TODO 44.8 KB
Newer Older
Bruce Momjian's avatar
Bruce Momjian committed
1

Bruce Momjian's avatar
Bruce Momjian committed
2 3
PostgreSQL TODO List
====================
Bruce Momjian's avatar
Bruce Momjian committed
4
Current maintainer:	Bruce Momjian (pgman@candle.pha.pa.us)
Bruce Momjian's avatar
Bruce Momjian committed
5
Last updated:		Thu Jan  5 11:39:45 EST 2006
Marc G. Fournier's avatar
Marc G. Fournier committed
6

Bruce Momjian's avatar
Bruce Momjian committed
7 8
The most recent version of this document can be viewed at
http://www.postgresql.org/docs/faqs.TODO.html.
Bruce Momjian's avatar
Bruce Momjian committed
9

Bruce Momjian's avatar
Bruce Momjian committed
10
#A hyphen, "-", marks changes that will appear in the upcoming 8.2 release.#
Bruce Momjian's avatar
Bruce Momjian committed
11
#A percent sign, "%", marks items that are easier to implement.#
12 13 14 15

Bracketed items, "[]", have more detail.

This list contains all known PostgreSQL bugs and feature requests. If
Bruce Momjian's avatar
Bruce Momjian committed
16
you would like to work on an item, please read the Developer's FAQ
17
first.
18

Bruce Momjian's avatar
Bruce Momjian committed
19

20 21
Administration
==============
Bruce Momjian's avatar
Bruce Momjian committed
22

Bruce Momjian's avatar
Bruce Momjian committed
23
* %Remove behavior of postmaster -o
24
* %Allow pooled connections to list all prepared statements
25 26

  This would allow an application inheriting a pooled connection to know
27
  the statements prepared in the current session.
28

Bruce Momjian's avatar
Bruce Momjian committed
29 30
* Allow major upgrades without dump/reload, perhaps using pg_upgrade 
  [pg_upgrade]
Bruce Momjian's avatar
Bruce Momjian committed
31
* Check for unreferenced table files created by transactions that were
Bruce Momjian's avatar
Bruce Momjian committed
32
  in-progress when the server terminated abruptly
Bruce Momjian's avatar
Bruce Momjian committed
33 34
* Allow administrators to safely terminate individual sessions either
  via an SQL function or SIGTERM 
35

Bruce Momjian's avatar
Bruce Momjian committed
36 37 38 39
  Lock table corruption following SIGTERM of an individual backend
  has been reported in 8.0.  A possible cause was fixed in 8.1, but
  it is unknown whether other problems exist.  This item mostly
  requires additional testing rather than of writing any new code.
40

Bruce Momjian's avatar
Bruce Momjian committed
41
* %Set proper permissions on non-system schemas during db creation
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
42

Bruce Momjian's avatar
Bruce Momjian committed
43 44
  Currently all schemas are owned by the super-user because they are
  copied from the template1 database.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
45

Bruce Momjian's avatar
Bruce Momjian committed
46 47 48
* Support table partitioning that allows a single table to be stored
  in subtables that are partitioned based on the primary key or a WHERE
  clause
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
49
* Add function to report the time of the most recent server reload
Bruce Momjian's avatar
Bruce Momjian committed
50

Bruce Momjian's avatar
Bruce Momjian committed
51

Bruce Momjian's avatar
Bruce Momjian committed
52
* Improve replication solutions
53

Bruce Momjian's avatar
Bruce Momjian committed
54
	o Load balancing
Bruce Momjian's avatar
Bruce Momjian committed
55

Bruce Momjian's avatar
Bruce Momjian committed
56 57 58
	  You can use any of the master/slave replication servers to use a
	  standby server for data warehousing. To allow read/write queries to
	  multiple servers, you need multi-master replication like pgcluster.
Bruce Momjian's avatar
Bruce Momjian committed
59

Bruce Momjian's avatar
Bruce Momjian committed
60
	o Allow replication over unreliable or non-persistent links
61 62


Bruce Momjian's avatar
Bruce Momjian committed
63
* Configuration files
64

Bruce Momjian's avatar
Bruce Momjian committed
65 66
	o %Add "include file" functionality in postgresql.conf
	o %Allow commenting of variables in postgresql.conf to restore them
Bruce Momjian's avatar
Bruce Momjian committed
67
	  to defaults
Bruce Momjian's avatar
Bruce Momjian committed
68

Bruce Momjian's avatar
Bruce Momjian committed
69 70
	  Currently, if a variable is commented out, it keeps the
	  previous uncommented value until a server restarted.
Bruce Momjian's avatar
Bruce Momjian committed
71

Bruce Momjian's avatar
Bruce Momjian committed
72
	o %Allow pg_hba.conf settings to be controlled via SQL
Bruce Momjian's avatar
Bruce Momjian committed
73

Bruce Momjian's avatar
Bruce Momjian committed
74 75 76 77 78
	  This would add a function to load the SQL table from
          pg_hba.conf, and one to writes its contents to the flat file.
	  The table should have a line number that is a float so rows
	  can be inserted between existing rows, e.g. row 2.5 goes
	  between row 2 and row 3.
Bruce Momjian's avatar
Bruce Momjian committed
79

Bruce Momjian's avatar
Bruce Momjian committed
80
	o %Allow postgresql.conf file values to be changed via an SQL
Bruce Momjian's avatar
Bruce Momjian committed
81
	  API, perhaps using SET GLOBAL
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
82
	o Allow the server to be stopped/restarted via an SQL API
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
83 84 85
	o Issue a warning if a change-on-restart-only postgresql.conf value
	  is modified  and the server config files are reloaded
	o Mark change-on-restart-only values in postgresql.conf
Bruce Momjian's avatar
Bruce Momjian committed
86

87

Bruce Momjian's avatar
Bruce Momjian committed
88
* Tablespaces
89

Bruce Momjian's avatar
Bruce Momjian committed
90 91 92
	* Allow a database in tablespace t1 with tables created in
	  tablespace t2 to be used as a template for a new database created
	  with default tablespace t2
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
93

Bruce Momjian's avatar
Bruce Momjian committed
94 95 96 97 98 99 100 101
	  All objects in the default database tablespace must have default
	  tablespace specifications. This is because new databases are
	  created by copying directories. If you mix default tablespace
	  tables and tablespace-specified tables in the same directory,
	  creating a new database from such a mixed directory would create a
	  new database with tables that had incorrect explicit tablespaces.
	  To fix this would require modifying pg_class in the newly copied
	  database, which we don't currently do.
102

Bruce Momjian's avatar
Bruce Momjian committed
103
	* Allow reporting of which objects are in which tablespaces
104

Bruce Momjian's avatar
Bruce Momjian committed
105 106 107 108 109
	  This item is difficult because a tablespace can contain objects
	  from multiple databases. There is a server-side function that
	  returns the databases which use a specific tablespace, so this
	  requires a tool that will call that function and connect to each
	  database to find the objects in each database for that tablespace.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
110

Bruce Momjian's avatar
Bruce Momjian committed
111
	o %Add a GUC variable to control the tablespace for temporary objects
Bruce Momjian's avatar
Bruce Momjian committed
112
	  and sort files
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
113

Bruce Momjian's avatar
Bruce Momjian committed
114 115
	  It could start with a random tablespace from a supplied list and
	  cycle through the list.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
116

Bruce Momjian's avatar
Bruce Momjian committed
117 118
	o Allow WAL replay of CREATE TABLESPACE to work when the directory
	  structure on the recovery computer is different from the original
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
119

Bruce Momjian's avatar
Bruce Momjian committed
120
	o Allow per-tablespace quotas
121 122


123 124
* Point-In-Time Recovery (PITR)

Bruce Momjian's avatar
Bruce Momjian committed
125 126
	  o Allow point-in-time recovery to archive partially filled
	    write-ahead logs [pitr]
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
127

Bruce Momjian's avatar
Bruce Momjian committed
128 129 130 131 132 133 134 135 136
	    Currently only full WAL files are archived. This means that the
	    most recent transactions aren't available for recovery in case
	    of a disk failure. This could be triggered by a user command or
	    a timer.

	  o Automatically force archiving of partially-filled WAL files when
	    pg_stop_backup() is called or the server is stopped

	    Doing this will allow administrators to know more easily when
137
	    the archive contains all the files needed for point-in-time
Bruce Momjian's avatar
Bruce Momjian committed
138 139
	    recovery.

