Release 9.1alpha4
Overview
PostgreSQL alpha releases are snapshots of development code. They
are intended to preview and test upcoming features and to provide
the possibility for early feedback. They should not be used in
production installations or active development projects. While the
PostgreSQL code is continuously subject to a number of automated
and manual tests, alpha releases might have serious bugs. Also
features may be changed incompatibly or removed at any time during
the development cycle.
The development cycle of a PostgreSQL major release alternates
between periods of development and periods of integration work,
called commit fests, normally one month each. Alpha releases are
planned to be produced at the end of every commit fest, thus every
two months. Since the first commit fest starts within a month from
the beginning of development altogether, early alpha releases are
not indicative of the likely feature set of the final release.
The release notes below highlight user visible changes and new
features. There are normally numerous bug fixes and performance
improvements in every new snapshot of PostgreSQL, and it would be
too bulky to attempt to list them all. Note that many bug fixes are
also backported to stable releases of PostgreSQL, and you should be
using those if you are looking for bug-fix-only upgrades for your
current installations.
These notes are cumulative over all alpha releases of the current
development cycle. Items that are new in the latest alpha release are
emphasized.
Migration
To upgrade from any release to an alpha release or from an alpha
release to any other release will most likely require a
dump/restore upgrade procedure. It may happen that this is not
necessary in particular cases, but that is not verified beforehand.
(The server will warn you in any case when a dump/restore is
necessary if you attempt to use it with an old data directory.)
Note, however, that the dump/restore upgrade procedure is expected
to work for alpha releases, and problems in this area should be
reported.
Testing
The primary reason we release alphas is to get users to test new
features as early as possible. If you are interested in helping
with organized testing, please see
the
testing information page.
Changes
Replication and Recovery
Support named restore points
Named restore points can be created using a new server function,
pg_create_restore_point. A new recovery.conf parameter,
recovery_target_name, allows recovery to stop at a named restore
point.
Add new recovery.conf parameter, pause_at_recovery_target,
which defaults to on
This is a behavior change from previous releases, which entered normal
running upon reaching the recovery target.
Add pg_is_xlog_replay_paused, pg_xlog_replay_pause, and
pg_xlog_replay_resume functions
Allow standby servers to send information on running
transactions back to the master
When enabled, this prevents most recovery conflicts on the slave,
but may cause MVCC bloat on the master.
Support streaming base backups
A new client, pg_basebackup, can be used to perform a streaming base
backup.
Synchronous replication
When enabled, transaction commit will wait for WAL records to be
replicated before informing the client that a transaction has
successfully committed.
When recovery_target_timelines='latest', periodically
rescan the archive for new timelines
SQL Features
Add ALTER ... SET SCHEMA support for conversions,
operators, operator classes, operator families, text search
configurations, text search dictionaries, text search
parsers, and text search templates
Make TRUNCATE ... RESTART IDENTITY restart sequences transactionally
Previously, the sequence restart did not roll back on error.
This meant that an error between truncating and committing
left the sequences out of sync with the table contents, with
potentially bad consequences as were noted in a Warning on
the TRUNCATE man page.
Prevent invoking I/O conversion casts via
functional/attribute notation
This was found to be confusing in some cases. The
functionality is still available via cast syntax.
Add ALTER TYPE ... ADD/DROP/ALTER/RENAME ATTRIBUTE
Support triggers on views
This adds the SQL-standard concept of an INSTEAD OF trigger,
which is fired instead of performing a physical
insert/update/delete. The trigger function is passed the
entire old and/or new rows of the view, and must figure out
what to do to the underlying tables to implement the update.
So this feature can be used to implement updatable views
using trigger programming style rather than rule hacking.
Allow WITH clauses to be attached to INSERT, UPDATE, DELETE
statements
Allow INSERT/UPDATE/DELETE to be used inside a WITH
clause
Allow new values to be added to an existing enum type
Recognize functional dependency on primary
keys
This allows a table's other columns to be referenced without listing
them in GROUP BY, so long as the primary key column(s) are listed in
GROUP BY.
Add CREATE TABLE IF NOT EXISTS
Allow a table's row type to be cast to the table's supertype
if it's a type table
This is analagous to the existing facility that allows casting a row
type to a supertable's row type.
Make foreign data wrappers functional, and support FOREIGN
TABLEs
Foreign tables are component of SQL/MED, and provide a framework to
allow data stored outside the database to be accessed via SQL.
contrib/file_fdw is provided as a sample foreign data wrapper.
