lp:~eric-launchpad-soroos/postgresql/lucid-9.0

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

256. By erics <erics@builder-1004-2-x64-kvm> on 2012-10-04

TAS code has changed in upstream, patch does not apply cleanly.
Only necessary for armel, which I'm not concerned with.

255. By Martin Pitt on 2011-12-05

releasing version 9.0.6-1

254. By Martin Pitt on 2011-12-02

new upstream release 9.0.6

253. By Martin Pitt on 2011-09-26

releasing version 9.0.5-1

252. By Martin Pitt on 2011-09-25

* New upstream bug fix release:
  - Fix catalog cache invalidation after a "VACUUM FULL" or "CLUSTER"
    on a system catalog. In some cases the relocation of a system catalog
    row to another place would not be recognized by concurrent server
    processes, allowing catalog corruption to occur if they then tried to
    update that row. The worst-case outcome could be as bad as complete loss
    of a table.
  - Fix incorrect order of operations during sinval reset processing,
    and ensure that TOAST OIDs are preserved in system catalogs.
    These mistakes could lead to transient failures after a "VACUUM
    FULL" or "CLUSTER" on a system catalog.
  - Fix bugs in indexing of in-doubt HOT-updated tuples.
    These bugs could result in index corruption after reindexing a
    system catalog. They are not believed to affect user indexes.
  - Fix multiple bugs in GiST index page split processing.
  - Fix possible buffer overrun in tsvector_concat().
  - Fix crash in xml_recv when processing a "standalone" parameter.
  - Make pg_options_to_table return NULL for an option with no value.
    Previously such cases would result in a server crash.
  - Avoid possibly accessing off the end of memory in "ANALYZE" and in
    SJIS-2004 encoding conversion.
    This fixes some very-low-probability server crash scenarios.
  - Protect pg_stat_reset_shared() against NULL input.
  - Fix possible failure when a recovery conflict deadlock is detected
    within a sub-transaction.
  - Avoid spurious conflicts while recycling btree index pages during
    hot standby.
  - Shut down WAL receiver if it's still running at end of recovery.
    The postmaster formerly panicked in this situation, but it's
    actually a legitimate case.
  - Fix race condition in relcache init file invalidation.
    There was a window wherein a new backend process could read a stale
    init file but miss the inval messages that would tell it the data
    is stale. The result would be bizarre failures in catalog accesses,
    typically "could not read block 0 in file ..." later during
    startup.
  - Fix memory leak at end of a GiST index scan.
  - Fix memory leak when encoding conversion has to be done on incoming
    command strings and "LISTEN" is active.
  - Fix incorrect memory accounting (leading to possible memory bloat)
    in tuplestores supporting holdable cursors and plpgsql's RETURN
    NEXT command.
  - Fix trigger WHEN conditions when both BEFORE and AFTER triggers
    exist. Evaluation of WHEN conditions for AFTER ROW UPDATE triggers
    could crash if there had been a BEFORE ROW trigger fired for the same
    update.
  - Fix performance problem when constructing a large, lossy bitmap.
  - Fix join selectivity estimation for unique columns.
  - Fix nested PlaceHolderVar expressions that appear only in
    sub-select target lists. This mistake could result in outputs of an
    outer join incorrectly appearing as NULL.
  - Allow the planner to assume that empty parent tables really are
    empty. Normally an empty table is assumed to have a certain minimum
    size for planning purposes; but this heuristic seems to do more harm
    than good for the parent table of an inheritance hierarchy, which
    often is permanently empty.
  - Allow nested EXISTS queries to be optimized properly.
  - Fix array- and path-creating functions to ensure padding bytes are
    zeroes. This avoids some situations where the planner will think that
    semantically-equal constants are not equal, resulting in poor
    optimization.
  - Fix "EXPLAIN" to handle gating Result nodes within inner-indexscan
    subplans. The usual symptom of this oversight was "bogus varno" errors.
  - Fix btree preprocessing of "indexedcol" IS NULL conditions.
    Such a condition is unsatisfiable if combined with any other type
    of btree-indexable condition on the same index column. The case was
    handled incorrectly in 9.0.0 and later, leading to query output
    where there should be none.
  - Work around gcc 4.6.0 bug that breaks WAL replay.
    This could lead to loss of committed transactions after a server
    crash.
  - Fix dump bug for VALUES in a view.
  - Disallow SELECT FOR UPDATE/SHARE on sequences.
    This operation doesn't work as expected and can lead to failures.