Bruce Momjian's avatar
Bruce Momjian committed
140
	  o %Create dump tool for write-ahead logs for use in determining
Bruce Momjian's avatar
Bruce Momjian committed
141
	    transaction id for point-in-time recovery
142
	  o Allow a warm standby system to also allow read-only statements
Bruce Momjian's avatar
Bruce Momjian committed
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
	    [pitr]

	    This is useful for checking PITR recovery.

	  o Allow the PITR process to be debugged and data examined


Monitoring
==========

* Allow server log information to be output as INSERT statements

  This would allow server log information to be easily loaded into
  a database for analysis.

Bruce Momjian's avatar
Bruce Momjian committed
158
* %Add ability to monitor the use of temporary sort files
Bruce Momjian's avatar
Bruce Momjian committed
159
* Allow server logs to be remotely read and removed using SQL commands
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
160
* Allow protocol-level BIND parameter values to be logged
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
161

Bruce Momjian's avatar
Bruce Momjian committed
162

163 164
Data Types
==========
165

Bruce Momjian's avatar
Bruce Momjian committed
166 167 168 169 170
* Improve the MONEY data type

  Change the MONEY data type to use DECIMAL internally, with special
  locale-aware output formatting.

171
* Change NUMERIC to enforce the maximum precision
Bruce Momjian's avatar
Bruce Momjian committed
172 173 174 175 176 177 178 179 180 181 182
* Add NUMERIC division operator that doesn't round?

  Currently NUMERIC _rounds_ the result to the specified precision.  
  This means division can return a result that multiplied by the 
  divisor is greater than the dividend, e.g. this returns a value > 10:

    SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;

  The positive modulus result returned by NUMERICs might be considered
  inaccurate, in one sense.

Bruce Momjian's avatar
Bruce Momjian committed
183
* %Disallow changing default expression of a SERIAL column?
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
184
* Fix data types where equality comparison isn't intuitive, e.g. box
Bruce Momjian's avatar
Bruce Momjian committed
185
* %Prevent INET cast to CIDR if the unmasked bits are not zero, or
Bruce Momjian's avatar
Bruce Momjian committed
186
  zero the bits
187
* %Prevent INET cast to CIDR from dropping netmask, SELECT '1.1.1.1'::inet::cidr
Bruce Momjian's avatar
Bruce Momjian committed
188
* Allow INET + INT4 to increment the host part of the address, or
Bruce Momjian's avatar
Bruce Momjian committed
189
  throw an error on overflow
Bruce Momjian's avatar
Bruce Momjian committed
190
* %Add 'tid != tid ' operator for use in corruption recovery
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
191 192
* Allow user-defined types to specify a type modifier at table creation
  time
Bruce Momjian's avatar
Bruce Momjian committed
193

Bruce Momjian's avatar
Bruce Momjian committed
194

195 196 197 198 199 200 201 202 203 204 205
* Dates and Times

	o Allow infinite dates just like infinite timestamps
	o Merge hardwired timezone names with the TZ database; allow either 
	  kind everywhere a TZ name is currently taken
	o Allow customization of the known set of TZ names (generalize the
	  present australian_timezones hack)
	o Allow TIMESTAMP WITH TIME ZONE to store the original timezone
	  information, either zone name or offset from UTC [timezone]

	  If the TIMESTAMP value is stored with a time zone name, interval 
Bruce Momjian's avatar
Bruce Momjian committed
206
	  computations should adjust based on the time zone rules.
207

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
208
	o Fix SELECT '0.01 years'::interval, '0.01 months'::interval
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
209
	o Fix SELECT INTERVAL '1' MONTH
Bruce Momjian's avatar
Bruce Momjian committed
210 211 212 213 214 215 216 217 218 219 220 221 222
	o Add a GUC variable to allow output of interval values in ISO8601 
	  format
	o Improve timestamptz subtraction to be DST-aware

	  Currently, subtracting one date from another that crosses a
	  daylight savings time adjustment can return '1 day 1 hour', but
	  adding that back to the first date returns a time one hour in
	  the future.  This is caused by the adjustment of '25 hours' to
	  '1 day 1 hour', and '1 day' is the same time the next day, even
	  if daylight savings adjustments are involved.

	o Fix interval display to support values exceeding 2^31 hours
	o Add overflow checking to timestamp and interval arithmetic	
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
	o Add ISO INTERVAL handling
		o Add support for day-time syntax, INTERVAL '1 2:03:04' DAY TO 
		  SECOND
		o Add support for year-month syntax, INTERVAL '50-6' YEAR TO MONTH
		o For syntax that isn't uniquely ISO or PG syntax, like '1:30' or
		  '1', treat as ISO if there is a range specification clause,
	          and as PG if there no clause is present, e.g. interpret 
			  '1:30' MINUTE TO SECOND as '1 minute 30 seconds', and 
			  interpret '1:30' as '1 hour, 30 minutes'
		o Interpret INTERVAL '1 year' MONTH as CAST (INTERVAL '1 year' AS
		  INTERVAL MONTH), and this should return '12 months'
		o Round or truncate values to the requested precision, e.g.
		  INTERVAL '11 months' AS YEAR should return one or zero
		o Support precision, CREATE TABLE foo (a INTERVAL MONTH(3))


* Arrays
Bruce Momjian's avatar
Bruce Momjian committed
240

Bruce Momjian's avatar
Done:  
Bruce Momjian committed
241
	o -Allow NULLs in arrays
Bruce Momjian's avatar
Bruce Momjian committed
242 243
	o Delay resolution of array expression's data type so assignment
	  coercion can be performed on empty array expressions
Bruce Momjian's avatar
Bruce Momjian committed
244

Bruce Momjian's avatar
Bruce Momjian committed
245

246 247
* Binary Data

Bruce Momjian's avatar
Bruce Momjian committed
248
	o Improve vacuum of large objects, like /contrib/vacuumlo?
Bruce Momjian's avatar
Bruce Momjian committed
249
	o Add security checking for large objects
Bruce Momjian's avatar
Bruce Momjian committed
250
	o Auto-delete large objects when referencing row is deleted
251

Bruce Momjian's avatar
Bruce Momjian committed
252
          /contrib/lo offers this functionality.
253 254 255

	o Allow read/write into TOAST values like large objects

256
	  This requires the TOAST column to be stored EXTERNAL.
257

258

Bruce Momjian's avatar
Bruce Momjian committed
259 260 261 262 263 264 265 266 267 268 269 270
Functions
=========

* Allow INET subnet tests using non-constants to be indexed
* Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
  functionality

  Current CURRENT_TIMESTAMP returns the start time of the current
  transaction, and gettimeofday() returns the wallclock time. This will
  make time reporting more consistent and will allow reporting of
  the statement start time.

Bruce Momjian's avatar
Bruce Momjian committed
271 272
* %Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
  pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
Bruce Momjian's avatar
Bruce Momjian committed
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
* Allow to_char() to print localized month names
* Allow functions to have a schema search path specified at creation time
* Allow substring/replace() to get/set bit values
* Allow to_char() on interval values to accumulate the highest unit
  requested

  Some special format flag would be required to request such
  accumulation.  Such functionality could also be added to EXTRACT. 
  Prevent accumulation that crosses the month/day boundary because of
  the uneven number of days in a month.

	o to_char(INTERVAL '1 hour 5 minutes', 'MI') => 65
	o to_char(INTERVAL '43 hours 20 minutes', 'MI' ) => 2600 
	o to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
	o to_char(INTERVAL '3 years 5 months','MM') => 41

Bruce Momjian's avatar
Bruce Momjian committed
289
* Add sleep() function, remove from regress.c
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
290 291
* Allow user-defined functions retuning a domain value to enforce domain
  constraints
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
292
* Add SPI_gettypmod() to return the typemod for a TupleDesc
Bruce Momjian's avatar
Bruce Momjian committed
293

Bruce Momjian's avatar
Bruce Momjian committed
294

295 296
Multi-Language Support
======================
Bruce Momjian's avatar
Bruce Momjian committed
297

298
* Add NCHAR (as distinguished from ordinary varchar),
Bruce Momjian's avatar
Bruce Momjian committed
299
* Allow locale to be set at database creation
300

Bruce Momjian's avatar
Bruce Momjian committed
301 302 303 304
  Currently locale can only be set during initdb.  No global tables have
  locale-aware columns.  However, the database template used during
  database creation might have locale-aware indexes.  The indexes would
  need to be reindexed to match the new locale.
305 306 307 308 309

* Allow encoding on a per-column basis

  Right now only one encoding is allowed per database.

