Release 9.3Release Date2013-XX-XX, CURRENT AS OF 2013-05-03Overview
ADD HERE Major enhancements include:
The above items are explained in more detail in the sections below.
Migration to Version 9.3
A dump/restore using pg_dumpall>, or use
of pg_upgrade>, is
required for those wishing to migrate data from any previous release.
Version 9.3 contains a number of changes that may affect compatibility
with previous releases. Observe the following incompatibilities:
Server Settings
Rename replication_timeout> to wal_sender_timeout>
(Amit Kapila)
This setting controls the WAL> sender timeout.
Improve grouping of sessions waiting for commit_delay>
(Peter Geoghegan)
This improves the usefulness and behavior of
commit_delay>.
Require superuser privileges to set commit_delay>
because it can now potentially delay other sessions (Simon Riggs)
Have session id> (%c>) in log_line_prefix>
always output four hex digits after the period (Bruce Momjian)
Other
Throw an error if expiring tuple is again updated or deleted (Kevin Grittner)
DETAILS?
Change ON UPDATE
SET NULL/SET DEFAULT> foreign key actions to affect
all referenced columns, not just those referenced in the
UPDATE> (Tom Lane)
Previously only columns referenced in the UPDATE> were
set to null or DEFAULT>.
Internally store default foreign key matches (non-FULL>,
non-PARTIAL>) as simple> (Tom Lane)
These were previously stored as "<unspecified>".
This changes the value stored in system column pg_constraint.confmatchtype>.
Store WAL> in a continuous
stream, rather than skipping the last 16MB segment every 4GB
(Heikki Linnakangas)
Previously, WAL> files ending in FF> were
not used. If you have WAL> backup or restore scripts
that took that skipping into account, they need to be adjusted.
Changes
Below you will find a detailed account of the changes between
PostgreSQL 9.3 and the previous major
release.
ServerLocking
Prevent non-key-field row updates from locking foreign key rows
(Álvaro Herrera, Noah Misch, Andres Freund, Alexander
Shulgin, Marti Raudsepp)
This improves concurrency and reduces the probability of deadlocks.
UPDATE>s on non-key columns use the new SELECT
FOR NO KEY UPDATE> lock type, and foreign key checks use the
new SELECT FOR KEY SHARE> lock mode.
Add configuration variable lock_timeout> to limit
lock wait duration (Zoltán Böszörményi)
Add cache of local locks (Jeff Janes)
This speeds lock release at statement completion in
transactions that hold many locks; it is particularly useful
for pg_dump>.
Indexes
Add SP-GiST>
support for range data types (Alexander Korotkov)
Allow unlogged GiST> indexes
(Jeevan Chalke)
Improve concurrency of hash indexes (Robert Haas)
Optimizer
Collect and use histograms for range
types (Alexander Korotkov)
Reduce optimizer overhead by discarding plans with unneeded cheaper
startup costs (Tom Lane)
Improve optimizer cost estimation for index access (Tom Lane)
General Performance
Add COPY FREEZE>
option to avoid the overhead of later marking tuples as committed
(Simon Riggs, Jeff Davis)
Improve performance of NUMERIC> calculations
(Kyotaro Horiguchi)
Allow in-memory sorts to use their full memory allocation (Jeff Janes)
Users who have set work_mem> based on the
previous behavior should revisit that setting.
Improve performance for transactions creating, rebuilding, or
dropping many relations (Jeff Janes, Tomas Vondra)
Improve performance of the CREATE TEMPORARY TABLE ... ON
COMMIT DELETE ROWS> clause by only issuing delete if
the temporary table was accessed (Heikki Linnakangas)
Have vacuum recheck visibility after it has removed expired tuples
(Pavan Deolasee)
This increases the chance of a page being marked as all-visible.
Split the pg_stat_tmp>
statistics file into per-database and global files (Tomas Vondra)
This reduces the statistics management read and write overhead.
Monitoring
Add optional ability to checksum data pages and
report corruption (Simon Riggs, Jeff Davis, Greg Smith, Ants Aasma)
The checksum option can be set during initdb.