  - Fix "VACUUM" so that it always updates pg_class.reltuples/relpages.
    This fixes some scenarios where autovacuum could make increasingly
    poor decisions about when to vacuum tables.
  - Defend against integer overflow when computing size of a hash table.
  - Fix cases where "CLUSTER" might attempt to access already-removed
    TOAST data.
  - Fix premature timeout failures during initial authentication
    transaction.
  - Fix portability bugs in use of credentials control messages for
    "peer" authentication. (see #627596, already fixed in Debian before)
  - Fix SSPI login when multiple roundtrips are required. The typical
    symptom of this problem was "The function requested is not supported"
    errors during SSPI login.
  - Fix failure when adding a new variable of a custom variable class
    to "postgresql.conf".
  - Throw an error if "pg_hba.conf" contains hostssl but SSL is
    disabled. This was concluded to be more user-friendly than the previous
    behavior of silently ignoring such lines.
  - Fix failure when "DROP OWNED BY" attempts to remove default
    privileges on sequences.
  - Fix typo in pg_srand48 seed initialization.
    This led to failure to use all bits of the provided seed. This
    function is not used on most platforms (only those without
    srandom), and the potential security exposure from a
    less-random-than-expected seed seems minimal in any case.
  - Avoid integer overflow when the sum of LIMIT and OFFSET values
    exceeds 2^63.
  - Add overflow checks to int4 and int8 versions of generate_series()
  - Fix trailing-zero removal in to_char(). In a format with FM and no digit
    positions after the decimal point, zeroes to the left of the decimal
    point could be removed incorrectly.
  - Fix pg_size_pretty() to avoid overflow for inputs close to 2^63.
  - Weaken plpgsql's check for typmod matching in record values.
    An overly enthusiastic check could lead to discarding length
    modifiers that should have been kept.
  - In pg_upgrade, avoid dumping orphaned temporary tables. This prevents
    situations wherein table OID assignments could get out of sync between old
    and new installations.
  - Fix pg_upgrade to preserve toast tables' relfrozenxids during an
    upgrade from 8.3. Failure to do this could lead to "pg_clog" files being
    removed too soon after the upgrade.
  - Fix psql's counting of script file line numbers during COPY from a
    different file.
  - Fix pg_restore's direct-to-database mode for
    standard_conforming_strings. pg_restore could emit incorrect commands
    when restoring directly to a database server from an archive file that
    had been made with standard_conforming_strings set to on.
  - Be more user-friendly about unsupported cases for parallel
    pg_restore. This change ensures that such cases are detected and
    reported before any restore actions have been taken.
  - Fix write-past-buffer-end and memory leak in libpq's LDAP service
    lookup code.
  - In libpq, avoid failures when using nonblocking I/O and an SSL
    connection.
  - Improve libpq's handling of failures during connection startup.
    In particular, the response to a server report of fork() failure
    during SSL connection startup is now saner.
  - Improve libpq's error reporting for SSL failures.
  - Fix PQsetvalue() to avoid possible crash when adding a new tuple to
    a PGresult originally obtained from a server query.
  - Make ecpglib write double values with 15 digits precision.
  - In ecpglib, be sure LC_NUMERIC setting is restored after an error.
  - Apply upstream fix for blowfish signed-character bug
    (CVE-2011-2483) (Tom Lane) (Closes: #636571)
    "contrib/pg_crypto"'s blowfish encryption code could give wrong
    results on platforms where char is signed (which is most), leading
    to encrypted passwords being weaker than they should be.
  - Fix memory leak in "contrib/seg".
  - Fix pgstatindex() to give consistent results for empty indexes.
  - Allow building with perl 5.14. (Closes: #628505)
* Drop 03-cmsgcred-size.patch. Fixed upstream in a different way.

251. By Martin Pitt on 2011-09-09

releasing version 9.0.4-3

250. By Martin Pitt on 2011-09-09

* debian/control: Stop building the client-side libraries, they are built by
  postgresql-9.1 now. Add libpq-dev build dependency.
* debian/rules: Drop check for uninstalled files, since it'd now break the
  build due to the uninstalled libraries.

249. By Martin Pitt on 2011-08-25

releasing version 9.0.4-2

248. By Martin Pitt on 2011-08-25

debian/control: Drop the versionless metapackages, they are built by -9.1
now.

247. By Martin Pitt on 2011-08-08

debian/control: Add explicit build dependency to jade. (Closes: #621492)