Bruce Momjian's avatar
Bruce Momjian committed
310
* Support multiple simultaneous character sets, per SQL92
Bruce Momjian's avatar
Bruce Momjian committed
311
* Improve UTF8 combined character handling?
312 313
* Add octet_length_server() and octet_length_client()
* Make octet_length_client() the same as octet_length()?
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
314
* Fix problems with wrong runtime encoding conversion for NLS message files
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
315

316

317 318
Views / Rules
=============
Bruce Momjian's avatar
Bruce Momjian committed
319

Bruce Momjian's avatar
Bruce Momjian committed
320
* %Automatically create rules on views so they are updateable, per SQL99
Bruce Momjian's avatar
Bruce Momjian committed
321 322 323 324

  We can only auto-create rules for simple views.  For more complex
  cases users will still have to write rules.

Bruce Momjian's avatar
Bruce Momjian committed
325
* Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
Bruce Momjian's avatar
Bruce Momjian committed
326
* Allow NOTIFY in rules involving conditionals
Bruce Momjian's avatar
Bruce Momjian committed
327
* Allow VIEW/RULE recompilation when the underlying tables change
328

Bruce Momjian's avatar
Bruce Momjian committed
329 330 331 332
  Another issue is whether underlying table changes should be reflected
  in the view, e.g. should SELECT * show additional columns if they
  are added after the view is created.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
333

334 335
SQL Commands
============
Bruce Momjian's avatar
Bruce Momjian committed
336

Bruce Momjian's avatar
Bruce Momjian committed
337
* Change LIMIT/OFFSET and FETCH/MOVE to use int8
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
338
* Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
339
* Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
Bruce Momjian's avatar
Bruce Momjian committed
340 341
* %Allow SET CONSTRAINTS to be qualified by schema/table name
* %Allow TRUNCATE ... CASCADE/RESTRICT
Bruce Momjian's avatar
Bruce Momjian committed
342 343 344

  This is like DELETE CASCADE, but truncates.

Bruce Momjian's avatar
Bruce Momjian committed
345
* %Add a separate TRUNCATE permission
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
346 347 348 349

  Currently only the owner can TRUNCATE a table because triggers are not
  called, and the table is locked in exclusive mode.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
350
* Allow PREPARE of cursors
Bruce Momjian's avatar
Bruce Momjian committed
351
* Allow PREPARE to automatically determine parameter types based on the SQL
352
  statement
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
353 354 355 356 357 358 359 360
* Allow finer control over the caching of prepared query plans

  Currently, queries prepared via the libpq API are planned on first
  execute using the supplied parameters --- allow SQL PREPARE to do the
  same.  Also, allow control over replanning prepared queries either
  manually or automatically when statistics for execute parameters
  differ dramatically from those used during planning.

361 362 363 364 365
* Allow LISTEN/NOTIFY to store info in memory rather than tables?

  Currently LISTEN/NOTIFY information is stored in pg_listener. Storing
  such information in memory would improve performance.

Bruce Momjian's avatar
Bruce Momjian committed
366
* Add optional textual message to NOTIFY
367 368 369 370 371

  This would allow an informational message to be added to the notify
  message, perhaps indicating the row modified or other custom
  information.

Bruce Momjian's avatar
Bruce Momjian committed
372
* Add a GUC variable to warn about non-standard SQL usage in queries
373
* Add SQL-standard MERGE command, typically used to merge two tables
Bruce Momjian's avatar
Bruce Momjian committed
374
  [merge]
375 376 377 378 379 380

  This is similar to UPDATE, then for unmatched rows, INSERT.
  Whether concurrent access allows modifications which could cause
  row loss is implementation independent.

* Add REPLACE or UPSERT command that does UPDATE, or on failure, INSERT
Bruce Momjian's avatar
Bruce Momjian committed
381
  [merge]
Bruce Momjian's avatar
Bruce Momjian committed
382 383

  To implement this cleanly requires that the table have a unique index
384 385 386
  so duplicate checking can be easily performed.  It is possible to
  do it without a unique index if we require the user to LOCK the table
  before the MERGE.
Bruce Momjian's avatar
Bruce Momjian committed
387

Bruce Momjian's avatar
Bruce Momjian committed
388 389
* Add NOVICE output level for helpful messages like automatic sequence/index
  creation
Bruce Momjian's avatar
Bruce Momjian committed
390
* %Add COMMENT ON for all cluster global objects (roles, databases
Bruce Momjian's avatar
Bruce Momjian committed
391
  and tablespaces)
Bruce Momjian's avatar
Done:  
Bruce Momjian committed
392
* -Make row-wise comparisons work per SQL spec
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
393 394 395 396 397

  Right now, '(a, b) < (1, 2)' is processed as 'a < 1 and b < 2', but
  the SQL standard requires it to be processed as a column-by-column
  comparison, so the proper comparison is '(a < 1) OR (a = 1 AND b < 2)'.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
398
* Add RESET CONNECTION command to reset all session state
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
399

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
400
  This would include resetting of all variables (RESET ALL), dropping of
401 402 403 404
  temporary tables, removing any NOTIFYs, cursors, open transactions,
  prepared queries, currval()s, etc.  This could be used  for connection
  pooling.  We could also change RESET ALL to have this functionality.  
  The difficult of this features is allowing RESET ALL to not affect 
Bruce Momjian's avatar
Bruce Momjian committed
405 406 407
  changes made by the interface driver for its internal use.  One idea 
  is for this to be a protocol-only feature.  Another approach is to 
  notify the protocol when a RESET CONNECTION command is used.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
408

409
* Add GUC to issue notice about statements that use unjoined tables
410
* Allow EXPLAIN to identify tables that were skipped because of 
411
  constraint_exclusion
412
* Allow EXPLAIN output to be more easily processed by scripts
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
413
* Eventually enable escape_string_warning and standard_conforming_strings
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
414
* Simplify dropping roles that have objects in several databases
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
415
* Allow COMMENT ON to accept an expression rather than just a string
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
416 417
* Allow the count returned by SELECT, etc to be to represent as an int64
  to allow a higher range of values
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
418
* Make CLUSTER preserve recently-dead tuples per MVCC requirements
Bruce Momjian's avatar
Bruce Momjian committed
419

Bruce Momjian's avatar
Bruce Momjian committed
420

421 422 423 424 425 426
* CREATE

	o Allow CREATE TABLE AS to determine column lengths for complex
	  expressions like SELECT col1 || col2
	o Use more reliable method for CREATE DATABASE to get a consistent
	  copy of db?
Bruce Momjian's avatar
Bruce Momjian committed
427
	o Add ON COMMIT capability to CREATE TABLE AS ... SELECT
428 429 430 431 432 433 434 435 436 437 438 439 440


* UPDATE
	o Allow UPDATE to handle complex aggregates [update]?
	o Allow an alias to be provided for the target table in
	  UPDATE/DELETE

	  This is not SQL-spec but many DBMSs allow it.

	o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
	  columns


Bruce Momjian's avatar
Bruce Momjian committed
441
* ALTER
442

Bruce Momjian's avatar
Bruce Momjian committed
443
	o %Have ALTER TABLE RENAME rename SERIAL sequence names
Bruce Momjian's avatar
Bruce Momjian committed
444
	o Add ALTER DOMAIN to modify the underlying data type
Bruce Momjian's avatar
Bruce Momjian committed
445 446
	o %Allow ALTER TABLE ... ALTER CONSTRAINT ... RENAME
	o %Allow ALTER TABLE to change constraint deferrability and actions
Bruce Momjian's avatar
Bruce Momjian committed
447
	o Add missing object types for ALTER ... SET SCHEMA
448
	o Allow ALTER TABLESPACE to move to different directories
449
	o Allow databases to be moved to different tablespaces
450 451
	o Allow moving system tables to other tablespaces, where possible

452
	  Currently non-global system tables must be in the default database
453
	  tablespace. Global system tables can never be moved.
454

Bruce Momjian's avatar
Bruce Momjian committed
455 456 457
	o %Disallow dropping of an inherited constraint
	o %Prevent child tables from altering or dropping constraints 
          like CHECK that were inherited from the parent table
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
458
	o Have ALTER INDEX update the name of a constraint using that index
Bruce Momjian's avatar
Bruce Momjian committed
459
	o Add ALTER TABLE RENAME CONSTRAINT, update index name also
Bruce Momjian's avatar
Bruce Momjian committed
460