Allow a unique or primary key constraint to be created
using an existing index
Implement Serializable Snapshot Isolation, in order to
provide a more robust serializable transaction mode
In previous releases, the REPEATABLE READ and SERIALIZABLE isolation
levels were identical. The historical behavior of REPEATABLE READ is
unchanged in this release, but SERIALIZABLE now provides stronger
guarantees.
Allow foreign key constraints to be added as initially
NOT VALID, and validated later
Allow multiple collations to be used within a single
database
Support extensions, which provide a cleaner method for
installing and upgrading add-on modules, such as those in contrib
Teach ALTER TABLE ... SET DATA TYPE to avoid a table write
in some cases where it isn't necessary
Add ENCODING option to COPY TO/FROM
Allow binary I/O of type "void"
Rearrange snapshot handlling to make rule expansion more
consistent
Portals, SQL functions, and SPI all now agree that a completely new
MVCC snapshot should be obtained only after fully executing each
original query. This is equivalent to the prior behavior of EXPLAIN
ANALYZE. It eliminates one source of concurrency hazards for
rules, and should also be marginally faster.
Performance
Optimize the commit_siblings configuration parameter
KNN GiST, otherwise known as order-by-operator support for GiST
This can be used to optimize nearest-neighbor searches.
Improve process shutdown speed
Speed up conversion of signed integers to C strings
Support MergeAppend plans, to allow sorted output
from append relations
This eliminates the former need to sort the output of an
Append scan when an ordered scan of an inheritance tree is
wanted. This should be particularly useful for fast-start
cases such as queries with LIMIT.
Support unlogged tables
The contents of an unlogged table are not WAL-logged; thus, they
are not replicated and are truncated whenever the database system
enters recovery. Indexes on unlogged tables are also unlogged, but
GIST indexes on unlogged tables are not currently supported.
Hash joins now support right and full outer joins
Previously, full joins could be implemented only as a merge join,
and left and right outer joins could only hash the table on the
nullable side of the join. This optimization is particularly
advantageous when the non-nullable side of a left or right outer
join is much smaller than the nullable side.
Reduce the memory requirement for large ispell dictionaries
Teach CLUSTER to use seqscan-and-sort when it's faster than
indexscan
Improve GIN indexscan cost estimation
Fix GIN to support null keys, empty and null items, and
full index scans
Teach GIN to combine duplicate keys even across different
quals
Performance improvements in pg_upgrade for uses with many tables
Reduce lock levels of CREATE TRIGGER and some ALTER
TABLE, CREATE RULE actions
Add some knowledge about prefix matches to
tsmatchsel()
Replace the naive HYPOT() macro with a
standards-conformant hypotenuse function
Avoid unnecessary detoast when comparing unequal-length text
or bytea values for equality or inequality
Try to compact the background writer's fsync queue when
it overflows, instead of doing an fsync per block written
Server Settings
Change the default value of
standard_conforming_strings> to
on>
This is a significant incompatibility with previous releases,
because it may break unwary applications in security-critical ways.
Add log_file_mode> parameter
This allows control of the file permissions set on log files created
by the syslogger process.
Add restart_after_crash> parameter
Normally, we automatically restart after a backend crash, but in
some cases when PostgreSQL is invoked by clusterware it may be
desirable to suppress this behavior, so we now provide an option which
does that.
Teach replication connections to respect the value of
log_connections
Previously, they were always logged.
Log restartpoints in the same fashion as
checkpoints
Previously, less information was logged for restartpoints.
By default, set wal_buffers to a reasonable value, based
on the setting for shared_buffers
Administration and Monitoring
Instrument checkpoint sync calls
Use the same lexer for recovery.conf that has been used for postgresql.conf
Among other things, this allows values in recovery.conf to be unquoted.
Add column buffers_backend_fsync to pg_stat_bgwriter
Add monitoring function pg_last_xact_replay_timestamp()
Allow pg_ctl to register the service in either AUTO or DEMAND start
type
Support host names and host key word all
in pg_hba.conf
Make EXPLAIN show the function call expression of a
FunctionScan plan node, but only in VERBOSE mode
Make EXPLAIN print PARAM_EXEC Params as
the referenced expressions, rather than just $N
Add vacuum and analyze counters to pg_stat_*_tables
views
Add stats functions and views to provide access to a
transaction's own statistics counts
Implement remaining fields of
information_schema.sequences view
Add a "conflicts" column to pg_stat_database
New system view pg_stat_database_conflicts
New system view pg_stat_replication
This view shows which replication slaves are connected to the master,
and reports their status and replication progress.
