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: Thu Jan 5 11:35:12 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>Improve replication solutions
55 <p> You can use any of the master/slave replication servers to use a
56 standby server for data warehousing. To allow read/write queries to
57 multiple servers, you need multi-master replication like pgcluster.
59 </li><li>Allow replication over unreliable or non-persistent links
61 </li><li>Configuration files
63 <li>%Add "include file" functionality in postgresql.conf
64 </li><li>%Allow commenting of variables in postgresql.conf to restore them
66 <p> Currently, if a variable is commented out, it keeps the
67 previous uncommented value until a server restarted.
69 </li><li>%Allow pg_hba.conf settings to be controlled via SQL
70 <p> This would add a function to load the SQL table from
71 pg_hba.conf, and one to writes its contents to the flat file.
72 The table should have a line number that is a float so rows
73 can be inserted between existing rows, e.g. row 2.5 goes
74 between row 2 and row 3.
76 </li><li>%Allow postgresql.conf file values to be changed via an SQL
77 API, perhaps using SET GLOBAL
78 </li><li>Allow the server to be stopped/restarted via an SQL API
79 </li><li>Issue a warning if a change-on-restart-only postgresql.conf value
80 is modified and the server config files are reloaded
81 </li><li>Mark change-on-restart-only values in postgresql.conf
85 <li>Allow a database in tablespace t1 with tables created in
86 tablespace t2 to be used as a template for a new database created
87 with default tablespace t2
88 <p> All objects in the default database tablespace must have default
89 tablespace specifications. This is because new databases are
90 created by copying directories. If you mix default tablespace
91 tables and tablespace-specified tables in the same directory,
92 creating a new database from such a mixed directory would create a
93 new database with tables that had incorrect explicit tablespaces.
94 To fix this would require modifying pg_class in the newly copied
95 database, which we don't currently do.
97 </li><li>Allow reporting of which objects are in which tablespaces
98 <p> This item is difficult because a tablespace can contain objects
99 from multiple databases. There is a server-side function that
100 returns the databases which use a specific tablespace, so this
101 requires a tool that will call that function and connect to each
102 database to find the objects in each database for that tablespace.
105 <li>%Add a GUC variable to control the tablespace for temporary objects
107 <p> It could start with a random tablespace from a supplied list and
108 cycle through the list.
110 </li><li>Allow WAL replay of CREATE TABLESPACE to work when the directory
111 structure on the recovery computer is different from the original
112 </li><li>Allow per-tablespace quotas
115 </li><li>Point-In-Time Recovery (PITR)
117 <li>Allow point-in-time recovery to archive partially filled
118 write-ahead logs [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
119 <p> Currently only full WAL files are archived. This means that the
120 most recent transactions aren't available for recovery in case
121 of a disk failure. This could be triggered by a user command or
124 </li><li>Automatically force archiving of partially-filled WAL files when
125 pg_stop_backup() is called or the server is stopped
126 <p> Doing this will allow administrators to know more easily when
127 the archive contains all the files needed for point-in-time
130 </li><li>%Create dump tool for write-ahead logs for use in determining
131 transaction id for point-in-time recovery
132 </li><li>Allow a warm standby system to also allow read-only statements
133 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?pitr">pitr</a>]
134 <p> This is useful for checking PITR recovery.
136 </li><li>Allow the PITR process to be debugged and data examined
139 <h1><a name="section_3">Monitoring</a></h1>
142 <li>Allow server log information to be output as INSERT statements
143 <p> This would allow server log information to be easily loaded into
144 a database for analysis.
146 </li><li>%Add ability to monitor the use of temporary sort files
147 </li><li>Allow server logs to be remotely read and removed using SQL commands
148 </li><li>Allow protocol-level BIND parameter values to be logged
150 <h1><a name="section_4">Data Types</a></h1>
153 <li>Improve the MONEY data type
154 <p> Change the MONEY data type to use DECIMAL internally, with special
155 locale-aware output formatting.
157 </li><li>Change NUMERIC to enforce the maximum precision
158 </li><li>Add NUMERIC division operator that doesn't round?
159 <p> Currently NUMERIC _rounds_ the result to the specified precision.
160 This means division can return a result that multiplied by the
161 divisor is greater than the dividend, e.g. this returns a value > 10:
164 <p> SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
166 <p> The positive modulus result returned by NUMERICs might be considered<br/>
167 inaccurate, in one sense.
170 <li>%Disallow changing default expression of a SERIAL column?
171 </li><li>Fix data types where equality comparison isn't intuitive, e.g. box
172 </li><li>%Prevent INET cast to CIDR if the unmasked bits are not zero, or
174 </li><li>%Prevent INET cast to CIDR from dropping netmask, SELECT '<a href="telnet://1.1.1.1">1.1.1.1</a>'::inet::cidr
175 </li><li>Allow INET + INT4 to increment the host part of the address, or
176 throw an error on overflow
177 </li><li>%Add 'tid != tid ' operator for use in corruption recovery
178 </li><li>Allow user-defined types to specify a type modifier at table creation
180 </li><li>Dates and Times
182 <li>Allow infinite dates just like infinite timestamps
183 </li><li>Merge hardwired timezone names with the TZ database; allow either
184 kind everywhere a TZ name is currently taken
185 </li><li>Allow customization of the known set of TZ names (generalize the
186 present australian_timezones hack)
187 </li><li>Allow TIMESTAMP WITH TIME ZONE to store the original timezone
188 information, either zone name or offset from UTC [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?timezone">timezone</a>]
189 <p> If the TIMESTAMP value is stored with a time zone name, interval
190 computations should adjust based on the time zone rules.
