PostgreSQL TODO List
====================
Current maintainer: Bruce Momjian (pgman@candle.pha.pa.us)
-Last updated: Fri Aug 26 14:52:30 EDT 2005
+Last updated: Fri Aug 26 14:59:39 EDT 2005
The most recent version of this document can be viewed at
http://www.postgresql.org/docs/faqs.TODO.html.
* %Remove behavior of postmaster -o after making postmaster/postgres
flags unique
-* -Allow limits on per-db/role connections
* %Allow pooled connections to list all prepared queries
This would allow an application inheriting a pooled connection to know
it is unknown whether other problems exist. This item mostly
requires additional testing rather than of writing any new code.
-* -Prevent dropping user that still owns objects, or auto-drop the objects
* %Set proper permissions on non-system schemas during db creation
Currently all schemas are owned by the super-user because they are
copied from the template1 database.
-* -Add the client IP address and port to pg_stat_activity
* 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
a database for analysis.
* %Add ability to monitor the use of temporary sort files
-* -Add session start time and last statement time to pg_stat_activity
-* -Add a function that returns the start time of the postmaster
* Allow server logs to be remotely read and removed using SQL commands
* Arrays
o Allow NULLs in arrays
- o -Allow MIN()/MAX() on arrays
o Delay resolution of array expression's data type so assignment
coercion can be performed on empty array expressions
- o -Modify array literal representation to handle array index lower bound
- of other than one
* Binary Data
Functions
=========
-* -Add function to return compressed length of TOAST data values
* Allow INET subnet tests using non-constants to be indexed
* Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
functionality
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
-* -Prevent to_char() on interval from returning meaningless values
-
- For example, to_char('1 month', 'mon') is meaningless. Basically,
- most date-related parameters to to_char() are meaningless for
- intervals because interval is not anchored to a date.
-
* Add sleep() function, remove from regress.c
* Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
* Allow NOTIFY in rules involving conditionals
-* -Have views on temporary tables exist in the temporary namespace
-* -Allow temporary views on non-temporary tables
* Allow VIEW/RULE recompilation when the underlying tables change
SQL Commands
============
-* -Add BETWEEN SYMMETRIC/ASYMMETRIC
* Change LIMIT/OFFSET and FETCH/MOVE to use int8
-* -Add E'' escape string marker so eventually ordinary strings can treat
- backslashes literally, for portability
-
-* -Allow additional tables to be specified in DELETE for joins
-
- UPDATE already allows this (UPDATE...FROM) but we need similar
- functionality in DELETE. It's been agreed that the keyword should
- be USING, to avoid anything as confusing as DELETE FROM a FROM b.
-
* Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
-* -Allow REINDEX to rebuild all database indexes
* Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
* %Allow SET CONSTRAINTS to be qualified by schema/table name
* %Allow TRUNCATE ... CASCADE/RESTRICT
creation
* %Add COMMENT ON for all cluster global objects (roles, databases
and tablespaces)
-* -Add an option to automatically use savepoints for each statement in a
- multi-statement transaction.
-
- When enabled, this would allow errors in multi-statement transactions
- to be automatically ignored.
-
* %Make row-wise comparisons work per SQL spec
* Add RESET CONNECTION command to reset all session state
o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
columns
- o -Allow FOR UPDATE queries to do NOWAIT locks
* ALTER
This requires the use of a savepoint before each COPY line is
processed, with ROLLBACK on COPY failure.
- o -Allow COPY to understand \x as a hex byte
o %Have COPY return the number of rows loaded/unloaded?
- o -Allow COPY to optionally include column headings in the first line
- o -Allow COPY FROM ... CSV to interpret newlines and carriage
- returns in data
* GRANT/REVOKE
* SHOW/SET
- o -Have SHOW ALL show descriptions for server-side variables
o Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
ANALYZE, and CLUSTER
o Add SET PATH for schemas?
* Server-Side Languages
- o -Allow PL/PgSQL's RAISE function to take expressions
-
- Currently only constants are supported.