Bruce Momjian's avatar
Bruce Momjian committed
461

Bruce Momjian's avatar
Bruce Momjian committed
462
* CLUSTER
463

Bruce Momjian's avatar
Bruce Momjian committed
464
	o Automatically maintain clustering on a table
465

466 467
	  This might require some background daemon to maintain clustering
	  during periods of low usage. It might also require tables to be only
468
	  partially filled for easier reorganization.  Another idea would
469
          be to create a merged heap/index data file so an index lookup would
Bruce Momjian's avatar
Bruce Momjian committed
470
	  automatically access the heap data too.  A third idea would be to
Bruce Momjian's avatar
Bruce Momjian committed
471 472
	  store heap rows in hashed groups, perhaps using a user-supplied
	  hash function.
473

Bruce Momjian's avatar
Bruce Momjian committed
474
	o %Add default clustering to system tables
Bruce Momjian's avatar
Bruce Momjian committed
475

476 477
	  To do this, determine the ideal cluster index for each system
	  table and set the cluster setting during initdb.
478

Bruce Momjian's avatar
Bruce Momjian committed
479

Bruce Momjian's avatar
Bruce Momjian committed
480
* COPY
481

482
	o Allow COPY to report error lines and continue
483

484 485
	  This requires the use of a savepoint before each COPY line is
	  processed, with ROLLBACK on COPY failure.
486

Bruce Momjian's avatar
Bruce Momjian committed
487
	o %Have COPY return the number of rows loaded/unloaded?
Bruce Momjian's avatar
Bruce Momjian committed
488
	o Allow COPY on a newly-created table to skip WAL logging
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
489

Bruce Momjian's avatar
Bruce Momjian committed
490
	  On crash recovery, the table involved in the COPY would
Bruce Momjian's avatar
Bruce Momjian committed
491
	  be removed or have its heap and index files truncated.  One
Bruce Momjian's avatar
Bruce Momjian committed
492 493 494
	  issue is that no other backend should be able to add to 
	  the table at the same time, which is something that is 
	  currently allowed.
Bruce Momjian's avatar
Bruce Momjian committed
495

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
496 497
	o Allow COPY to output from views

498
	  Another idea would be to allow actual SELECT statements in a COPY.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
499

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
500

Bruce Momjian's avatar
Bruce Momjian committed
501
* GRANT/REVOKE
502

Bruce Momjian's avatar
Bruce Momjian committed
503
	o Allow column-level privileges
Bruce Momjian's avatar
Bruce Momjian committed
504
	o %Allow GRANT/REVOKE permissions to be applied to all schema objects
Bruce Momjian's avatar
Done:  
Bruce Momjian committed
505
	  with one command
Bruce Momjian's avatar
Bruce Momjian committed
506 507 508 509 510 511 512 513 514

	  The proposed syntax is:
		GRANT SELECT ON ALL TABLES IN public TO phpuser;
		GRANT SELECT ON NEW TABLES IN public TO phpuser;

	* Allow GRANT/REVOKE permissions to be inherited by objects based on
	  schema permissions


Bruce Momjian's avatar
Bruce Momjian committed
515
* CURSOR
516

517
	o Allow UPDATE/DELETE WHERE CURRENT OF cursor
518

519 520 521 522
	  This requires using the row ctid to map cursor rows back to the
	  original heap row. This become more complicated if WITH HOLD cursors
	  are to be supported because WITH HOLD cursors have a copy of the row
	  and no FOR UPDATE lock.
523 524

	o Prevent DROP TABLE from dropping a row referenced by its own open
Bruce Momjian's avatar
Bruce Momjian committed
525
	  cursor?
526

Bruce Momjian's avatar
Bruce Momjian committed
527
	o %Allow pooled connections to list all open WITH HOLD cursors
528

529 530
	  Because WITH HOLD cursors exist outside transactions, this allows
	  them to be listed so they can be closed.
Bruce Momjian's avatar
Bruce Momjian committed
531

Bruce Momjian's avatar
Bruce Momjian committed
532

Bruce Momjian's avatar
Bruce Momjian committed
533
* INSERT
534

535
	o Allow INSERT/UPDATE of the system-generated oid value for a row
Bruce Momjian's avatar
Bruce Momjian committed
536
	o Allow INSERT INTO tab (col1, ..) VALUES (val1, ..), (val2, ..)
537
	o Allow INSERT/UPDATE ... RETURNING new.col or old.col
538

539 540 541
	  This is useful for returning the auto-generated key for an INSERT.
	  One complication is how to handle rules that run as part of
	  the insert.
Bruce Momjian's avatar
Bruce Momjian committed
542

Bruce Momjian's avatar
Bruce Momjian committed
543

Bruce Momjian's avatar
Bruce Momjian committed
544
* SHOW/SET
545

Bruce Momjian's avatar
Bruce Momjian committed
546
	o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
Bruce Momjian's avatar
Bruce Momjian committed
547
	  ANALYZE, and CLUSTER
Bruce Momjian's avatar
Bruce Momjian committed
548
	o Add SET PATH for schemas?
549

550
	  This is basically the same as SET search_path.
Bruce Momjian's avatar
Bruce Momjian committed
551

Bruce Momjian's avatar
Bruce Momjian committed
552

553 554
* Server-Side Languages

Bruce Momjian's avatar
Bruce Momjian committed
555
	o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
Bruce Momjian's avatar
Bruce Momjian committed
556
	o Allow function parameters to be passed by name,
Bruce Momjian's avatar
Bruce Momjian committed
557
	  get_employee_salary(emp_id => 12345, tax_year => 2001)
558
	o Add Oracle-style packages
Bruce Momjian's avatar
Bruce Momjian committed
559
	o Add table function support to pltcl, plpython
Bruce Momjian's avatar
Bruce Momjian committed
560
	o Add capability to create and call PROCEDURES
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
561
	o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
562
	o Allow function argument names to be statements from PL/PgSQL
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
563
	o Add MOVE to PL/pgSQL
Bruce Momjian's avatar
Bruce Momjian committed
564 565 566 567
	o Add support for polymorphic arguments and return types to
	  languages other than PL/PgSQL
	o Add support for OUT and INOUT parameters to languages other 
	  than PL/PgSQL
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
568
	o Add single-step debugging of PL/PgSQL functions
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
569
	o Allow PL/PgSQL to support WITH HOLD cursors
Bruce Momjian's avatar
Bruce Momjian committed
570

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
571

572 573
Clients
=======
574

575
* -Have initdb set the input DateStyle (MDY or DMY) based on locale
576 577 578
* Have pg_ctl look at PGHOST in case it is a socket directory?
* Allow pg_ctl to work properly with configuration files located outside
  the PGDATA directory
579

580 581 582 583
  pg_ctl can not read the pid file because it isn't located in the
  config directory but in the PGDATA directory.  The solution is to
  allow pg_ctl to read and understand postgresql.conf to find the
  data_directory value.
584

585 586

* psql
Bruce Momjian's avatar
Bruce Momjian committed
587

588 589 590 591 592 593 594 595 596 597 598
	o Have psql show current values for a sequence
	o Move psql backslash database information into the backend, use
	  mnemonic commands? [psql]

	  This would allow non-psql clients to pull the same information out
	  of the database as psql.

	o Fix psql's display of schema information (Neil)
	o Allow psql \pset boolean variables to set to fixed values, rather
	  than toggle
	o Consistently display privilege information for all objects in psql
599
	o Improve psql's handling of multi-line statements
Bruce Momjian's avatar
Bruce Momjian committed
600

601 602
	  Currently, while \e saves a single statement as one entry, interactive
	  statements are saved one line at a time.  Ideally all statements
603
	  would be saved like \e does.
Bruce Momjian's avatar
Bruce Momjian committed
604 605 606 607 608 609 610

	o Allow multi-line column values to align in the proper columns

	  If the second output column value is 'a\nb', the 'b' should appear
	  in the second display column, rather than the first column as it
	  does now.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
611 612 613 614 615
	o Display IN, INOUT, and OUT parameters in \df+

	  It probably requires psql to output newlines in the proper
	  column, which is already on the TODO list.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
616 617 618 619 620
	o Add auto-expanded mode so expanded output is used if the row
	  length is wider than the screen width.  

	  Consider using auto-expanded mode for backslash commands like \df+.