192 </li><li>Fix SELECT '0.01 years'::interval, '0.01 months'::interval
193 </li><li>Fix SELECT INTERVAL '1' MONTH
194 </li><li>Add a GUC variable to allow output of interval values in ISO8601
196 </li><li>Improve timestamptz subtraction to be DST-aware
197 <p> Currently, subtracting one date from another that crosses a
198 daylight savings time adjustment can return '1 day 1 hour', but
199 adding that back to the first date returns a time one hour in
200 the future. This is caused by the adjustment of '25 hours' to
201 '1 day 1 hour', and '1 day' is the same time the next day, even
202 if daylight savings adjustments are involved.
204 </li><li>Fix interval display to support values exceeding 2^31 hours
205 </li><li>Add overflow checking to timestamp and interval arithmetic
206 </li><li>Add ISO INTERVAL handling
208 <li>Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO
210 </li><li>Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
211 </li><li>For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
212 '1', treat as ISO if there is a range specification clause,
213 and as PG if there no clause is present, e.g. interpret
214 <p> '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and
215 interpret '1:30' as '1 hour, 30 minutes'
216 <li>Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
217 INTERVAL MONTH), and this should return '12 months'
218 </li><li>Round or truncate values to the requested precision, e.g.
219 INTERVAL '11 months' AS YEAR should return one or zero
220 </li><li>Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))
225 <li>-<em>Allow NULLs in arrays</em>
226 </li><li>Delay resolution of array expression's data type so assignment
227 coercion can be performed on empty array expressions
231 <li>Improve vacuum of large objects, like /contrib/vacuumlo?
232 </li><li>Add security checking for large objects
233 </li><li>Auto-delete large objects when referencing row is deleted
234 <p> /contrib/lo offers this functionality.
236 </li><li>Allow read/write into TOAST values like large objects
237 <p> This requires the TOAST column to be stored EXTERNAL.
241 <h1><a name="section_5">Functions</a></h1>
244 <li>Allow INET subnet tests using non-constants to be indexed
245 </li><li>Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
247 <p> Current CURRENT_TIMESTAMP returns the start time of the current
248 transaction, and gettimeofday() returns the wallclock time. This will
249 make time reporting more consistent and will allow reporting of
250 the statement start time.
252 </li><li>%Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
253 pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
254 </li><li>Allow to_char() to print localized month names
255 </li><li>Allow functions to have a schema search path specified at creation time
256 </li><li>Allow substring/replace() to get/set bit values
257 </li><li>Allow to_char() on interval values to accumulate the highest unit
259 <p> Some special format flag would be required to request such
260 accumulation. Such functionality could also be added to EXTRACT.
261 Prevent accumulation that crosses the month/day boundary because of
262 the uneven number of days in a month.
265 <li>to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
266 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600
267 </li><li>to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
268 </li><li>to_char(INTERVAL '3 years 5 months','MM') => 41
270 </li><li>Add sleep() function, remove from regress.c
271 </li><li>Allow user-defined functions retuning a domain value to enforce domain
273 </li><li>Add SPI_gettypmod() to return the typemod for a TupleDesc
275 <h1><a name="section_6">Multi-Language Support</a></h1>
278 <li>Add NCHAR (as distinguished from ordinary varchar),
279 </li><li>Allow locale to be set at database creation
280 <p> Currently locale can only be set during initdb. No global tables have
281 locale-aware columns. However, the database template used during
282 database creation might have locale-aware indexes. The indexes would
283 need to be reindexed to match the new locale.
285 </li><li>Allow encoding on a per-column basis
286 <p> Right now only one encoding is allowed per database.
288 </li><li>Support multiple simultaneous character sets, per SQL92
289 </li><li>Improve UTF8 combined character handling?
290 </li><li>Add octet_length_server() and octet_length_client()
291 </li><li>Make octet_length_client() the same as octet_length()?
292 </li><li>Fix problems with wrong runtime encoding conversion for NLS message files
294 <h1><a name="section_7">Views / Rules</a></h1>
297 <li>%Automatically create rules on views so they are updateable, per SQL99
298 <p> We can only auto-create rules for simple views. For more complex
299 cases users will still have to write rules.
301 </li><li>Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
302 </li><li>Allow NOTIFY in rules involving conditionals
303 </li><li>Allow VIEW/RULE recompilation when the underlying tables change
304 <p> Another issue is whether underlying table changes should be reflected
305 in the view, e.g. should SELECT * show additional columns if they
306 are added after the view is created.
309 <h1><a name="section_8">SQL Commands</a></h1>
312 <li>Change LIMIT/OFFSET and FETCH/MOVE to use int8
313 </li><li>Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
314 </li><li>Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
315 </li><li>%Allow SET CONSTRAINTS to be qualified by schema/table name
316 </li><li>%Allow TRUNCATE ... CASCADE/RESTRICT
317 <p> This is like DELETE CASCADE, but truncates.
319 </li><li>%Add a separate TRUNCATE permission
320 <p> Currently only the owner can TRUNCATE a table because triggers are not
321 called, and the table is locked in exclusive mode.
323 </li><li>Allow PREPARE of cursors
324 </li><li>Allow PREPARE to automatically determine parameter types based on the SQL
326 </li><li>Allow finer control over the caching of prepared query plans
327 <p> Currently, queries prepared via the libpq API are planned on first
328 execute using the supplied parameters --- allow SQL PREPARE to do the
329 same. Also, allow control over replanning prepared queries either
330 manually or automatically when statistics for execute parameters
331 differ dramatically from those used during planning.
333 </li><li>Allow LISTEN/NOTIFY to store info in memory rather than tables?
334 <p> Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
335 such information in memory would improve performance.