When an autovacuum worker (other than one performing an
anti-wraparound vacuum) is unable to obtain a lock on the target
relation without blocking, skip the relation
This avoids pinning down an autovacuum worker. The next autovacuum
cycle will try again.
Track time of last statistics reset on databases and
bgwriter
Security
Add SECURITY LABEL command
This is intended as infrastructure to support integration
with label-based mandatory access control systems such as
SE-Linux.
Add server authentication over Unix-domain
sockets
This adds a libpq connection parameter requirepeer> that
specifies the user name that the server process is expected to run
under.
Add REPLICATION privilege for ROLEs
This makes it possible for replication to be performed by a
non-superuser account.
Built-in Functions
Add pg_read_binary_file() and whole-file-at-once
versions of pg_read_file()
Add new function format(), for sprintf-like string formatting
Add pg_describe_object() function to obtain textual
descriptions of objects as stored in pg_depend
Accept public as a pseudo-role
name in has_table_privilege() and friends
to see if a particular privilege has been granted to
PUBLIC
Add string functions: concat(), concat_ws(), left(),
right(), and reverse()
Add three-parameter forms of array_to_string and
string_to_array
Add XMLEXISTS function
Add xpath_exists() function
This is equivalent to XMLEXISTS except that it offers support for
namespace mapping.
Add xml_is_well_formed, xml_is_well_formed_document,
xml_is_well_formed_content functions to the core XML
code
These supersede a version previously offered by contrib/xml2.
Change pg_last_xlog_receive_location not to move
backwards
Data Types
Allow numeric to use a more compact, 2-byte header in
many cases
Add support for dividing money by money (yielding a
float8 result) and for casting between money and numeric
Server Tools
Add options to force quoting of all
identifiers
This includes a quote_all_identifiers parameter which affects the
behavior of the backend, and a --quote-all-identifiers argument to
pg_dump and pg_dumpall which sets the parameter and also affects the
quoting done internally by those applications.
Add a directory output format to pg_dump
psql
Add S
option to \dn command in psql
System schemas are now hidden by default by \dn.
Add \conninfo command to psql, to show current connection
info
Extend psql's \e and \ef commands so that a line number
can be specified, and the editor's cursor will be initially placed
on that line
To avoid assumptions about what switch the user's editor takes for
this purpose, there is a new psql variable EDITOR_LINENUMBER_SWITCH
with (at present) no default value.
Add a \sf (show function) command to psql, for those
times when you need to look at a function but don't wish to fire up
an editor
Make psql distinguish between unique indices and unique
constraints in \d printout
Add a \dL command to psql, to list procedural
languages
Allow psql to automatically deduce the client encoding from
the locale
libpq
Add PQlibVersion() function
When reporting the server as not responding, if the
hostname was supplied, also print the IP address
This allows IPv4 and IPv6 failures to be distinguished. Also
useful when a hostname resolves to multiple IP addresses.
Add functions PQping and PQpingParams to allow
detection of the server's status, including a status where
the server is running but refuses ordinary connections
pg_ctl uses the new function. This fixes the case where
pg_ctl reports that the server is not running (cannot
connect) but in fact it is running.
ecpg
Allow ecpg to accept dynamic cursor names even in WHERE
CURRENT OF clauses
Procedural Languages
PL/pgSQL
Modify the handling of RAISE without parameters so that
the error it throws can be caught in the same places that could
catch an ordinary RAISE ERROR in the same location
The previous coding insisted on throwing the error from the block
containing the active exception handler; which is arguably more
surprising, and definitely unlike Oracle's behavior.
Remove ancient PL/pgSQL line numbering hack
PL/pgSQL used to count the second line of the function body as
line 1>, if the first line was zero-length.
While this hack arguably has some benefit in terms of making
PL/pgsql's line numbering match the programmer's expectations, it
also makes PL/pgsql inconsistent with the remaining PLs, making it
difficult for clients to reliably determine where the error
actually is. On balance, it seems better to be consistent.
PL/Perl
Allow generic record arguments to PL/Perl functions
Convert PostgreSQL arrays passed to PL/perl functions to
Perl arrays
PL/Python
Fix an error when a set-returning function fails halfway
through
Make plpy.Fatal() raise FATAL, rather than ERROR
Skip dropped attributes when converting Python objects to
tuples
Improve the handling of exceptions
Fix the way that exceptions are handled by the plpy module for Python
3. Get rid of the global variable holding the error state, and
instead arrange to report PostgreSQL errors to Python immediately.