Bruce Momjian's avatar
Bruce Momjian committed
621

Bruce Momjian's avatar
Bruce Momjian committed
622
* pg_dump
623

Bruce Momjian's avatar
Bruce Momjian committed
624 625
	o %Have pg_dump use multi-statement transactions for INSERT dumps
	o %Allow pg_dump to use multiple -t and -n switches [pg_dump]
Bruce Momjian's avatar
Bruce Momjian committed
626
	o %Add dumping of comments on index columns and composite type columns
Bruce Momjian's avatar
Bruce Momjian committed
627
	o %Add full object name to the tag field.  eg. for operators we need
Bruce Momjian's avatar
Bruce Momjian committed
628
	  '=(integer, integer)', instead of just '='.
Bruce Momjian's avatar
Bruce Momjian committed
629
	o Add pg_dumpall custom format dumps?
Bruce Momjian's avatar
Bruce Momjian committed
630
	o %Add CSV output format
Bruce Momjian's avatar
Bruce Momjian committed
631
	o Update pg_dump and psql to use the new COPY libpq API (Christopher)
Bruce Momjian's avatar
Bruce Momjian committed
632 633
	o Remove unnecessary function pointer abstractions in pg_dump source
	  code
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
634 635 636 637
	o Allow selection of individual object(s) of all types, not just
	  tables
	o In a selective dump, allow dumping of an object and all its 
	  dependencies
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
638
	o Add options like pg_restore -l and -L to pg_dump
Bruce Momjian's avatar
Bruce Momjian committed
639
	o Stop dumping CASCADE on DROP TYPE commands in clean mode
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
640 641
	o Allow pg_dump --clean to drop roles that own objects or have
	  privileges
Bruce Momjian's avatar
Bruce Momjian committed
642
	o Add -f to pg_dumpall
Bruce Momjian's avatar
Done:  
Bruce Momjian committed
643

Bruce Momjian's avatar
Bruce Momjian committed
644

645 646
* ecpg

Michael Meskes's avatar
Michael Meskes committed
647
	o Docs
648

649 650
	  Document differences between ecpg and the SQL standard and
	  information about the Informix-compatibility module.
651

Bruce Momjian's avatar
Bruce Momjian committed
652
	o Solve cardinality > 1 for input descriptors / variables?
Bruce Momjian's avatar
Bruce Momjian committed
653
	o Add a semantic check level, e.g. check if a table really exists
Bruce Momjian's avatar
Bruce Momjian committed
654
	o fix handling of DB attributes that are arrays
Bruce Momjian's avatar
Bruce Momjian committed
655
	o Use backend PREPARE/EXECUTE facility for ecpg where possible
Bruce Momjian's avatar
Bruce Momjian committed
656
	o Implement SQLDA
Michael Meskes's avatar
Michael Meskes committed
657
	o Fix nested C comments
Bruce Momjian's avatar
Bruce Momjian committed
658
	o %sqlwarn[6] should be 'W' if the PRECISION or SCALE value specified
659
	o Make SET CONNECTION thread-aware, non-standard?
660
	o Allow multidimensional arrays
Bruce Momjian's avatar
Bruce Momjian committed
661
	o Add internationalized message strings
Bruce Momjian's avatar
Bruce Momjian committed
662

Bruce Momjian's avatar
Bruce Momjian committed
663

664 665 666 667 668 669 670 671 672
libpq

	o Add a function to support Parse/DescribeStatement capability
	o Add PQescapeIdentifier()
	o Prevent PQfnumber() from lowercasing unquoted the column name

	  PQfnumber() should never have been doing lowercasing, but 
	  historically it has so we need a way to prevent it

673
	o Allow statement results to be automatically batched to the client
674

675
	  Currently, all statement results are transferred to the libpq
676 677
	  client before libpq makes the results available to the 
	  application.  This feature would allow the application to make
678
	  use of the first result rows while the rest are transferred, or
679
	  held on the server waiting for them to be requested by libpq.
680
	  One complexity is that a statement like SELECT 1/col could error
681 682 683
	  out mid-way through the result set.


684 685
Referential Integrity
=====================
Bruce Momjian's avatar
Bruce Momjian committed
686

Bruce Momjian's avatar
Bruce Momjian committed
687
* Add MATCH PARTIAL referential integrity
688 689 690 691 692 693
* Add deferred trigger queue file

  Right now all deferred trigger information is stored in backend
  memory.  This could exhaust memory for very large trigger queues.
  This item involves dumping large queues into files.

Bruce Momjian's avatar
Bruce Momjian committed
694
* Change foreign key constraint for array -> element to mean element
Bruce Momjian's avatar
Bruce Momjian committed
695 696
  in array?
* Allow DEFERRABLE UNIQUE constraints?
697
* Allow triggers to be disabled in only the current session.
698

699 700 701
  This is currently possible by starting a multi-statement transaction,
  modifying the system tables, performing the desired SQL, restoring the
  system tables, and committing the transaction.  ALTER TABLE ...
Bruce Momjian's avatar
Bruce Momjian committed
702
  TRIGGER requires a table lock so it is not ideal for this usage.
703

704
* With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN KEY
705 706 707 708

  If the dump is known to be valid, allow foreign keys to be added
  without revalidating the data.

Bruce Momjian's avatar
Bruce Momjian committed
709
* Allow statement-level triggers to access modified rows
Bruce Momjian's avatar
Bruce Momjian committed
710
* Support triggers on columns (Greg Sabino Mullane)
711
* Enforce referential integrity for system tables
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
712
* Allow AFTER triggers on system tables
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
713

714 715 716 717 718
  System tables are modified in many places in the backend without going
  through the executor and therefore not causing triggers to fire. To
  complete this item, the functions that modify system tables will have
  to fire triggers.

Bruce Momjian's avatar
Bruce Momjian committed
719

720 721
Dependency Checking
===================
Bruce Momjian's avatar
Bruce Momjian committed
722

Bruce Momjian's avatar
Bruce Momjian committed
723 724 725
* Flush cached query plans when the dependent objects change,
  when the cardinality of parameters changes dramatically, or
  when new ANALYZE statistics are available
Bruce Momjian's avatar
Bruce Momjian committed
726 727 728 729

  A more complex solution would be to save multiple plans for different
  cardinality and use the appropriate plan based on the EXECUTE values.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
730
* Track dependencies in function bodies and recompile/invalidate
731

Bruce Momjian's avatar
Bruce Momjian committed
732 733
  This is particularly important for references to temporary tables
  in PL/PgSQL because PL/PgSQL caches query plans.  The only workaround
734 735 736
  in PL/PgSQL is to use EXECUTE.  One complexity is that a function
  might itself drop and recreate dependent tables, causing it to
  invalidate its own query plan.
Bruce Momjian's avatar
Bruce Momjian committed
737

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
738

739 740
Exotic Features
===============
741

742 743
* Add SQL99 WITH clause to SELECT
* Add SQL99 WITH RECURSIVE to SELECT
Bruce Momjian's avatar
Bruce Momjian committed
744
* Add pre-parsing phase that converts non-ISO syntax to supported
745 746 747 748 749
  syntax

  This could allow SQL written for other databases to run without
  modification.

Bruce Momjian's avatar
Bruce Momjian committed
750
* Allow plug-in modules to emulate features from other databases
Bruce Momjian's avatar
Bruce Momjian committed
751
* SQL*Net listener that makes PostgreSQL appear as an Oracle database
Bruce Momjian's avatar
Bruce Momjian committed
752
  to clients
753
* Allow statements across databases or servers with transaction
754
  semantics
755

756
  This can be done using dblink and two-phase commit.
757

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
758
* Add the features of packages
759

760
	o  Make private objects accessible only to objects in the same schema
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
761 762 763
	o  Allow current_schema.objname to access current schema objects
	o  Add session variables
	o  Allow nested schemas
764

Bruce Momjian's avatar
Bruce Momjian committed
765

Bruce Momjian's avatar
Bruce Momjian committed
766 767 768 769 770 771 772 773 774 775 776 777
Indexes
=======

* Allow inherited tables to inherit index, UNIQUE constraint, and primary
  key, foreign key
* UNIQUE INDEX on base column not honored on INSERTs/UPDATEs from
  inherited table:  INSERT INTO inherit_table (unique_index_col) VALUES
  (dup) should fail

  The main difficulty with this item is the problem of creating an index
  that can span more than one table.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
778
* Allow SELECT ... FOR UPDATE on inherited tables
Bruce Momjian's avatar
Bruce Momjian committed
779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799
* Add UNIQUE capability to non-btree indexes
* Prevent index uniqueness checks when UPDATE does not modify the column

  Uniqueness (index) checks are done when updating a column even if the
  column is not modified by the UPDATE.