-
- o -Change PL/PgSQL to use palloc() instead of malloc()
o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
o Allow function parameters to be passed by name,
get_employee_salary(emp_id => 12345, tax_year => 2001)
o Add Oracle-style packages
o Add table function support to pltcl, plpython
- o -Allow PL/pgSQL EXECUTE query_var INTO record_var;
o Add capability to create and call PROCEDURES
o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
o Allow function argument names to be queries from PL/PgSQL
o Add MOVE to PL/pgSQL
- o -Pass arrays natively instead of as text between plperl and postgres
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
o %Add dumping of comments on index columns
o %Replace crude DELETE FROM method of pg_dumpall --clean for
cleaning of roles with separate DROP commands
- o -Add dumping and restoring of LOB comments
o Stop dumping CASCADE on DROP TYPE commands in clean mode
o %Add full object name to the tag field. eg. for operators we need
'=(integer, integer)', instead of just '='.
memory. This could exhaust memory for very large trigger queues.
This item involves dumping large queues into files.
-* -Implement shared row locks and use them in RI triggers
* Change foreign key constraint for array -> element to mean element
in array?
* Allow DEFERRABLE UNIQUE constraints?
-* -Allow triggers to be disabled
* Allow triggers to be disabled in only the current session.
This is currently possible by starting a multi-statement transaction,
This can be done using dblink and two-phase commit.
-* -Add two-phase commit
-
-
* Add the features of packages
o Make private objects accessable only to objects in the same schema
that can span more than one table.
* Allow SELECT ... FOR UPDATE on inherited tables
-* -Prevent inherited tables from expanding temporary subtables of other
- sessions
* Add UNIQUE capability to non-btree indexes
-* -Use indexes for MIN() and MAX()
-
- MIN/MAX queries can already be rewritten as SELECT col FROM tab ORDER
- BY col {DESC} LIMIT 1. Completing this item involves doing this
- transformation automatically.
-
-* -Use index to restrict rows returned by multi-key index when used with
- non-consecutive keys to reduce heap accesses
-
- For an index on col1,col2,col3, and a WHERE clause of col1 = 5 and
- col3 = 9, spin though the index checking for col1 and col3 matches,
- rather than just col1; also called skip-scanning.
-
* 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.
-* -Fetch heap pages matching index entries in sequential order
-
- Rather than randomly accessing heap pages based on index entries, mark
- heap pages needing access in a bitmap and do the lookups in sequential
- order. Another method would be to sort heap ctids matching the index
- before accessing the heap rows.
-
-* -Allow non-bitmap indexes to be combined by creating bitmaps in memory
-
- This feature allows separate indexes to be ANDed or ORed together. This
- is particularly useful for data warehousing applications that need to
- query the database in an many permutations. This feature scans an index
- and creates an in-memory bitmap, and allows that bitmap to be combined
- with other bitmap created in a similar way. The bitmap can either index
- all TIDs, or be lossy, meaning it records just page numbers and each
- page tuple has to be checked for validity in a separate pass.
-
* Allow the creation of on-disk bitmap indexes which can be quickly
combined with other bitmap indexes
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
-* -Fix incorrect rtree results due to wrong assumptions about "over"
- operator semantics
* Allow constraint_exclusion to work for UNIONs like it does for
inheritance, and allow it to work for UPDATE and DELETE queries
* GIST
o Add more GIST index support for geometric data types
- o -Add concurrency to GIST
o Allow GIST indexes to create certain complex index types, like
digital trees (see Aoki)
Ideally this requires a separate test program that can be run
at initdb time or optionally later.
-* -Allow multiple blocks to be written to WAL with one write()
* %Add an option to sync() before fsync()'ing checkpoint files
* Add program to test if fsync has a delay compared to non-fsync
backends that also have the file open, and the feature is not supported
on all operating systems.
-* -Consider use of open/fcntl(O_DIRECT) to minimize OS caching,
- for WAL writes
-
- O_DIRECT doesn't have the same media write guarantees as fsync, so it
- is in addition to the fsync method, not in place of it.
-
-* -Cache last known per-tuple offsets to speed long tuple access
* Speed up COUNT(*)
We could use a fixed row count and a +/- count to follow MVCC
o Query execute plan
o Query results
-* -Allow the size of the buffer cache used by temporary objects to be
- specified as a GUC variable
-
- Larger local buffer cache sizes requires more efficient handling of
- local cache lookups.