Call PLy_spi_execute_fetch_result inside the try/catch block, so that
errors from fetching tuples are correct reported as errors in the
SPI call. Use the built-in TypeError, not SPIError, for errors
having to do with argument counts or types. Use SPIError, not simply
plpy.Error, for errors in PLy_spi_execute_plan. Do not set
a Python exception if PyArg_ParseTuple failed, as it already sets
the correct exception.
Improve error reporting
Better error messages for errors in compiling anonymous
PL/Python blocks. Avoid prefixing error messages with the string
"PL/Python: ", which is redundant, given the error
context. Provide a separate exception class for each error code
the backend defines, and make it possible to get the SQLSTATE from
the exception object. Report Python errors from iterators
using PLy_elog, so that the Python exception is included in the
exception, and to avoid setting the errcode to
ERRCODE_DATA_EXCEPTION.
Check PL/Python functions for syntax errors at creation
time
Previously, even gross syntax errors would not be detected until
the function was executed.
Allow Python try/catch blocks to catch exceptions arising
from SPI calls
Invalidate PL/python functions with composite type
arguments when the type changes
Add quoting functions plpy.quote_ident, plpy.quote_literal,
and plpy.quote_nullable
Support table functions
Add a plpy.subtransaction() context manager
Assorted code cleanup
We now use HTABs instead of Python dictionary objects to cache
procedures. palloc is used in TopMemoryContext instead of malloc.
Avoid use of undocumented API PyObject_NEW. Free plan values in
the PlyPlanObject dealloc function. Add braces around an if block,
for readability. Refactor PLy_spi_prepare to save two levels of
indentation. Avoid palloc(0).
Additional Supplied Modules
New module contrib/auth_delay
New module contrib/sepgsql, an SE-Linux integration for
PostgreSQL
Updates to contrib/isn ISBN tables
Add the ability to compute per-statement latencies (ie,
average execution times) to pgbench
Teach levenshtein() about multi-byte
characters
Add btree_gist support for searching on not
equals
Add ssl_cipher() and ssl_version() functions to
contrib/sslinfo
Remove the arbitrary (and undocumented) limit on the
number of parameter=value pairs that can be handled by
xslt_process()
New module contrib/pg_test_fsync, replacing
src/tools/test_fsync
In contrib/pageinspect, fix the heap_page_item to return
infomasks as 32 bit values
In contrib/btree_gist, support KNN GiST searches
In contrib/pg_trgm, support LIKE, ILIKE, and KNN GiST
searches
Several improvements to contrib/intarray
Correct the behavior of array containment operators, so that an empty
array is contained in any other array. Fix bogus failures where
operators would claim that nulls were present when they really
weren't. Remove arbitrary restriction to 1-D arrays.
Fix pg_upgrade not to insist on pg_dumpall and psql in the
old cluster, since we don't need them, but do check for pg_resetxlog
in the old and new clusters, since we do. Avoid a useless attempt
to call pg_config in the old cluster.
Source Code, Build Options
Rewrite the GiST insertion logic so that the
post-recovery cleanup stage to finish incomplete inserts or
splits is not needed anymore
Add support for collecting minidump
-style crash dumps
on Windows
Replace pg_class.relistemp column with relpersistence
This is for upcoming work on unlogged and global temporary tables.
GNU make 3.80 or newer is now required
Improved support for parallel make, make -k, and make -q
Include the backend ID in the relpath of temporary
relations
This allows us to reliably remove all leftover temporary relation
files on cluster startup without reference to system catalogs or
WAL; therefore, we no longer include temporary relations in
XLOG_XACT_COMMIT and XLOG_XACT_ABORT WAL records.
Rewrite COMMENT ON object> code for better
modularity, and add necessary locking
Add a hook in ExecCheckRTPerms() for eventual use by
external security-checking modules
Standardize names and calling conventions of
get_whatever_oid functions
Make NestLoop plan nodes pass outer-relation variables
into their inner relation using the general PARAM_EXEC executor
parameter mechanism, rather than the ad-hoc kluge of passing the
outer tuple down through ExecReScan
Make details of the Numeric representation private to
numeric.c
Suppress some compiler warnings from
clang>
Allow make check in PL directories
Add a make check-world target
PGXS support for contrib/hstore