* Allow the creation of on-disk bitmap indexes which can be quickly
  combined with other bitmap indexes

  Such indexes could be more compact if there are only a few distinct values.
  Such indexes can also be compressed.  Keeping such indexes updated can be
  costly.

* Allow use of indexes to search for NULLs

  One solution is to create a partial index on an IS NULL expression.

* Allow accurate statistics to be collected on indexes with more than
  one column or expression indexes, perhaps using per-index statistics
* Add fillfactor to control reserved free space during index creation
* Allow the creation of indexes with mixed ascending/descending specifiers
800
* Allow constraint_exclusion to work for UNIONs like it does for
801 802
  inheritance, allow it to work for UPDATE and DELETE statements, and allow
  it to be used for all statements with little performance impact
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
803 804
* Allow CREATE INDEX to take an additional parameter for use with
  special index types
Bruce Momjian's avatar
Bruce Momjian committed
805
* Consider compressing indexes by storing key values duplicated in
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
806
  several rows as a single index entry
Bruce Momjian's avatar
Bruce Momjian committed
807

Bruce Momjian's avatar
Bruce Momjian committed
808 809
  This is difficult because it requires datatype-specific knowledge.

Bruce Momjian's avatar
Bruce Momjian committed
810

811 812 813 814 815
* GIST

	o Add more GIST index support for geometric data types
	o Allow GIST indexes to create certain complex index types, like
	  digital trees (see Aoki)
Bruce Momjian's avatar
Bruce Momjian committed
816 817

* Hash
818

Bruce Momjian's avatar
Bruce Momjian committed
819 820
	o Pack hash index buckets onto disk pages more efficiently

Bruce Momjian's avatar
Bruce Momjian committed
821
	  Currently only one hash bucket can be stored on a page. Ideally
Bruce Momjian's avatar
Bruce Momjian committed
822 823 824 825 826 827 828 829
	  several hash buckets could be stored on a single page and greater
	  granularity used for the hash algorithm.

	o Consider sorting hash buckets so entries can be found using a
	  binary search, rather than a linear scan

	o In hash indexes, consider storing the hash value with or instead
	  of the key itself
830

Bruce Momjian's avatar
Bruce Momjian committed
831 832 833
	o Add WAL logging for crash recovery
	o Allow multi-column hash indexes

834

835 836
Fsync
=====
837

838
* Improve commit_delay handling to reduce fsync()
Bruce Momjian's avatar
Bruce Momjian committed
839
* Determine optimal fdatasync/fsync, O_SYNC/O_DSYNC options
Bruce Momjian's avatar
Bruce Momjian committed
840 841

  Ideally this requires a separate test program that can be run
Bruce Momjian's avatar
Bruce Momjian committed
842 843
  at initdb time or optionally later.  Consider O_SYNC when
  O_DIRECT exists.
Bruce Momjian's avatar
Bruce Momjian committed
844

Bruce Momjian's avatar
Bruce Momjian committed
845
* %Add an option to sync() before fsync()'ing checkpoint files
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
846
* Add program to test if fsync has a delay compared to non-fsync
Bruce Momjian's avatar
Bruce Momjian committed
847

848

Bruce Momjian's avatar
Bruce Momjian committed
849 850
Cache Usage
===========
Bruce Momjian's avatar
Bruce Momjian committed
851

Bruce Momjian's avatar
Bruce Momjian committed
852
* Allow free-behind capability for large sequential scans, perhaps using
Bruce Momjian's avatar
Bruce Momjian committed
853 854
  posix_fadvise()

Bruce Momjian's avatar
Bruce Momjian committed
855
  Posix_fadvise() can control both sequential/random file caching and
Bruce Momjian's avatar
Bruce Momjian committed
856 857 858 859
  free-behind behavior, but it is unclear how the setting affects other
  backends that also have the file open, and the feature is not supported
  on all operating systems.

860 861 862 863
* Speed up COUNT(*)

  We could use a fixed row count and a +/- count to follow MVCC
  visibility rules, or a single cached value could be used and
864 865 866 867 868
  invalidated if anyone modifies the table.  Another idea is to
  get a count directly from a unique index, but for this to be
  faster than a sequential scan it must avoid access to the heap
  to obtain tuple visibility information.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
869 870
* Add estimated_count(*) to return an estimate of COUNT(*)

871
  This would use the planner ANALYZE statistics to return an estimated
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
872 873
  count.

874 875
* Allow data to be pulled directly from indexes

Bruce Momjian's avatar
Bruce Momjian committed
876 877
  Currently indexes do not have enough tuple visibility information 
  to allow data to be pulled from the index without also accessing 
878
  the heap.  One way to allow this is to set a bit on index tuples 
Bruce Momjian's avatar
Bruce Momjian committed
879 880
  to indicate if a tuple is currently visible to all transactions 
  when the first valid heap lookup happens.  This bit would have to 
881 882 883 884 885 886 887 888
  be cleared when a heap tuple is expired.  

  Another idea is to maintain a bitmap of heap pages where all rows
  are visible to all backends, and allow index lookups to reference 
  that bitmap to avoid heap lookups, perhaps the same bitmap we might
  add someday to determine which heap pages need vacuuming.  Frequently
  accessed bitmaps would have to be stored in shared memory.  One 8k
  page of bitmaps could track 512MB of heap pages.
Bruce Momjian's avatar
Bruce Momjian committed
889

890
* Consider automatic caching of statements at various levels:
Bruce Momjian's avatar
Bruce Momjian committed
891

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
892 893 894 895
	o Parsed query tree
	o Query execute plan
	o Query results

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
896
* Allow sequential scans to take advantage of other concurrent
897
  sequential scans, also called "Synchronised Scanning"
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
898

899 900 901 902
  One possible implementation is to start sequential scans from the lowest
  numbered buffer in the shared cache, and when reaching the end wrap
  around to the beginning, rather than always starting sequential scans
  at the start of the table.
Bruce Momjian's avatar
Bruce Momjian committed
903

Bruce Momjian's avatar
Bruce Momjian committed
904

905 906
Vacuum
======
Bruce Momjian's avatar
Bruce Momjian committed
907

908 909
* Improve speed with indexes

910
  For large table adjustments during VACUUM FULL, it is faster to 
Bruce Momjian's avatar
Bruce Momjian committed
911
  reindex rather than update the index.
912

Bruce Momjian's avatar
Bruce Momjian committed
913 914
* Reduce lock time during VACUUM FULL by moving tuples with read lock,
  then write lock and truncate table
915 916 917 918 919

  Moved tuples are invisible to other backends so they don't require a
  write lock. However, the read lock promotion to write lock could lead
  to deadlock situations.

Bruce Momjian's avatar
Bruce Momjian committed
920 921
* Auto-fill the free space map by scanning the buffer cache or by
  checking pages written by the background writer
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
922 923 924 925 926 927
* Create a bitmap of pages that need vacuuming

  Instead of sequentially scanning the entire table, have the background
  writer or some other process record pages that have expired rows, then
  VACUUM can look at just those pages rather than the entire table.  In
  the event of a system crash, the bitmap would probably be invalidated.
Bruce Momjian's avatar
Bruce Momjian committed
928 929 930 931
  One complexity is that index entries still have to be vacuumed, and
  doing this without an index scan (by using the heap values to find the
  index entry) might be slow and unreliable, especially for user-defined
  index functions.
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
932

Bruce Momjian's avatar
Bruce Momjian committed
933
* %Add system view to show free space map contents
Bruce Momjian's avatar
Bruce Momjian committed
934 935


Bruce Momjian's avatar
Add:  
Bruce Momjian committed
936
* Auto-vacuum
937

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
938
	o Use free-space map information to guide refilling
Bruce Momjian's avatar
Bruce Momjian committed
939 940
	o %Issue log message to suggest VACUUM FULL if a table is nearly
	  empty?
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
941 942
	o Improve xid wraparound detection by recording per-table rather
	  than per-database
Bruce Momjian's avatar
Bruce Momjian committed
943
	o Consider logging activity either to the logs or a system view
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
944

Bruce Momjian's avatar
Bruce Momjian committed
945

946 947
Locking
=======
948

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
949
* Fix priority ordering of read and write light-weight locks (Neil)
950

951

Bruce Momjian's avatar
Bruce Momjian committed
952 953
Startup Time Improvements
=========================
954

Bruce Momjian's avatar
Bruce Momjian committed
955
* Experiment with multi-threaded backend [thread]
956 957 958

  This would prevent the overhead associated with process creation. Most
  operating systems have trivial process creation time compared to
959
  database startup overhead, but a few operating systems (Win32,
Bruce Momjian's avatar
Bruce Momjian committed
960
  Solaris) might benefit from threading.  Also explore the idea of
961
  a single session using multiple threads to execute a statement faster.
962

Bruce Momjian's avatar
Bruce Momjian committed
963
* Add connection pooling
964 965 966 967

  It is unclear if this should be done inside the backend code or done
  by something external like pgpool. The passing of file descriptors to
  existing backends is one of the difficulties with a backend approach.
968

969

970 971
Write-Ahead Log
===============
972

973 974
* Eliminate need to write full pages to WAL before page modification [wal]

Bruce Momjian's avatar
Bruce Momjian committed
975
  Currently, to protect against partial disk page writes, we write
976
  full page images to WAL before they are modified so we can correct any
977 978
  partial page writes during recovery.  These pages can also be
  eliminated from point-in-time archive files.
979

Bruce Momjian's avatar
Bruce Momjian committed
980 981
	o  When off, write CRC to WAL and check file system blocks
	   on recovery
Bruce Momjian's avatar
Bruce Momjian committed
982 983 984 985

	   If CRC check fails during recovery, remember the page in case
	   a later CRC for that page properly matches.