337 </li><li>Add optional textual message to NOTIFY
338 <p> This would allow an informational message to be added to the notify
339 message, perhaps indicating the row modified or other custom
342 </li><li>Add a GUC variable to warn about non-standard SQL usage in queries
343 </li><li>Add SQL-standard MERGE command, typically used to merge two tables
344 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?merge">merge</a>]
345 <p> This is similar to UPDATE, then for unmatched rows, INSERT.
346 Whether concurrent access allows modifications which could cause
347 row loss is implementation independent.
349 </li><li>Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
350 [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?merge">merge</a>]
351 <p> To implement this cleanly requires that the table have a unique index
352 so duplicate checking can be easily performed. It is possible to
353 do it without a unique index if we require the user to LOCK the table
356 </li><li>Add NOVICE output level for helpful messages like automatic sequence/index
358 </li><li>%Add COMMENT ON for all cluster global objects (roles, databases
360 </li><li>-<em>Make row-wise comparisons work per SQL spec</em>
361 <p> Right now, '(a, b) < (1, 2)' is processed as 'a < 1 and b < 2', but
362 the SQL standard requires it to be processed as a column-by-column
363 comparison, so the proper comparison is '(a < 1) OR (a = 1 AND b < 2)'.
365 </li><li>Add RESET CONNECTION command to reset all session state
366 <p> This would include resetting of all variables (RESET ALL), dropping of
367 temporary tables, removing any NOTIFYs, cursors, open transactions,
368 prepared queries, currval()s, etc. This could be used for connection
369 pooling. We could also change RESET ALL to have this functionality.
370 The difficult of this features is allowing RESET ALL to not affect
371 changes made by the interface driver for its internal use. One idea
372 is for this to be a protocol-only feature. Another approach is to
373 notify the protocol when a RESET CONNECTION command is used.
375 </li><li>Add GUC to issue notice about statements that use unjoined tables
376 </li><li>Allow EXPLAIN to identify tables that were skipped because of
378 </li><li>Allow EXPLAIN output to be more easily processed by scripts
379 </li><li>Eventually enable escape_string<u>warning and standard</u>conforming_strings
380 </li><li>Simplify dropping roles that have objects in several databases
381 </li><li>Allow COMMENT ON to accept an expression rather than just a string
382 </li><li>Allow the count returned by SELECT, etc to be to represent as an int64
383 to allow a higher range of values
384 </li><li>Make CLUSTER preserve recently-dead tuples per MVCC requirements
387 <li>Allow CREATE TABLE AS to determine column lengths for complex
388 expressions like SELECT col1 || col2
389 </li><li>Use more reliable method for CREATE DATABASE to get a consistent
391 </li><li>Add ON COMMIT capability to CREATE TABLE AS ... SELECT
395 <li>Allow UPDATE to handle complex aggregates [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?update">update</a>]?
396 </li><li>Allow an alias to be provided for the target table in
398 <p> This is not SQL-spec but many DBMSs allow it.
400 </li><li>Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
405 <li>%Have ALTER TABLE RENAME rename SERIAL sequence names
406 </li><li>Add ALTER DOMAIN to modify the underlying data type
407 </li><li>%Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
408 </li><li>%Allow ALTER TABLE to change constraint deferrability and actions
409 </li><li>Add missing object types for ALTER ... SET SCHEMA
410 </li><li>Allow ALTER TABLESPACE to move to different directories
411 </li><li>Allow databases to be moved to different tablespaces
412 </li><li>Allow moving system tables to other tablespaces, where possible
413 <p> Currently non-global system tables must be in the default database
414 tablespace. Global system tables can never be moved.
416 </li><li>%Disallow dropping of an inherited constraint
417 </li><li>%Prevent child tables from altering or dropping constraints
418 like CHECK that were inherited from the parent table
419 </li><li>Have ALTER INDEX update the name of a constraint using that index
420 </li><li>Add ALTER TABLE RENAME CONSTRAINT, update index name also
424 <li>Automatically maintain clustering on a table
425 <p> This might require some background daemon to maintain clustering
426 during periods of low usage. It might also require tables to be only
427 partially filled for easier reorganization. Another idea would
428 be to create a merged heap/index data file so an index lookup would
429 automatically access the heap data too. A third idea would be to
430 store heap rows in hashed groups, perhaps using a user-supplied
433 </li><li>%Add default clustering to system tables
434 <p> To do this, determine the ideal cluster index for each system
435 table and set the cluster setting during initdb.
440 <li>Allow COPY to report error lines and continue
441 <p> This requires the use of a savepoint before each COPY line is
442 processed, with ROLLBACK on COPY failure.
444 </li><li>%Have COPY return the number of rows loaded/unloaded?
445 </li><li>Allow COPY on a newly-created table to skip WAL logging
446 <p> On crash recovery, the table involved in the COPY would
447 be removed or have its heap and index files truncated. One
448 issue is that no other backend should be able to add to
449 the table at the same time, which is something that is
452 </li><li>Allow COPY to output from views
453 <p> Another idea would be to allow actual SELECT statements in a COPY.
456 </li><li>GRANT/REVOKE
458 <li>Allow column-level privileges
459 </li><li>%Allow GRANT/REVOKE permissions to be applied to all schema objects
461 <p> The proposed syntax is:
462 </p><p> GRANT SELECT ON ALL TABLES IN public TO phpuser;
463 GRANT SELECT ON NEW TABLES IN public TO phpuser;
466 <li>Allow GRANT/REVOKE permissions to be inherited by objects based on
472 <li>Allow UPDATE/DELETE WHERE CURRENT OF cursor
473 <p> This requires using the row ctid to map cursor rows back to the
474 original heap row. This become more complicated if WITH HOLD cursors
475 are to be supported because WITH HOLD cursors have a copy of the row
476 and no FOR UPDATE lock.