Allow pg_terminate_backend()>
to terminate other backends with the same role (Dan Farina)
Previously, only superusers could terminate other sessions.
Allow the statistics
collector to operate properly in cases where the system
clock goes backwards (Tom Lane)
Previously statistics collection would stop until the time again
reached the previously-stored latest time.
Authentication
Improve LDAP> error
reporting and documentation (Peter Eisentraut)
Add support for LDAP> authentication to be specified
in URL> format (Peter Eisentraut)
Change the ssl_ciphers> parameter
to start with DEFAULT>, rather than ALL>,
then remove insecure ciphers (Magnus Hagander)
It is assumed DEFAULT> is more appropriate cipher set.
Parse and load pg_ident.conf>
once, not during each connection (Amit Kapila)
This is similar to how pg_hba.conf>
is processed.
Server Settings
Dramatically reduce System V shared
memory usage (Robert Haas)
Instead, on Unix-like systems, mmap()> is used for
shared memory. This eliminates the requirement of increasing
System V shared memory kernel parameters.
Allow the postmaster to listen on multiple Unix-domain sockets
(Honza Horák)
This renames configuration parameter
unix_socket_directory> to unix_socket_directories>,
which accepts a list of directories.
Allow a directory of configuration files to be processed (Magnus
Hagander, Greg Smith, Selena Deckelmann)
The directory is specified as include_dir> in server
configuration file.
Increase the maximum initdb-configured value for shared_buffers>
to 128MB (Robert Haas)
This is the maximum value initdb attempts to set in postgresql.conf>;
the previous value was 32MB.
Remove the external
PID> file on postmaster exit (Peter Eisentraut)
Replication and Recovery
Allow a streaming replication standbys to follow a timeline switch
(Heikki Linnakangas)
This allows streaming standbys to feed from newly-promoted slaves.
Previously slaves required access to a WAL> archive directory to
accomplish this.
Add SQL> functions pg_is_in_backup()>
and pg_backup_start_time()>
(Gilles Darold)
These functions report the status of base backups.
Improve performance of streaming log shipping with synchronous_commit>
disabled (Andres Freund)
Allow much faster promotion of a streaming standby to primary (Simon
Riggs, Kyotaro Horiguchi)
Add the last checkpoint's redo location to pg_controldata>'s
output (Fujii Masao)
This information is useful for determining the WAL>
files needed for restore.
Allow tooling like pg_receivexlog>
to run on computers with different architectures (Heikki
Linnakangas)
Have pg_basebackup>
This simplifies setting up a standby server.
Allow pg_receivexlog>
and pg_basebackup>
Add wal_receiver_timeout>
parameter to control the WAL> receiver timeout
(Amit Kapila)
This allows more rapid detection of
connection failure. No longer set wal_receiver_status_interval>?
Write-Ahead Log
(WAL>)
Change the WAL> record format to allow splitting the record header
across pages (Heikki Linnakangas)
The new format is slightly more compact.
Queries
Implement SQL>-standard LATERAL> for
FROM>-clause subqueries and function calls (Tom Lane)
This is useful in allowing FROM> subqueries and
functions to reference output from tables in the FROM>
clause. The LATERAL> keyword is optional for functions.
Add support for piping COPY> and psql> \copy>
to/from an external program (Etsuro Fujita)
Improve query string error location reporting (Tom Lane)
Only CREATE SCHEMA ... IF NOT EXISTS>? If so, remove it.
Object Manipulation
Add support to event triggers
(Dimitri Fontaine, Robert Haas, Álvaro Herrera)
This allows C functions to be called when specific commands are run.
Allow a multi-row VALUES> clause in rules
to reference OLD>/NEW> (Tom Lane)
Add CREATE SCHEMA ... IF
NOT EXISTS> clause (Fabrízio de Royes Mello)
Have REASSIGN
OWNED> also change ownership of shared objects
(Álvaro Herrera)
CREATE TABLE>
No longer output messages about implicit index and sequence creation (Robert Haas)
These messages now appear with DEBUG1>-level output.