Bruce Momjian's avatar
Bruce Momjian committed
986 987 988 989
	o  Write full pages during file system write and not when
	   the page is modified in the buffer cache

	   This allows most full page writes to happen in the background
Bruce Momjian's avatar
Bruce Momjian committed
990 991 992
	   writer.  It might cause problems for applying WAL on recovery
	   into a partially-written page, but later the full page will be
	   replaced from WAL.
Bruce Momjian's avatar
Bruce Momjian committed
993

Bruce Momjian's avatar
Bruce Momjian committed
994
* Allow WAL traffic to be streamed to another server for stand-by
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
995
  replication
996
* Reduce WAL traffic so only modified values are written rather than
Bruce Momjian's avatar
Bruce Momjian committed
997
  entire rows?
998 999
* Allow the pg_xlog directory location to be specified during initdb
  with a symlink back to the /data location
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1000
* Allow WAL information to recover corrupted pg_controldata
Bruce Momjian's avatar
Bruce Momjian committed
1001 1002
* Find a way to reduce rotational delay when repeatedly writing
  last WAL page
Bruce Momjian's avatar
Bruce Momjian committed
1003

1004 1005 1006
  Currently fsync of WAL requires the disk platter to perform a full
  rotation to fsync again. One idea is to write the WAL to different
  offsets that might reduce the rotational delay.
1007

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1008 1009 1010 1011
* Allow buffered WAL writes and fsync

  Instead of guaranteeing recovery of all committed transactions, this
  would provide improved performance by delaying WAL writes and fsync
Bruce Momjian's avatar
Bruce Momjian committed
1012
  so an abrupt operating system restart might lose a few seconds of
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1013 1014 1015 1016
  committed transactions but still be consistent.  We could perhaps
  remove the 'fsync' parameter (which results in an an inconsistent
  database) in favor of this capability.

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1017 1018 1019 1020 1021
* Allow control over which tables are WAL-logged

  Allow tables to bypass WAL writes and just fsync() dirty pages on
  commit.  To do this, only a single writer can modify the table, and
  writes must happen only on new pages.  Readers can continue accessing
Bruce Momjian's avatar
Bruce Momjian committed
1022 1023 1024 1025
  the table.  This would affect COPY, and perhaps INSERT/UPDATE too.
  Another option is to avoid transaction logging entirely and truncate 
  or drop the table on crash recovery.  These should be implemented 
  using ALTER TABLE, e.g. ALTER TABLE PERSISTENCE [ DROP | TRUNCATE | 
Bruce Momjian's avatar
Bruce Momjian committed
1026 1027
  STABLE | DEFAULT ].  Tables using non-default logging should not use
  referential integrity with default-logging tables, and tables using
Bruce Momjian's avatar
Bruce Momjian committed
1028
  stable logging probably can not have indexes.  [walcontrol]
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1029

Bruce Momjian's avatar
Bruce Momjian committed
1030

1031 1032
Optimizer / Executor
====================
1033

Bruce Momjian's avatar
Bruce Momjian committed
1034
* Improve selectivity functions for geometric operators
Bruce Momjian's avatar
Bruce Momjian committed
1035
* Allow ORDER BY ... LIMIT # to select high/low value without sort or
1036 1037
  index using a sequential scan for highest/lowest values

Bruce Momjian's avatar
Bruce Momjian committed
1038
  Right now, if no index exists, ORDER BY ... LIMIT # requires we sort
1039 1040
  all values to return the high/low value.  Instead The idea is to do a 
  sequential scan to find the high/low value, thus avoiding the sort.
Bruce Momjian's avatar
Bruce Momjian committed
1041
  MIN/MAX already does this, but not for LIMIT > 1.
1042

1043
* Precompile SQL functions to avoid overhead
Bruce Momjian's avatar
Bruce Momjian committed
1044
* Create utility to compute accurate random_page_cost value
Bruce Momjian's avatar
Bruce Momjian committed
1045
* Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1046
* Have EXPLAIN ANALYZE highlight poor optimizer estimates
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1047 1048
* Consider using hash buckets to do DISTINCT, rather than sorting

Bruce Momjian's avatar
Bruce Momjian committed
1049 1050
  This would be beneficial when there are few distinct values.  This is
  already used by GROUP BY.
Bruce Momjian's avatar
Bruce Momjian committed
1051

1052
* Log statements where the optimizer row estimates were dramatically
Bruce Momjian's avatar
Bruce Momjian committed
1053
  different from the number of rows actually found?
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1054

Bruce Momjian's avatar
Bruce Momjian committed
1055

Bruce Momjian's avatar
Bruce Momjian committed
1056 1057
Miscellaneous Performance
=========================
1058 1059

* Do async I/O for faster random read-ahead of data
1060 1061 1062

  Async I/O allows multiple I/O requests to be sent to the disk with
  results coming back asynchronously.
Bruce Momjian's avatar
Bruce Momjian committed
1063

Bruce Momjian's avatar
Bruce Momjian committed
1064
* Use mmap() rather than SYSV shared memory or to write WAL files?
1065 1066

  This would remove the requirement for SYSV SHM but would introduce
Bruce Momjian's avatar
Bruce Momjian committed
1067 1068
  portability issues. Anonymous mmap (or mmap to /dev/zero) is required
  to prevent I/O overhead.
Bruce Momjian's avatar
Bruce Momjian committed
1069 1070 1071

* Consider mmap()'ing files into a backend?

Bruce Momjian's avatar
Bruce Momjian committed
1072 1073 1074
  Doing I/O to large tables would consume a lot of address space or
  require frequent mapping/unmapping.  Extending the file also causes
  mapping problems that might require mapping only individual pages,
Bruce Momjian's avatar
Bruce Momjian committed
1075
  leading to thousands of mappings.  Another problem is that there is no
Bruce Momjian's avatar
Bruce Momjian committed
1076
  way to _prevent_ I/O to disk from the dirty shared buffers so changes
Bruce Momjian's avatar
Bruce Momjian committed
1077
  could hit disk before WAL is written.
1078

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1079
* Add a script to ask system configuration questions and tune postgresql.conf
Bruce Momjian's avatar
Bruce Momjian committed
1080 1081 1082 1083 1084 1085 1086 1087 1088
* Merge xmin/xmax/cmin/cmax back into three header fields

  Before subtransactions, there used to be only three fields needed to
  store these four values. This was possible because only the current
  transaction looks at the cmin/cmax values. If the current transaction
  created and expired the row the fields stored where xmin (same as
  xmax), cmin, cmax, and if the transaction was expiring a row from a
  another transaction, the fields stored were xmin (cmin was not
  needed), xmax, and cmax. Such a system worked because a transaction
Bruce Momjian's avatar
Bruce Momjian committed
1089
  could only see rows from another completed transaction. However,
Bruce Momjian's avatar
Bruce Momjian committed
1090 1091 1092 1093 1094
  subtransactions can see rows from outer transactions, and once the
  subtransaction completes, the outer transaction continues, requiring
  the storage of all four fields. With subtransactions, an outer
  transaction can create a row, a subtransaction expire it, and when the
  subtransaction completes, the outer transaction still has to have
Bruce Momjian's avatar
Bruce Momjian committed
1095
  proper visibility of the row's cmin, for example, for cursors.
Bruce Momjian's avatar
Bruce Momjian committed
1096 1097