478 </li><li>Prevent DROP TABLE from dropping a row referenced by its own open
480 </li><li>%Allow pooled connections to list all open WITH HOLD cursors
481 <p> Because WITH HOLD cursors exist outside transactions, this allows
482 them to be listed so they can be closed.
487 <li>Allow INSERT/UPDATE of the system-generated oid value for a row
488 </li><li>Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
489 </li><li>Allow INSERT/UPDATE ... RETURNING new.col or old.col
490 <p> This is useful for returning the auto-generated key for an INSERT.
491 One complication is how to handle rules that run as part of
497 <li>Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
499 </li><li>Add SET PATH for schemas?
500 <p> This is basically the same as SET search_path.
503 </li><li>Server-Side Languages
505 <li>Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
506 </li><li>Allow function parameters to be passed by name,
507 get_employee_salary(emp_id => 12345, tax_year => 2001)
508 </li><li>Add Oracle-style packages
509 </li><li>Add table function support to pltcl, plpython
510 </li><li>Add capability to create and call PROCEDURES
511 </li><li>Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?"></a>]
512 </li><li>Allow function argument names to be statements from PL/PgSQL
513 </li><li>Add MOVE to PL/pgSQL
514 </li><li>Add support for polymorphic arguments and return types to
515 languages other than PL/PgSQL
516 </li><li>Add support for OUT and INOUT parameters to languages other
518 </li><li>Add single-step debugging of PL/PgSQL functions
519 </li><li>Allow PL/PgSQL to support WITH HOLD cursors
522 <h1><a name="section_9">Clients</a></h1>
525 <li>-<em>Have initdb set the input DateStyle (MDY or DMY) based on locale</em>
526 </li><li>Have pg_ctl look at PGHOST in case it is a socket directory?
527 </li><li>Allow pg_ctl to work properly with configuration files located outside
529 <p> pg_ctl can not read the pid file because it isn't located in the
530 config directory but in the PGDATA directory. The solution is to
531 allow pg_ctl to read and understand postgresql.conf to find the
532 data_directory value.
536 <li>Have psql show current values for a sequence
537 </li><li>Move psql backslash database information into the backend, use
538 mnemonic commands? [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?psql">psql</a>]
539 <p> This would allow non-psql clients to pull the same information out
540 of the database as psql.
542 </li><li>Fix psql's display of schema information (Neil)
543 </li><li>Allow psql \pset boolean variables to set to fixed values, rather
545 </li><li>Consistently display privilege information for all objects in psql
546 </li><li>Improve psql's handling of multi-line statements
547 <p> Currently, while \e saves a single statement as one entry, interactive
548 statements are saved one line at a time. Ideally all statements
549 would be saved like \e does.
551 </li><li>Allow multi-line column values to align in the proper columns
552 <p> If the second output column value is 'a\nb', the 'b' should appear
553 in the second display column, rather than the first column as it
556 </li><li>Display IN, INOUT, and OUT parameters in \df+
557 <p> It probably requires psql to output newlines in the proper
558 column, which is already on the TODO list.
560 </li><li>Add auto-expanded mode so expanded output is used if the row
561 length is wider than the screen width.
562 <p> Consider using auto-expanded mode for backslash commands like \df+.
567 <li>%Have pg_dump use multi-statement transactions for INSERT dumps
568 </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>]
569 </li><li>%Add dumping of comments on index columns and composite type columns
570 </li><li>%Add full object name to the tag field. eg. for operators we need
571 '=(integer, integer)', instead of just '='.
572 </li><li>Add pg_dumpall custom format dumps?
573 </li><li>%Add CSV output format
574 </li><li>Update pg_dump and psql to use the new COPY libpq API (Christopher)
575 </li><li>Remove unnecessary function pointer abstractions in pg_dump source
577 </li><li>Allow selection of individual object(s) of all types, not just
579 </li><li>In a selective dump, allow dumping of an object and all its
581 </li><li>Add options like pg_restore -l and -L to pg_dump
582 </li><li>Stop dumping CASCADE on DROP TYPE commands in clean mode
583 </li><li>Allow pg_dump --clean to drop roles that own objects or have
585 </li><li>Add -f to pg_dumpall
590 <p> Document differences between ecpg and the SQL standard and
591 information about the Informix-compatibility module.
593 </li><li>Solve cardinality > 1 for input descriptors / variables?
594 </li><li>Add a semantic check level, e.g. check if a table really exists
595 </li><li>fix handling of DB attributes that are arrays
596 </li><li>Use backend PREPARE/EXECUTE facility for ecpg where possible
597 </li><li>Implement SQLDA
598 </li><li>Fix nested C comments
599 </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
600 </li><li>Make SET CONNECTION thread-aware, non-standard?
601 </li><li>Allow multidimensional arrays
602 </li><li>Add internationalized message strings
608 <li>Add a function to support Parse/DescribeStatement capability
609 </li><li>Add PQescapeIdentifier()
610 </li><li>Prevent PQfnumber() from lowercasing unquoted the column name
611 <p> PQfnumber() should never have been doing lowercasing, but
612 historically it has so we need a way to prevent it
614 </li><li>Allow statement results to be automatically batched to the client
615 <p> Currently, all statement results are transferred to the libpq
616 client before libpq makes the results available to the
617 application. This feature would allow the application to make
618 use of the first result rows while the rest are transferred, or
619 held on the server waiting for them to be requested by libpq.
620 One complexity is that a statement like SELECT 1/col could error
621 out mid-way through the result set.