-
-* -Improve the background writer
-
- Allow the background writer to more efficiently write dirty buffers
- from the end of the LRU cache and use a clock sweep algorithm to
- write other dirty buffers to reduced checkpoint I/O
-
* Allow sequential scans to take advantage of other concurrent
sequentiqal scans, also called "Synchronised Scanning"
write lock. However, the read lock promotion to write lock could lead
to deadlock situations.
-* -Add a warning when the free space map is too small
* Maintain a map of recently-expired rows
This allows vacuum to target specific pages for possible free space
* Auto-vacuum
- o -Move into the backend code
o Use free-space map information to guide refilling
o %Issue log message to suggest VACUUM FULL if a table is nearly
empty?
Locking
=======
-* -Make locking of shared data structures more fine-grained
-
- This requires that more locks be acquired but this would reduce lock
- contention, improving concurrency.
-
* Add code to detect an SMP machine and handle spinlocks accordingly
from distributted.net, http://www1.distributed.net/source,
in client/common/cpucheck.cpp
while on non-SMP machines, the backend should sleep so the process
holding the lock can complete and release it.
-* -Improve SMP performance on i386 machines
-
- i386-based SMP machines can generate excessive context switching
- caused by lock failure in high concurrency situations. This may be
- caused by CPU cache line invalidation inefficiencies.
-
* Research use of sched_yield() for spinlock acquisition failure
* Fix priority ordering of read and write light-weight locks (Neil)
partial page writes during recovery. These pages can also be
eliminated from point-in-time archive files.
- o -Add ability to turn off full page writes
o When off, write CRC to WAL and check file system blocks
on recovery
remove the 'fsync' parameter (which results in an an inconsistent
database) in favor of this capability.
-* -Eliminate WAL logging for CREATE TABLE AS when not doing WAL archiving
-* -Change WAL to use 32-bit CRC, for performance reasons
-
Optimizer / Executor
====================
* Create utility to compute accurate random_page_cost value
* Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
* Have EXPLAIN ANALYZE highlight poor optimizer estimates
-* -Use CHECK constraints to influence optimizer decisions
-
- CHECK constraints contain information about the distribution of values
- within the table. This is also useful for implementing subtables where
- a tables content is distributed across several subtables.
-
* Consider using hash buckets to do DISTINCT, rather than sorting
This would be beneficial when there are few distinct values.
-* -ANALYZE should record a pg_statistic entry for an all-NULL column
* Log queries where the optimizer row estimates were dramatically
different from the number of rows actually found?
* Allow constraint_elimination to be automatically performed
* %Promote debug_query_string into a server-side function current_query()
* %Allow the identifier length to be increased via a configure option
* Remove Win32 rename/unlink looping if unnecessary
-* -Remove kerberos4 from source tree
* Allow cross-compiling by generating the zic database on the target system
* Improve NLS maintenace of libpgport messages linked onto applications
* Allow ecpg to work with MSVC and BCC
-* -Make src/port/snprintf.c thread-safe
* Add xpath_array() to /contrib/xml2 to return results as an array
* Allow building in directories containing spaces
spaces and then copy the install to a directory with spaces.
* Fix sgmltools so PDFs can be generated with bookmarks
-* -Add C code on Unix to copy directories for use in creating new databases
* %Clean up compiler warnings (especially with gcc version 4)
backslashes
o Re-enable timezone output on log_line_prefix '%t' when a
shorter timezone string is available
- o -Improve dlerror() reporting string
o Fix problem with shared memory on the Win32 Terminal Server
o %Add support for Unicode
<body bgcolor="#FFFFFF" text="#000000" link="#FF0000" vlink="#A00000" alink="#0000FF">
<h1><a name="section_1">PostgreSQL TODO List</a></h1>
<p>Current maintainer: Bruce Momjian (<a href="mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</a>)<br/>
-Last updated: Fri Aug 26 14:52:30 EDT 2005
+Last updated: Fri Aug 26 14:59:39 EDT 2005
</p>
<p>The most recent version of this document can be viewed at<br/>
<a href="http://www.postgresql.org/docs/faqs.TODO.html">http://www.postgresql.org/docs/faqs.TODO.html</a>.
<ul>
<li>%Remove behavior of postmaster -o after making postmaster/postgres
flags unique
- </li><li>-<em>Allow limits on per-db/role connections</em>
</li><li>%Allow pooled connections to list all prepared queries
<p> This would allow an application inheriting a pooled connection to know
the queries prepared in the current session.
it is unknown whether other problems exist. This item mostly
requires additional testing rather than of writing any new code.