  One possible solution is to create a phantom cid which represents a
Bruce Momjian's avatar
Bruce Momjian committed
1098 1099
  cmin/cmax pair and is stored in local memory.  Another idea is to
  store both cmin and cmax only in local memory.
Bruce Momjian's avatar
Bruce Momjian committed
1100

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1101
* Research storing disk pages with no alignment/padding
1102

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1103

1104
Source Code
1105 1106
===========

Bruce Momjian's avatar
Bruce Momjian committed
1107
* Add use of 'const' for variables in source tree
Bruce Momjian's avatar
Bruce Momjian committed
1108
* Rename some /contrib modules from pg* to pg_*
Bruce Momjian's avatar
Bruce Momjian committed
1109
* Move some things from /contrib into main tree
Bruce Momjian's avatar
Bruce Momjian committed
1110
* Move some /contrib modules out to their own project sites
Bruce Momjian's avatar
Bruce Momjian committed
1111
* %Remove warnings created by -Wcast-align
Bruce Momjian's avatar
Bruce Momjian committed
1112
* Move platform-specific ps status display info from ps_status.c to ports
Bruce Momjian's avatar
Bruce Momjian committed
1113
* Add optional CRC checksum to heap and index pages
Bruce Momjian's avatar
Bruce Momjian committed
1114
* Improve documentation to build only interfaces (Marc)
1115
* Remove or relicense modules that are not under the BSD license, if possible
Bruce Momjian's avatar
Bruce Momjian committed
1116
* %Remove memory/file descriptor freeing before ereport(ERROR)
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1117
* Acquire lock on a relation before building a relcache entry for it
Bruce Momjian's avatar
Bruce Momjian committed
1118 1119
* %Promote debug_query_string into a server-side function current_query()
* %Allow the identifier length to be increased via a configure option
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1120
* Allow cross-compiling by generating the zic database on the target system
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1121
* Improve NLS maintenace of libpgport messages linked onto applications
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1122
* Allow ecpg to work with MSVC and BCC
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1123
* Add xpath_array() to /contrib/xml2 to return results as an array
Bruce Momjian's avatar
Bruce Momjian committed
1124 1125 1126 1127 1128
* Allow building in directories containing spaces

  This is probably not possible because 'gmake' and other compiler tools
  do not fully support quoting of paths with spaces.

Bruce Momjian's avatar
Done:  
Bruce Momjian committed
1129
* -Allow installing to directories containing spaces
Bruce Momjian's avatar
Bruce Momjian committed
1130 1131 1132 1133 1134

  This is possible if proper quoting is added to the makefiles for the
  install targets.  Because PostgreSQL supports relocatable installs, it
  is already possible to install into a directory that doesn't contain 
  spaces and then copy the install to a directory with spaces.
Bruce Momjian's avatar
Bruce Momjian committed
1135

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1136
* Fix sgmltools so PDFs can be generated with bookmarks
Bruce Momjian's avatar
Bruce Momjian committed
1137
* %Clean up compiler warnings (especially with gcc version 4)
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1138 1139
* Use UTF8 encoding for NLS messages so all server encodings can
  read them properly
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1140
* Update Bonjour to work with newer cross-platform SDK
Bruce Momjian's avatar
Done:  
Bruce Momjian committed
1141
* -Remove BeOS and QNX-specific code
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1142

Bruce Momjian's avatar
Bruce Momjian committed
1143

1144
* Win32
1145

1146 1147 1148 1149 1150 1151 1152
	o Remove configure.in check for link failure when cause is found
	o Remove readdir() errno patch when runtime/mingwex/dirent.c rev
	  1.4 is released
	o Remove psql newline patch when we find out why mingw outputs an
	  extra newline
	o Allow psql to use readline once non-US code pages work with
	  backslashes
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1153 1154
	o Re-enable timezone output on log_line_prefix '%t' when a
	  shorter timezone string is available
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1155
	o Fix problem with shared memory on the Win32 Terminal Server
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1156 1157
	o Improve signal handling,
	  http://archives.postgresql.org/pgsql-patches/2005-06/msg00027.php
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1158 1159 1160 1161
	o Add long file support for binary pg_dump output

	  While Win32 supports 64-bit files, the MinGW API does not,
	  meaning we have to build an fseeko replacement on top of the
Bruce Momjian's avatar
Bruce Momjian committed
1162 1163 1164
	  Win32 API, and we have to make sure MinGW handles it.  Another
	  option is to wait for the MinGW project to fix it, or use the
	  code from the LibGW32C project as a guide.
Bruce Momjian's avatar
Bruce Momjian committed
1165

Bruce Momjian's avatar
Bruce Momjian committed
1166

Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1167
* Wire Protocol Changes
1168

1169
	o Allow dynamic character set handling
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1170
	o Add decoded type, length, precision
1171
	o Use compression?
Bruce Momjian's avatar
Bruce Momjian committed
1172
	o Update clients to use data types, typmod, schema.table.column names
1173
	  of result sets using new statement protocol
Bruce Momjian's avatar
Add:  
Bruce Momjian committed
1174

1175

1176 1177 1178 1179 1180
---------------------------------------------------------------------------


Developers who have claimed items are:
--------------------------------------
Bruce Momjian's avatar
Bruce Momjian committed
1181
* Alvaro is Alvaro Herrera <alvherre@dcc.uchile.cl>
Bruce Momjian's avatar
Bruce Momjian committed
1182
* Andrew is Andrew Dunstan <andrew@dunslane.net>
Bruce Momjian's avatar
Bruce Momjian committed
1183 1184 1185 1186
* Bruce is Bruce Momjian <pgman@candle.pha.pa.us> of Software Research Assoc.
* Christopher is Christopher Kings-Lynne <chriskl@familyhealth.com.au> of
    Family Health Network
* D'Arcy is D'Arcy J.M. Cain <darcy@druid.net> of The Cain Gang Ltd.
Bruce Momjian's avatar
Bruce Momjian committed
1187
* Fabien is Fabien Coelho <coelho@cri.ensmp.fr>
Bruce Momjian's avatar
Bruce Momjian committed
1188 1189
* Gavin is Gavin Sherry <swm@linuxworld.com.au> of Alcove Systems Engineering
* Greg is Greg Sabino Mullane <greg@turnstep.com>
Bruce Momjian's avatar
Bruce Momjian committed
1190
* Jan is Jan Wieck <JanWieck@Yahoo.com> of Afilias, Inc.
Bruce Momjian's avatar
Bruce Momjian committed
1191
* Joe is Joe Conway <mail@joeconway.com>
1192
* Karel is Karel Zak <zakkr@zf.jcu.cz>
Bruce Momjian's avatar
Fix :  
Bruce Momjian committed
1193
* Magnus is Magnus Hagander <mha@sollentuna.net>
Bruce Momjian's avatar
Bruce Momjian committed
1194
* Marc is Marc Fournier <scrappy@hub.org> of PostgreSQL, Inc.
Bruce Momjian's avatar
Bruce Momjian committed
1195
* Matthew T. O'Connor <matthew@zeut.net>
Bruce Momjian's avatar
Bruce Momjian committed
1196
* Michael is Michael Meskes <meskes@postgresql.org> of Credativ
Bruce Momjian's avatar
Bruce Momjian committed
1197
* Neil is Neil Conway <neilc@samurai.com>
Bruce Momjian's avatar
Bruce Momjian committed
1198
* Oleg is Oleg Bartunov <oleg@sai.msu.su>
1199
* Peter is Peter Eisentraut <peter_e@gmx.net>
Bruce Momjian's avatar
Bruce Momjian committed
1200
* Philip is Philip Warner <pjw@rhyme.com.au> of Albatross Consulting Pty. Ltd.
Bruce Momjian's avatar
Bruce Momjian committed
1201
* Rod is Rod Taylor <pg@rbt.ca>
Bruce Momjian's avatar
Bruce Momjian committed
1202
* Simon is Simon Riggs <simon@2ndquadrant.com>
Bruce Momjian's avatar
Bruce Momjian committed
1203
* Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>
Bruce Momjian's avatar
Bruce Momjian committed
1204
* Tatsuo is Tatsuo Ishii <t-ishii@sra.co.jp> of Software Research Assoc.
Bruce Momjian's avatar
Bruce Momjian committed
1205
* Teodor is Teodor Sigaev <teodor@sigaev.ru>
Bruce Momjian's avatar
Bruce Momjian committed
1206
* Tom is Tom Lane <tgl@sss.pgh.pa.us> of Red Hat