624 <h1><a name="section_10">Referential Integrity</a></h1>
627 <li>Add MATCH PARTIAL referential integrity
628 </li><li>Add deferred trigger queue file
629 <p> Right now all deferred trigger information is stored in backend
630 memory. This could exhaust memory for very large trigger queues.
631 This item involves dumping large queues into files.
633 </li><li>Change foreign key constraint for array -> element to mean element
635 </li><li>Allow DEFERRABLE UNIQUE constraints?
636 </li><li>Allow triggers to be disabled in only the current session.
637 <p> This is currently possible by starting a multi-statement transaction,
638 modifying the system tables, performing the desired SQL, restoring the
639 system tables, and committing the transaction. ALTER TABLE ...
640 TRIGGER requires a table lock so it is not ideal for this usage.
642 </li><li>With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
643 <p> If the dump is known to be valid, allow foreign keys to be added
644 without revalidating the data.
646 </li><li>Allow statement-level triggers to access modified rows
647 </li><li>Support triggers on columns (Greg Sabino Mullane)
648 </li><li>Enforce referential integrity for system tables
649 </li><li>Allow AFTER triggers on system tables
650 <p> System tables are modified in many places in the backend without going
651 through the executor and therefore not causing triggers to fire. To
652 complete this item, the functions that modify system tables will have
656 <h1><a name="section_11">Dependency Checking</a></h1>
659 <li>Flush cached query plans when the dependent objects change,
660 when the cardinality of parameters changes dramatically, or
661 when new ANALYZE statistics are available
662 <p> A more complex solution would be to save multiple plans for different
663 cardinality and use the appropriate plan based on the EXECUTE values.
665 </li><li>Track dependencies in function bodies and recompile/invalidate
666 <p> This is particularly important for references to temporary tables
667 in PL/PgSQL because PL/PgSQL caches query plans. The only workaround
668 in PL/PgSQL is to use EXECUTE. One complexity is that a function
669 might itself drop and recreate dependent tables, causing it to
670 invalidate its own query plan.
673 <h1><a name="section_12">Exotic Features</a></h1>
676 <li>Add SQL99 WITH clause to SELECT
677 </li><li>Add SQL99 WITH RECURSIVE to SELECT
678 </li><li>Add pre-parsing phase that converts non-ISO syntax to supported
680 <p> This could allow SQL written for other databases to run without
683 </li><li>Allow plug-in modules to emulate features from other databases
684 </li><li>SQL*Net listener that makes PostgreSQL appear as an Oracle database
686 </li><li>Allow statements across databases or servers with transaction
688 <p> This can be done using dblink and two-phase commit.
690 </li><li>Add the features of packages
692 <li>Make private objects accessible only to objects in the same schema
693 </li><li>Allow current_schema.objname to access current schema objects
694 </li><li>Add session variables
695 </li><li>Allow nested schemas
698 <h1><a name="section_13">Indexes</a></h1>
701 <li>Allow inherited tables to inherit index, UNIQUE constraint, and primary
703 </li><li>UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
704 inherited table: INSERT INTO inherit_table (unique_index_col) VALUES
706 <p> The main difficulty with this item is the problem of creating an index
707 that can span more than one table.
709 </li><li>Allow SELECT ... FOR UPDATE on inherited tables
710 </li><li>Add UNIQUE capability to non-btree indexes
711 </li><li>Prevent index uniqueness checks when UPDATE does not modify the column
712 <p> Uniqueness (index) checks are done when updating a column even if the
713 column is not modified by the UPDATE.
715 </li><li>Allow the creation of on-disk bitmap indexes which can be quickly
716 combined with other bitmap indexes
717 <p> Such indexes could be more compact if there are only a few distinct values.
718 Such indexes can also be compressed. Keeping such indexes updated can be
721 </li><li>Allow use of indexes to search for NULLs
722 <p> One solution is to create a partial index on an IS NULL expression.
724 </li><li>Allow accurate statistics to be collected on indexes with more than
725 one column or expression indexes, perhaps using per-index statistics
726 </li><li>Add fillfactor to control reserved free space during index creation
727 </li><li>Allow the creation of indexes with mixed ascending/descending specifiers
728 </li><li>Allow constraint_exclusion to work for UNIONs like it does for
729 inheritance, allow it to work for UPDATE and DELETE statements, and allow
730 it to be used for all statements with little performance impact
731 </li><li>Allow CREATE INDEX to take an additional parameter for use with
733 </li><li>Consider compressing indexes by storing key values duplicated in
734 several rows as a single index entry
735 <p> This is difficult because it requires datatype-specific knowledge.
739 <li>Add more GIST index support for geometric data types
740 </li><li>Allow GIST indexes to create certain complex index types, like
741 digital trees (see Aoki)
745 <li>Pack hash index buckets onto disk pages more efficiently
746 <p> Currently only one hash bucket can be stored on a page. Ideally
747 several hash buckets could be stored on a single page and greater
748 granularity used for the hash algorithm.
750 </li><li>Consider sorting hash buckets so entries can be found using a
751 binary search, rather than a linear scan
752 </li><li>In hash indexes, consider storing the hash value with or instead
754 </li><li>Add WAL logging for crash recovery
755 </li><li>Allow multi-column hash indexes
758 <h1><a name="section_14">Fsync</a></h1>
761 <li>Improve commit_delay handling to reduce fsync()
762 </li><li>Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
763 <p> Ideally this requires a separate test program that can be run
764 at initdb time or optionally later. Consider O_SYNC when
767 </li><li>%Add an option to sync() before fsync()'ing checkpoint files
768 </li><li>Add program to test if fsync has a delay compared to non-fsync
770 <h1><a name="section_15">Cache Usage</a></h1>
773 <li>Allow free-behind capability for large sequential scans, perhaps using
775 <p> Posix_fadvise() can control both sequential/random file caching and
776 free-behind behavior, but it is unclear how the setting affects other
777 backends that also have the file open, and the feature is not supported
778 on all operating systems.