</p>
- </li><li>-<em>Prevent dropping user that still owns objects, or auto-drop the objects</em>
</li><li>%Set proper permissions on non-system schemas during db creation
<p> Currently all schemas are owned by the super-user because they are
copied from the template1 database.
</p>
- </li><li>-<em>Add the client IP address and port to pg_stat_activity</em>
</li><li>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
a database for analysis.
</p>
</li><li>%Add ability to monitor the use of temporary sort files
- </li><li>-<em>Add session start time and last statement time to pg_stat_activity</em>
- </li><li>-<em>Add a function that returns the start time of the postmaster</em>
</li><li>Allow server logs to be remotely read and removed using SQL commands
</li></ul>
<h1><a name="section_4">Data Types</a></h1>
</li><li>Arrays
<ul>
<li>Allow NULLs in arrays
- </li><li>-<em>Allow MIN()/MAX() on arrays</em>
</li><li>Delay resolution of array expression's data type so assignment
coercion can be performed on empty array expressions
- </li><li>-<em>Modify array literal representation to handle array index lower bound</em>
- of other than one
</li></ul>
</li><li>Binary Data
<ul>
<h1><a name="section_5">Functions</a></h1>
<ul>
- <li>-<em>Add function to return compressed length of TOAST data values</em>
- </li><li>Allow INET subnet tests using non-constants to be indexed
+ <li>Allow INET subnet tests using non-constants to be indexed
</li><li>Add transaction_timestamp(), statement_timestamp(), clock_timestamp()
functionality
<p> Current CURRENT_TIMESTAMP returns the start time of the current
</li><li>to_char(INTERVAL '43 hours 20 minutes', 'WK:DD:HR:MI') => 0:1:19:20
</li><li>to_char(INTERVAL '3 years 5 months','MM') => 41
</li></ul>
- </li><li>-<em>Prevent to_char() on interval from returning meaningless values</em>
-<p> For example, to_char('1 month', 'mon') is meaningless. Basically,
- most date-related parameters to to_char() are meaningless for
- intervals because interval is not anchored to a date.
-</p>
</li><li>Add sleep() function, remove from regress.c
</li></ul>
<h1><a name="section_6">Multi-Language Support</a></h1>
</p>
</li><li>Add the functionality for WITH CHECK OPTION clause of CREATE VIEW
</li><li>Allow NOTIFY in rules involving conditionals
- </li><li>-<em>Have views on temporary tables exist in the temporary namespace</em>
- </li><li>-<em>Allow temporary views on non-temporary tables</em>
</li><li>Allow VIEW/RULE recompilation when the underlying tables change
</li></ul>
<h1><a name="section_8">SQL Commands</a></h1>
<ul>
- <li>-<em>Add BETWEEN SYMMETRIC/ASYMMETRIC</em>
- </li><li>Change LIMIT/OFFSET and FETCH/MOVE to use int8
- </li><li>-<em>Add E'' escape string marker so eventually ordinary strings can treat</em>
- backslashes literally, for portability
- </li><li>-<em>Allow additional tables to be specified in DELETE for joins</em>
-<p> UPDATE already allows this (UPDATE...FROM) but we need similar
- functionality in DELETE. It's been agreed that the keyword should
- be USING, to avoid anything as confusing as DELETE FROM a FROM b.
-</p>
+ <li>Change LIMIT/OFFSET and FETCH/MOVE to use int8
</li><li>Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT
- </li><li>-<em>Allow REINDEX to rebuild all database indexes</em>
</li><li>Add ROLLUP, CUBE, GROUPING SETS options to GROUP BY
</li><li>%Allow SET CONSTRAINTS to be qualified by schema/table name
</li><li>%Allow TRUNCATE ... CASCADE/RESTRICT
creation
</li><li>%Add COMMENT ON for all cluster global objects (roles, databases
and tablespaces)
- </li><li>-<em>Add an option to automatically use savepoints for each statement in a</em>
- multi-statement transaction.
-<p> When enabled, this would allow errors in multi-statement transactions
- to be automatically ignored.