Allow CREATE TABLE IF NOT EXISTS> to succeed for a
non-existent schema (Bruce Momjian)
Previously, it threw an error if the schema did not exist.
Constraints
Provide clients with constraint violation details
as separate fields (Pavel Stehule)
This allows clients to retrieve table, column, data type, or constraint
name error details. Previously such information had to be extracted from
error strings. Client language support is required to access these
fields.
ALTER>
Add the ALTER TYPE ... IF
NOT EXISTS> clause when adding enumerated type labels
(Andrew Dunstan)
This is useful for conditional label creation in transaction blocks.
Add ALTER ROLE ALL
SET> to add settings to all users (Peter Eisentraut)
This allows settings to apply to all users in all databases. ALTER DATABASE SET>
already allowed addition of settings for all users in a single
database. postgresql.conf> has a similar effect.
Add support for ALTER RULE
... RENAME> (Ali Dar)
VIEWs>
Add a materialized
view relations (Kevin Grittner)
Unlike ordinary views, where the base tables are read on every access,
materialized views create physical tables at creation or refresh time.
Access to the materialized view reads from these materialized physical
tables. There is no facility for incrementally refreshing materialized
views or auto-accessing them via base table access.
Make simple views auto-updatable
(Dean Rasheed)
Simple views that reference some or all columns from a
single base table are now updatable by default. More
complex views can be made updatable using INSTEAD OF> triggers
or INSTEAD> rules.
Allow views and rules to handle cases where referenced tables are
renamed, or columns are renamed, added, or dropped (Tom Lane)
HOW DOES IT HANDLE A DROPPED REFERENCED COLUMN?
Add CREATE RECURSIVE
VIEW> syntax (Peter Eisentraut)
Internally this is translated into CREATE VIEW ... WITH
RECURSIVE>.
When converting a table to a view, remove its system columns (Tom Lane)
KEEP?
Data Types
Increase the maximum length of large
objects from 2GB to 4TB (Nozomi Anzai, Yugo Nagata)
This change includes new libpq and server-side 64-bit-enabled functions.
Allow text timezone
designations, e.g. America/Chicago> when using
the ISO> T> timestamptz format (Bruce Momjian)
JSON>
Add operators and functions
to extract values from JSON> data strings (Andrew Dunstan)
Allow JSON> data strings to be converted into records
(Andrew Dunstan)
Add functions
to convert values, records, and hstore data to JSON>
(Andrew Dunstan)
Functions
Add array_remove()>
and array_replace()>
functions (Marco Nenciarini, Gabriele Bartolini)
Allow concat()>
and format()>
to properly expand VARIADIC>-labeled arguments
(Pavel Stehule)
Improve format()>
to handle field width and left/right alignment (Pavel Stehule)
Have to_char()>,
to_date()>,
and to_timestamp()>
properly handle negative century designations (CC>)
(Bruce Momjian)
Previously the behavior was either wrong or inconsistent
with positive/AD> handling, e.g. format mask
IYYY-IW-DY>.
Have to_date()>
and to_timestamp()>
return proper results when mixing ISO> and Gregorian
week/day designations (Bruce Momjian)
Cause pg_get_viewdef()>
to default to wrapping after every SELECT> target list and
FROM> entry (Marko Tiikkaja)
This causes view output, like from pg_dump>, to be
more narrow.
Have map_sql_value_to_xml_value()> properly expand
domains (Pavel Stehule) WHAT IS THE USER-VISIBLE AFFECT HERE?
map_sql_value_to_xml_value IS A C FUNCTION
Force cached functions to be replanned if the search_path> changes
(Tom Lane)
Previously functions already run in the current session ignored
search_path> changes.
(Bruce Momjian)
Server-Side Languages
Allow SPI>
functions to access the number of rows processed by
COPY> (Pavel Stehule)
PL/pgSQL Server-Side Language
Allow PL/pgSQL to use RETURN> with a composite-type
expressions (Asif Rehman)
Previously RETURN> could only reference composite-type variables.