780 </li><li>Speed up COUNT(*)
781 <p> We could use a fixed row count and a +/- count to follow MVCC
782 visibility rules, or a single cached value could be used and
783 invalidated if anyone modifies the table. Another idea is to
784 get a count directly from a unique index, but for this to be
785 faster than a sequential scan it must avoid access to the heap
786 to obtain tuple visibility information.
788 </li><li>Add estimated_count(*) to return an estimate of COUNT(*)
789 <p> This would use the planner ANALYZE statistics to return an estimated
792 </li><li>Allow data to be pulled directly from indexes
793 <p> Currently indexes do not have enough tuple visibility information
794 to allow data to be pulled from the index without also accessing
795 the heap. One way to allow this is to set a bit on index tuples
796 to indicate if a tuple is currently visible to all transactions
797 when the first valid heap lookup happens. This bit would have to
798 be cleared when a heap tuple is expired.
800 <p> Another idea is to maintain a bitmap of heap pages where all rows
801 are visible to all backends, and allow index lookups to reference
802 that bitmap to avoid heap lookups, perhaps the same bitmap we might
803 add someday to determine which heap pages need vacuuming. Frequently
804 accessed bitmaps would have to be stored in shared memory. One 8k
805 page of bitmaps could track 512MB of heap pages.
807 </li><li>Consider automatic caching of statements at various levels:
809 <li>Parsed query tree
810 </li><li>Query execute plan
811 </li><li>Query results
813 </li><li>Allow sequential scans to take advantage of other concurrent
814 sequential scans, also called "Synchronised Scanning"
815 <p> One possible implementation is to start sequential scans from the lowest
816 numbered buffer in the shared cache, and when reaching the end wrap
817 around to the beginning, rather than always starting sequential scans
818 at the start of the table.
821 <h1><a name="section_16">Vacuum</a></h1>
824 <li>Improve speed with indexes
825 <p> For large table adjustments during VACUUM FULL, it is faster to
826 reindex rather than update the index.
828 </li><li>Reduce lock time during VACUUM FULL by moving tuples with read lock,
829 then write lock and truncate table
830 <p> Moved tuples are invisible to other backends so they don't require a
831 write lock. However, the read lock promotion to write lock could lead
832 to deadlock situations.
834 </li><li>Auto-fill the free space map by scanning the buffer cache or by
835 checking pages written by the background writer
836 </li><li>Create a bitmap of pages that need vacuuming
837 <p> Instead of sequentially scanning the entire table, have the background
838 writer or some other process record pages that have expired rows, then
839 VACUUM can look at just those pages rather than the entire table. In
840 the event of a system crash, the bitmap would probably be invalidated.
841 One complexity is that index entries still have to be vacuumed, and
842 doing this without an index scan (by using the heap values to find the
843 index entry) might be slow and unreliable, especially for user-defined
846 </li><li>%Add system view to show free space map contents
849 <li>Use free-space map information to guide refilling
850 </li><li>%Issue log message to suggest VACUUM FULL if a table is nearly
852 </li><li>Improve xid wraparound detection by recording per-table rather
854 </li><li>Consider logging activity either to the logs or a system view
857 <h1><a name="section_17">Locking</a></h1>
860 <li>Fix priority ordering of read and write light-weight locks (Neil)
862 <h1><a name="section_18">Startup Time Improvements</a></h1>
865 <li>Experiment with multi-threaded backend [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?thread">thread</a>]
866 <p> This would prevent the overhead associated with process creation. Most
867 operating systems have trivial process creation time compared to
868 database startup overhead, but a few operating systems (Win32,
869 Solaris) might benefit from threading. Also explore the idea of
870 a single session using multiple threads to execute a statement faster.
872 </li><li>Add connection pooling
873 <p> It is unclear if this should be done inside the backend code or done
874 by something external like pgpool. The passing of file descriptors to
875 existing backends is one of the difficulties with a backend approach.
878 <h1><a name="section_19">Write-Ahead Log</a></h1>
881 <li>Eliminate need to write full pages to WAL before page modification [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?wal">wal</a>]
882 <p> Currently, to protect against partial disk page writes, we write
883 full page images to WAL before they are modified so we can correct any
884 partial page writes during recovery. These pages can also be
885 eliminated from point-in-time archive files.
888 <li>When off, write CRC to WAL and check file system blocks
890 <p> If CRC check fails during recovery, remember the page in case
891 a later CRC for that page properly matches.
893 </li><li>Write full pages during file system write and not when
894 the page is modified in the buffer cache
895 <p> This allows most full page writes to happen in the background
896 writer. It might cause problems for applying WAL on recovery
897 into a partially-written page, but later the full page will be
901 </li><li>Allow WAL traffic to be streamed to another server for stand-by
903 </li><li>Reduce WAL traffic so only modified values are written rather than
905 </li><li>Allow the pg_xlog directory location to be specified during initdb
906 with a symlink back to the /data location
907 </li><li>Allow WAL information to recover corrupted pg_controldata
908 </li><li>Find a way to reduce rotational delay when repeatedly writing
910 <p> Currently fsync of WAL requires the disk platter to perform a full
911 rotation to fsync again. One idea is to write the WAL to different
912 offsets that might reduce the rotational delay.