-</p>
</li><li>%Make row-wise comparisons work per SQL spec
</li><li>Add RESET CONNECTION command to reset all session state
<p> This would include resetting of all variables (RESET ALL), dropping of
</p>
</li><li>Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
columns
- </li><li>-<em>Allow FOR UPDATE queries to do NOWAIT locks</em>
</li></ul>
</li><li>ALTER
<ul>
<p> This requires the use of a savepoint before each COPY line is
processed, with ROLLBACK on COPY failure.
</p>
- </li><li>-<em>Allow COPY to understand \x as a hex byte</em>
</li><li>%Have COPY return the number of rows loaded/unloaded?
- </li><li>-<em>Allow COPY to optionally include column headings in the first line</em>
- </li><li>-<em>Allow COPY FROM ... CSV to interpret newlines and carriage</em>
- returns in data
</li></ul>
</li><li>GRANT/REVOKE
<ul>
</li></ul>
</li><li>SHOW/SET
<ul>
- <li>-<em>Have SHOW ALL show descriptions for server-side variables</em>
- </li><li>Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
+ <li>Add SET PERFORMANCE_TIPS option to suggest INDEX, VACUUM, VACUUM
ANALYZE, and CLUSTER
</li><li>Add SET PATH for schemas?
<p> This is basically the same as SET search_path.
</li></ul>
</li><li>Server-Side Languages
<ul>
- <li>-<em>Allow PL/PgSQL's RAISE function to take expressions</em>
-<p> Currently only constants are supported.
-</p>
- </li><li>-<em>Change PL/PgSQL to use palloc() instead of malloc()</em>
- </li><li>Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
+ <li>Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
</li><li>Allow function parameters to be passed by name,
get_employee_salary(emp_id => 12345, tax_year => 2001)
</li><li>Add Oracle-style packages
</li><li>Add table function support to pltcl, plpython
- </li><li>-<em>Allow PL/pgSQL EXECUTE query_var INTO record_var;</em>
</li><li>Add capability to create and call PROCEDURES
</li><li>Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[<a href="http://momjian.postgresql.org/cgi-bin/pgtodo?"></a>]
</li><li>Allow function argument names to be queries from PL/PgSQL
</li><li>Add MOVE to PL/pgSQL
- </li><li>-<em>Pass arrays natively instead of as text between plperl and postgres</em>
</li><li>Add support for polymorphic arguments and return types to
languages other than PL/PgSQL
</li><li>Add support for OUT and INOUT parameters to languages other
</li><li>%Add dumping of comments on index columns
</li><li>%Replace crude DELETE FROM method of pg_dumpall --clean for
cleaning of roles with separate DROP commands
- </li><li>-<em>Add dumping and restoring of LOB comments</em>
</li><li>Stop dumping CASCADE on DROP TYPE commands in clean mode
</li><li>%Add full object name to the tag field. eg. for operators we need
'=(integer, integer)', instead of just '='.
memory. This could exhaust memory for very large trigger queues.
This item involves dumping large queues into files.
</p>
- </li><li>-<em>Implement shared row locks and use them in RI triggers</em>
</li><li>Change foreign key constraint for array -> element to mean element
in array?
</li><li>Allow DEFERRABLE UNIQUE constraints?
- </li><li>-<em>Allow triggers to be disabled</em>
</li><li>Allow triggers to be disabled in only the current session.
<p> This is currently possible by starting a multi-statement transaction,
modifying the system tables, performing the desired SQL, restoring the
semantics
<p> This can be done using dblink and two-phase commit.
</p>
- </li><li>-<em>Add two-phase commit</em>
</li><li>Add the features of packages
<ul>
<li>Make private objects accessable only to objects in the same schema
that can span more than one table.
</p>
</li><li>Allow SELECT ... FOR UPDATE on inherited tables
- </li><li>-<em>Prevent inherited tables from expanding temporary subtables of other</em>
- sessions
</li><li>Add UNIQUE capability to non-btree indexes
- </li><li>-<em>Use indexes for MIN() and MAX()</em>
-<p> MIN/MAX queries can already be rewritten as SELECT col FROM tab ORDER
- BY col {DESC} LIMIT 1. Completing this item involves doing this
- transformation automatically.