Allow PL/pgSQL to access the number of rows processed by
COPY> (Pavel Stehule)
The command is GET DIAGNOSTICS
x = ROW_COUNT>.
Allow greater flexibility in where keywords can be used in PL/pgSQL (Tom Lane)
PL/Python Server-Side Language
Add PL/Python result object string handler (Peter Eisentraut)
This allows plpy.debug(rv) to output something reasonable.
Have PL/Python convert oid values to a proper Python numeric type
(Peter Eisentraut)
Handle SPI> errors raised
explicitly with PL/Python's RAISE> the same as as
internal SPI> errors (Oskari Saarenmaa and Jan Urbanski)
Client Applications
Add command-line utility pg_isready> to
check if the server is ready to accept connections (Phil Sorber)
Support multiple
This is similar to the pg_dump>
Add
Add libpq function PQconninfo()>
to return connection information (Zoltán
Böszörményi, Magnus Hagander)
psql>
Adjust function cost settings so psql> tab
completion and pattern searching is more efficient (Tom Lane)
Improve psql> tab completion coverage (Jeff Janes,
Peter Eisentraut)
Allow the psql>
Previously this option only worked when reading from a file.
Remove psql> warning when connecting to an older
server (Peter Eisentraut)
The warning when connecting to a newer server was retained.
Backslash Commands
Add psql> \watch> command to repeatedly
execute commands (Will Leinweber)
Add psql> command \gset> to store query
results in psql> variables (Pavel Stehule)
Add SSL> information to psql>'s
\conninfo> command (Alastair Turner)
Add Security> label to psql> \df+>
output (Jon Erdman)
Allow psql> \l> to accept a database
name pattern (Peter Eisentraut)
In psql>, do not allow \connect> to
use defaults if there is no active connection (Bruce Momjian)
This might be the case if the server had crashed.
Properly reset state if psql>'s \g file>
command failed (Tom Lane)
Previously failed commands discarded output from subsequent commands.
Output
Add a latex-longtable> output format to
psql> (Bruce Momjian)
Longtable allows tables to span multiple pages.
Add a border=3> output mode to the psql>
latex> format (Bruce Momjian)
In psql> tuples-only and expanded modes, no longer
output (No rows)> (Peter Eisentraut)
In psql>, no longer print an empty line for
unaligned, expanded output for zero rows (Peter Eisentraut)
pg_dump>
Add pg_dump>
Have pg_dump> output functions in a predictable
order (Joel Jacobson)
Fix tar files emitted by pg_dump> and pg_basebackup>
to be POSIX> conformant (Brian Weaver, Tom Lane)
Add
The database name could already be supplied last without a flag.
initdb>
Have initdb fsync the newly created data directory (Jeff Davis)
This can be disabled by using
Add initdb
This is used by pg_upgrade>.
Have initdb issue a warning about placing the data directory at the
top of file system mount points (Bruce Momjian)
Source Code
Add an embedded list interface (Andres Freund)
Add infrastructure to better support plug-in background worker processes
(Álvaro Herrera)
Create a centralized timeout API> (Zoltán
Böszörményi)
Create libpgcommon and move pg_malloc()> and other
functions there (Álvaro Herrera, Andres Freund)
This allows libpgport to be used solely for porting code.
Standardize on naming of client-side memory allocation functions (Tom Lane)
Add compiler designations to indicate some ereport()>
and elog()> calls do not return (Peter Eisentraut,
Andres Freund, Tom Lane, Heikki Linnakangas)
Allow options to be passed to the regression
test output comparison utility via PG_REGRESS_DIFF_OPTS>
(Peter Eisentraut)
Add isolation tests for CREATE INDEX
CONCURRENTLY> (Abhijit Menon-Sen)
Remove typedefs for int2>/int4> as they are better
represented as int16>/int32> (Peter Eisentraut)
Fix install-strip on Mac OS
X> (Peter Eisentraut)
Remove configure flag
Rewrite pgindent in Perl> (Andrew Dunstan)
Add emacs macro to match PostgreSQL> perltidy
formatting (Peter Eisentraut)
Run tool to check the keyword list when the backend grammar is
changed (Tom Lane)
Centralize flex> and bison>
make> rules (Peter Eisentraut)
This is useful for pgxs> authors.