914 </li><li>Allow buffered WAL writes and fsync
915 <p> Instead of guaranteeing recovery of all committed transactions, this
916 would provide improved performance by delaying WAL writes and fsync
917 so an abrupt operating system restart might lose a few seconds of
918 committed transactions but still be consistent. We could perhaps
919 remove the 'fsync' parameter (which results in an an inconsistent
920 database) in favor of this capability.
922 </li><li>Allow control over which tables are WAL-logged
923 <p> Allow tables to bypass WAL writes and just fsync() dirty pages on
924 commit. To do this, only a single writer can modify the table, and
925 writes must happen only on new pages. Readers can continue accessing
926 the table. This would affect COPY, and perhaps INSERT/UPDATE too.
927 Another option is to avoid transaction logging entirely and truncate
928 or drop the table on crash recovery. These should be implemented
929 using ALTER TABLE, e.g. ALTER TABLE PERSISTENCE [ DROP | TRUNCATE |
930 STABLE | DEFAULT ]. Tables using non-default logging should not use
931 referential integrity with default-logging tables, and tables using
932 stable logging probably can not have indexes. [<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?wallog">wallog</a>]
935 <h1><a name="section_20">Optimizer / Executor</a></h1>
938 <li>Improve selectivity functions for geometric operators
939 </li><li>Allow ORDER BY ... LIMIT # to select high/low value without sort or
940 index using a sequential scan for highest/lowest values
941 <p> Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
942 all values to return the high/low value. Instead The idea is to do a
943 sequential scan to find the high/low value, thus avoiding the sort.
944 MIN/MAX already does this, but not for LIMIT > 1.
946 </li><li>Precompile SQL functions to avoid overhead
947 </li><li>Create utility to compute accurate random_page_cost value
948 </li><li>Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
949 </li><li>Have EXPLAIN ANALYZE highlight poor optimizer estimates
950 </li><li>Consider using hash buckets to do DISTINCT, rather than sorting
951 <p> This would be beneficial when there are few distinct values. This is
952 already used by GROUP BY.
954 </li><li>Log statements where the optimizer row estimates were dramatically
955 different from the number of rows actually found?
957 <h1><a name="section_21">Miscellaneous Performance</a></h1>
960 <li>Do async I/O for faster random read-ahead of data
961 <p> Async I/O allows multiple I/O requests to be sent to the disk with
962 results coming back asynchronously.
964 </li><li>Use mmap() rather than SYSV shared memory or to write WAL files?
965 <p> This would remove the requirement for SYSV SHM but would introduce
966 portability issues. Anonymous mmap (or mmap to /dev/zero) is required
967 to prevent I/O overhead.
969 </li><li>Consider mmap()'ing files into a backend?
970 <p> Doing I/O to large tables would consume a lot of address space or
971 require frequent mapping/unmapping. Extending the file also causes
972 mapping problems that might require mapping only individual pages,
973 leading to thousands of mappings. Another problem is that there is no
974 way to _prevent_ I/O to disk from the dirty shared buffers so changes
975 could hit disk before WAL is written.
977 </li><li>Add a script to ask system configuration questions and tune postgresql.conf
978 </li><li>Merge xmin/xmax/cmin/cmax back into three header fields
979 <p> Before subtransactions, there used to be only three fields needed to
980 store these four values. This was possible because only the current
981 transaction looks at the cmin/cmax values. If the current transaction
982 created and expired the row the fields stored where xmin (same as
983 xmax), cmin, cmax, and if the transaction was expiring a row from a
984 another transaction, the fields stored were xmin (cmin was not
985 needed), xmax, and cmax. Such a system worked because a transaction
986 could only see rows from another completed transaction. However,
987 subtransactions can see rows from outer transactions, and once the
988 subtransaction completes, the outer transaction continues, requiring
989 the storage of all four fields. With subtransactions, an outer
990 transaction can create a row, a subtransaction expire it, and when the
991 subtransaction completes, the outer transaction still has to have
992 proper visibility of the row's cmin, for example, for cursors.
994 <p> One possible solution is to create a phantom cid which represents a
995 cmin/cmax pair and is stored in local memory. Another idea is to
996 store both cmin and cmax only in local memory.
998 </li><li>Research storing disk pages with no alignment/padding
1000 <h1><a name="section_22">Source Code</a></h1>
1003 <li>Add use of 'const' for variables in source tree
1004 </li><li>Rename some /contrib modules from pg* to pg_*
1005 </li><li>Move some things from /contrib into main tree
1006 </li><li>Move some /contrib modules out to their own project sites
1007 </li><li>%Remove warnings created by -Wcast-align
1008 </li><li>Move platform-specific ps status display info from ps_status.c to ports
1009 </li><li>Add optional CRC checksum to heap and index pages
1010 </li><li>Improve documentation to build only interfaces (Marc)
1011 </li><li>Remove or relicense modules that are not under the BSD license, if possible
1012 </li><li>%Remove memory/file descriptor freeing before ereport(ERROR)
1013 </li><li>Acquire lock on a relation before building a relcache entry for it
1014 </li><li>%Promote debug_query_string into a server-side function current_query()
1015 </li><li>%Allow the identifier length to be increased via a configure option
1016 </li><li>Allow cross-compiling by generating the zic database on the target system
1017 </li><li>Improve NLS maintenace of libpgport messages linked onto applications
1018 </li><li>Allow ecpg to work with MSVC and BCC
1019 </li><li>Add xpath_array() to /contrib/xml2 to return results as an array
1020 </li><li>Allow building in directories containing spaces
1021 <p> This is probably not possible because 'gmake' and other compiler tools
1022 do not fully support quoting of paths with spaces.