-</p>
- </li><li>-<em>Use index to restrict rows returned by multi-key index when used with</em>
- non-consecutive keys to reduce heap accesses
-<p> For an index on col1,col2,col3, and a WHERE clause of col1 = 5 and
- col3 = 9, spin though the index checking for col1 and col3 matches,
- rather than just col1; also called skip-scanning.
-</p>
</li><li>Prevent index uniqueness checks when UPDATE does not modify the column
<p> Uniqueness (index) checks are done when updating a column even if the
column is not modified by the UPDATE.
-</p>
- </li><li>-<em>Fetch heap pages matching index entries in sequential order</em>
-<p> Rather than randomly accessing heap pages based on index entries, mark
- heap pages needing access in a bitmap and do the lookups in sequential
- order. Another method would be to sort heap ctids matching the index
- before accessing the heap rows.
-</p>
- </li><li>-<em>Allow non-bitmap indexes to be combined by creating bitmaps in memory</em>
-<p> This feature allows separate indexes to be ANDed or ORed together. This
- is particularly useful for data warehousing applications that need to
- query the database in an many permutations. This feature scans an index
- and creates an in-memory bitmap, and allows that bitmap to be combined
- with other bitmap created in a similar way. The bitmap can either index
- all TIDs, or be lossy, meaning it records just page numbers and each
- page tuple has to be checked for validity in a separate pass.
</p>
</li><li>Allow the creation of on-disk bitmap indexes which can be quickly
combined with other bitmap indexes
one column or expression indexes, perhaps using per-index statistics
</li><li>Add fillfactor to control reserved free space during index creation
</li><li>Allow the creation of indexes with mixed ascending/descending specifiers
- </li><li>-<em>Fix incorrect rtree results due to wrong assumptions about "over"</em>
- operator semantics
</li><li>Allow constraint_exclusion to work for UNIONs like it does for
inheritance, and allow it to work for UPDATE and DELETE queries
</li><li>GIST
<ul>
<li>Add more GIST index support for geometric data types
- </li><li>-<em>Add concurrency to GIST</em>
</li><li>Allow GIST indexes to create certain complex index types, like
digital trees (see Aoki)
</li></ul>
<p> Ideally this requires a separate test program that can be run
at initdb time or optionally later.
</p>
- </li><li>-<em>Allow multiple blocks to be written to WAL with one write()</em>
</li><li>%Add an option to sync() before fsync()'ing checkpoint files
</li><li>Add program to test if fsync has a delay compared to non-fsync
</li></ul>
backends that also have the file open, and the feature is not supported
on all operating systems.
</p>
- </li><li>-<em>Consider use of open/fcntl(O_DIRECT) to minimize OS caching,</em>
- for WAL writes
-<p> O_DIRECT doesn't have the same media write guarantees as fsync, so it
- is in addition to the fsync method, not in place of it.
-</p>
- </li><li>-<em>Cache last known per-tuple offsets to speed long tuple access</em>
</li><li>Speed up COUNT(*)
<p> We could use a fixed row count and a +/- count to follow MVCC
visibility rules, or a single cached value could be used and
</li><li>Query execute plan
</li><li>Query results
</li></ul>
- </li><li>-<em>Allow the size of the buffer cache used by temporary objects to be</em>
- specified as a GUC variable
-<p> Larger local buffer cache sizes requires more efficient handling of
- local cache lookups.
-</p>
- </li><li>-<em>Improve the background writer</em>
-<p> Allow the background writer to more efficiently write dirty buffers
- from the end of the LRU cache and use a clock sweep algorithm to
- write other dirty buffers to reduced checkpoint I/O
-</p>
</li><li>Allow sequential scans to take advantage of other concurrent
sequentiqal scans, also called "Synchronised Scanning"
<p> One possible implementation is to start sequential scans from the lowest
write lock. However, the read lock promotion to write lock could lead
to deadlock situations.
</p>
- </li><li>-<em>Add a warning when the free space map is too small</em>
</li><li>Maintain a map of recently-expired rows
<p> This allows vacuum to target specific pages for possible free space
without requiring a sequential scan.
</li><li>%Add system view to show free space map contents
</li><li>Auto-vacuum
<ul>
- <li>-<em>Move into the backend code</em>
- </li><li>Use free-space map information to guide refilling
+ <li>Use free-space map information to guide refilling
</li><li>%Issue log message to suggest VACUUM FULL if a table is nearly
empty?