Add Assert()> functionality to client-side code for
frontend code (Andrew Dunstan)
Change many internal many backend functions to return OID>s
rather than void (Dimitri Fontaine)
This is useful for event triggers.
Invent pre-commit/pre-prepare/pre-subcommit events for transaction
callbacks (Tom Lane)
Loadable modules that use transaction callbacks might need modification
to handle these new event types.
Add function pg_identify_object()>
to dump an object in machine-readable format (Álvaro
Herrera)
Add post-ALTER>-object server hooks (KaiGai Kohei)
Implement a generic binary heap and use it for Merge-Append
operations (Abhijit Menon-Sen)
Improve ability to detect official timezone abbreviation changes
(Tom Lane)
Add pkg-config> support libpq>
and ecpg> libraries (Peter Eisentraut)
Remove src/tool/backend>, now that the content is on
the PostgreSQL> wiki (Bruce Momjian)
Split out WAL> reading as
an independent facility (Heikki Linnakangas, Andres Freund)
Use a 64-bit integer to represent WAL> positions
(XLogRecPtr>) instead of two 32-bit integers
(Heikki Linnakangas)
Generally, tools that need to read the WAL> format
will need to be adjusted.
Allow PL/Python to support
platform-specific include directories (Peter Eisentraut)
Allow PL/Python on OS
X> to build against custom versions of Python>
(Peter Eisentraut)
Additional Modules
Allow write-enabled foreign data wrappers
to support writes (KaiGai Kohei)
Add a Postgres> foreign
data wrapper contrib module (Shigeru Hanada)
This foreign data wrapper allows writes.
Add pg_xlogdump>
contrib program (Andres Freund)
Add support for indexing of regular-expression searches in
extension pg_trgm>
(Alexander Korotkov)
Improve pg_trgm>
handling of multibyte characters (Tom Lane)
Add pgstattuple function to report the
size of the GIN> pending index insertion list (Fujii Masao)
Have oid2name>,
pgbench>, and
vacuumlo> set
fallback_application_name> (Amit Kapila)
Improve output of pg_test_timing>
(Bruce Momjian)
Improve output of pg_test_fsync>
(Peter Geoghegan)
Improve dblink option validator
(Tom Lane)
Details?
pg_upgrade>
Allow pg_upgrade>
This allows parallel schema dump/restore of databases, as well as
parallel copy/link of data files per tablespace.
Have pg_upgrade> create unix-domain sockets in
the current directory (Bruce Momjian, Tom Lane)
This reduces the possibility that someone will accidentally connect
during the upgrade.
Have pg_upgrade>
Improve performance of pg_upgrade> for databases
with many tables (Bruce Momjian)
Increase pg_upgrade> logging content by showing
executed command (Álvaro Herrera)
Improve pg_upgrade>'s status display during
copy/link (Bruce Momjian)
pgbench>
Add
This adds foreign key constraints to tables, for use in foreign
key performance testing.
Allow pgbench> to aggregate performance statistics
and produce output every
Add pg_bench>
Reduce and improve the status message output of
pgbench>'s initialization mode (Robert Haas,
Peter Eisentraut)
Add pgbench>
Output pgbench> elapsed and estimated remaining
time during initialization logging (Tomas Vondra)
Allow pgbench> to use a larger scale factor
(Greg Smith)
Documentation
Allow EPUB>-format documentation to be created
(Peter Eisentraut)
Update FreeBSD> kernel configuration documentation
(Brad Davis)
Improve WINDOW>
function documentation (Bruce Momjian, Tom Lane)
Add instructions for setting
up the documentation tool chain on Mac OS X>
(Peter Eisentraut)
Improve commit_delay>
documentation (Peter Geoghegan)