">schema.table.column</a>/ names
1105 of result sets using new statement protocol
1110 <h2><a name="section_22_1">Developers who have claimed items are:</a></h2>
1112 <li>Alvaro is Alvaro Herrera <<a href="mailto:alvherre@dcc.uchile.cl">alvherre@dcc.uchile.cl</a>>
1113 </li><li>Andrew is Andrew Dunstan <<a href="mailto:andrew@dunslane.net">andrew@dunslane.net</a>>
1114 </li><li>Bruce is Bruce Momjian <<a href="mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</a>> of Software Research Assoc.
1115 </li><li>Christopher is Christopher Kings-Lynne <<a href="mailto:chriskl@familyhealth.com.au">chriskl@familyhealth.com.au</a>> of
1116 Family Health Network
1117 </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.
1118 </li><li>Fabien is Fabien Coelho <<a href="mailto:coelho@cri.ensmp.fr">coelho@cri.ensmp.fr</a>>
1119 </li><li>Gavin is Gavin Sherry <<a href="mailto:swm@linuxworld.com.au">swm@linuxworld.com.au</a>> of Alcove Systems Engineering
1120 </li><li>Greg is Greg Sabino Mullane <<a href="mailto:greg@turnstep.com">greg@turnstep.com</a>>
1121 </li><li>Jan is Jan Wieck <<a href="mailto:JanWieck@Yahoo.com">JanWieck@Yahoo.com</a>> of Afilias, Inc.
1122 </li><li>Joe is Joe Conway <<a href="mailto:mail@joeconway.com">mail@joeconway.com</a>>
1123 </li><li>Karel is Karel Zak <<a href="mailto:zakkr@zf.jcu.cz">zakkr@zf.jcu.cz</a>>
1124 </li><li>Magnus is Magnus Hagander <<a href="mailto:mha@sollentuna.net">mha@sollentuna.net</a>>
1125 </li><li>Marc is Marc Fournier <<a href="mailto:scrappy@hub.org">scrappy@hub.org</a>> of PostgreSQL, Inc.
1126 </li><li>Matthew T. O'Connor <<a href="mailto:matthew@zeut.net">matthew@zeut.net</a>>
1127 </li><li>Michael is Michael Meskes <<a href="mailto:meskes@postgresql.org">meskes@postgresql.org</a>> of Credativ
1128 </li><li>Neil is Neil Conway <<a href="mailto:neilc@samurai.com">neilc@samurai.com</a>>
1129 </li><li>Oleg is Oleg Bartunov <<a href="mailto:oleg@sai.msu.su">oleg@sai.msu.su</a>>
1130 </li><li>Peter is Peter Eisentraut <<a href="mailto:peter_e@gmx.net">peter_e@gmx.net</a>>
1131 </li><li>Philip is Philip Warner <<a href="mailto:pjw@rhyme.com.au">pjw@rhyme.com.au</a>> of Albatross Consulting Pty. Ltd.
1132 </li><li>Rod is Rod Taylor <<a href="mailto:pg@rbt.ca">pg@rbt.ca</a>>
1133 </li><li>Simon is Simon Riggs <<a href="mailto:simon@2ndquadrant.com">simon@2ndquadrant.com</a>>
1134 </li><li>Stephan is Stephan Szabo <<a href="mailto:sszabo@megazone23.bigpanda.com">sszabo@megazone23.bigpanda.com</a>>
1135 </li><li>Tatsuo is Tatsuo Ishii <<a href="mailto:t-ishii@sra.co.jp">t-ishii@sra.co.jp</a>> of Software Research Assoc.
1136 </li><li>Teodor is Teodor Sigaev <<a href="mailto:teodor@sigaev.ru">teodor@sigaev.ru</a>>
1137 </li><li>Tom is Tom Lane <<a href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>> of Red Hat
1139 </li></ul></li></ul>