</li><li>Improve xid wraparound detection by recording per-table rather
<h1><a name="section_17">Locking</a></h1>
<ul>
- <li>-<em>Make locking of shared data structures more fine-grained</em>
-<p> This requires that more locks be acquired but this would reduce lock
- contention, improving concurrency.
-</p>
- </li><li>Add code to detect an SMP machine and handle spinlocks accordingly
+ <li>Add code to detect an SMP machine and handle spinlocks accordingly
from distributted.net, <a href="http://www1.distributed.net/source">http://www1.distributed.net/source</a>,
in client/common/cpucheck.cpp
<p> On SMP machines, it is possible that locks might be released shortly,
while on non-SMP machines, the backend should sleep so the process
holding the lock can complete and release it.
-</p>
- </li><li>-<em>Improve SMP performance on i386 machines</em>
-<p> i386-based SMP machines can generate excessive context switching
- caused by lock failure in high concurrency situations. This may be
- caused by CPU cache line invalidation inefficiencies.
</p>
</li><li>Research use of sched_yield() for spinlock acquisition failure
</li><li>Fix priority ordering of read and write light-weight locks (Neil)
eliminated from point-in-time archive files.
</p>
<ul>
- <li>-Add ability to turn off full page writes
- </li><li>When off, write CRC to WAL and check file system blocks
+ <li>When off, write CRC to WAL and check file system blocks
on recovery
<p> If CRC check fails during recovery, remember the page in case
a later CRC for that page properly matches.
remove the 'fsync' parameter (which results in an an inconsistent
database) in favor of this capability.
</p>
- </li><li>-<em>Eliminate WAL logging for CREATE TABLE AS when not doing WAL archiving</em>
- </li><li>-<em>Change WAL to use 32-bit CRC, for performance reasons</em>
</li></ul>
<h1><a name="section_20">Optimizer / Executor</a></h1>
</li><li>Create utility to compute accurate random_page_cost value
</li><li>Improve ability to display optimizer analysis using OPTIMIZER_DEBUG
</li><li>Have EXPLAIN ANALYZE highlight poor optimizer estimates
- </li><li>-<em>Use CHECK constraints to influence optimizer decisions</em>
-<p> CHECK constraints contain information about the distribution of values
- within the table. This is also useful for implementing subtables where
- a tables content is distributed across several subtables.
-</p>
</li><li>Consider using hash buckets to do DISTINCT, rather than sorting
<p> This would be beneficial when there are few distinct values.
</p>
- </li><li>-<em>ANALYZE should record a pg_statistic entry for an all-NULL column</em>
</li><li>Log queries where the optimizer row estimates were dramatically
different from the number of rows actually found?
</li><li>Allow constraint_elimination to be automatically performed
</li><li>%Promote debug_query_string into a server-side function current_query()
</li><li>%Allow the identifier length to be increased via a configure option
</li><li>Remove Win32 rename/unlink looping if unnecessary
- </li><li>-<em>Remove kerberos4 from source tree</em>
</li><li>Allow cross-compiling by generating the zic database on the target system
</li><li>Improve NLS maintenace of libpgport messages linked onto applications
</li><li>Allow ecpg to work with MSVC and BCC
- </li><li>-<em>Make src/port/snprintf.c thread-safe</em>
</li><li>Add xpath_array() to /contrib/xml2 to return results as an array
</li><li>Allow building in directories containing spaces
<p> This is probably not possible because 'gmake' and other compiler tools
spaces and then copy the install to a directory with spaces.
</p>
</li><li>Fix sgmltools so PDFs can be generated with bookmarks
- </li><li>-<em>Add C code on Unix to copy directories for use in creating new databases</em>
</li><li>%Clean up compiler warnings (especially with gcc version 4)
</li><li>Win32
<ul>
backslashes
</li><li>Re-enable timezone output on log_line_prefix '%t' when a
shorter timezone string is available
- </li><li>-<em>Improve dlerror() reporting string</em>
</li><li>Fix problem with shared memory on the Win32 Terminal Server
</li><li>%Add support for Unicode
<p> To fix this, the data needs to be converted to/from UTF16/UTF8
projects / postgresql / commitdiff