1024 </li><li>-<em>Allow installing to directories containing spaces</em>
1025 <p> This is possible if proper quoting is added to the makefiles for the
1026 install targets. Because PostgreSQL supports relocatable installs, it
1027 is already possible to install into a directory that doesn't contain
1028 spaces and then copy the install to a directory with spaces.
1030 </li><li>Fix sgmltools so PDFs can be generated with bookmarks
1031 </li><li>%Clean up compiler warnings (especially with gcc version 4)
1032 </li><li>Use UTF8 encoding for NLS messages so all server encodings can
1034 </li><li>Update Bonjour to work with newer cross-platform SDK
1035 </li><li>-<em>Remove BeOS and QNX-specific code</em>
1038 <li>Remove configure.in check for link failure when cause is found
1039 </li><li>Remove readdir() errno patch when runtime/mingwex/dirent.c rev
1041 </li><li>Remove psql newline patch when we find out why mingw outputs an
1043 </li><li>Allow psql to use readline once non-US code pages work with
1045 </li><li>Re-enable timezone output on log_line_prefix '%t' when a
1046 shorter timezone string is available
1047 </li><li>Fix problem with shared memory on the Win32 Terminal Server
1048 </li><li>Improve signal handling,
1049 <a href="http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php">http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php</a>
1050 </li><li>Add long file support for binary pg_dump output
1051 <p> While Win32 supports 64-bit files, the MinGW API does not,
1052 meaning we have to build an fseeko replacement on top of the
1053 Win32 API, and we have to make sure MinGW handles it. Another
1054 option is to wait for the MinGW project to fix it, or use the
1055 code from the LibGW32C project as a guide.
1058 </li><li>Wire Protocol Changes
1060 <li>Allow dynamic character set handling
1061 </li><li>Add decoded type, length, precision
1062 </li><li>Use compression?
1063 </li><li>Update clients to use data types, typmod, <a href="http://schema.table.column">schema.table.column</a>/ names
1064 of result sets using new statement protocol
1069 <h2><a name="section_22_1">Developers who have claimed items are:</a></h2>
1071 <li>Alvaro is Alvaro Herrera <<a href="mailto:alvherre@dcc.uchile.cl">alvherre@dcc.uchile.cl</a>>
1072 </li><li>Andrew is Andrew Dunstan <<a href="mailto:andrew@dunslane.net">andrew@dunslane.net</a>>
1073 </li><li>Bruce is Bruce Momjian <<a href="mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</a>> of Software Research Assoc.
1074 </li><li>Christopher is Christopher Kings-Lynne <<a href="mailto:chriskl@familyhealth.com.au">chriskl@familyhealth.com.au</a>> of
1075 Family Health Network
1076 </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.
1077 </li><li>Fabien is Fabien Coelho <<a href="mailto:coelho@cri.ensmp.fr">coelho@cri.ensmp.fr</a>>
1078 </li><li>Gavin is Gavin Sherry <<a href="mailto:swm@linuxworld.com.au">swm@linuxworld.com.au</a>> of Alcove Systems Engineering
1079 </li><li>Greg is Greg Sabino Mullane <<a href="mailto:greg@turnstep.com">greg@turnstep.com</a>>
1080 </li><li>Jan is Jan Wieck <<a href="mailto:JanWieck@Yahoo.com">JanWieck@Yahoo.com</a>> of Afilias, Inc.
1081 </li><li>Joe is Joe Conway <<a href="mailto:mail@joeconway.com">mail@joeconway.com</a>>
1082 </li><li>Karel is Karel Zak <<a href="mailto:zakkr@zf.jcu.cz">zakkr@zf.jcu.cz</a>>
1083 </li><li>Magnus is Magnus Hagander <<a href="mailto:mha@sollentuna.net">mha@sollentuna.net</a>>
1084 </li><li>Marc is Marc Fournier <<a href="mailto:scrappy@hub.org">scrappy@hub.org</a>> of PostgreSQL, Inc.
1085 </li><li>Matthew T. O'Connor <<a href="mailto:matthew@zeut.net">matthew@zeut.net</a>>
1086 </li><li>Michael is Michael Meskes <<a href="mailto:meskes@postgresql.org">meskes@postgresql.org</a>> of Credativ
1087 </li><li>Neil is Neil Conway <<a href="mailto:neilc@samurai.com">neilc@samurai.com</a>>
1088 </li><li>Oleg is Oleg Bartunov <<a href="mailto:oleg@sai.msu.su">oleg@sai.msu.su</a>>
1089 </li><li>Peter is Peter Eisentraut <<a href="mailto:peter_e@gmx.net">peter_e@gmx.net</a>>
1090 </li><li>Philip is Philip Warner <<a href="mailto:pjw@rhyme.com.au">pjw@rhyme.com.au</a>> of Albatross Consulting Pty. Ltd.
1091 </li><li>Rod is Rod Taylor <<a href="mailto:pg@rbt.ca">pg@rbt.ca</a>>
1092 </li><li>Simon is Simon Riggs <<a href="mailto:simon@2ndquadrant.com">simon@2ndquadrant.com</a>>
1093 </li><li>Stephan is Stephan Szabo <<a href="mailto:sszabo@megazone23.bigpanda.com">sszabo@megazone23.bigpanda.com</a>>
1094 </li><li>Tatsuo is Tatsuo Ishii <<a href="mailto:t-ishii@sra.co.jp">t-ishii@sra.co.jp</a>> of Software Research Assoc.
1095 </li><li>Teodor is Teodor Sigaev <<a href="mailto:teodor@sigaev.ru">teodor@sigaev.ru</a>>
1096 </li><li>Tom is Tom Lane <<a href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>> of Red Hat
1098 </li></ul></li></ul>