Comment 2 for bug 250483

This bug was fixed in the package postgresql-8.3 - 8.3.4-1ubuntu1

---------------
postgresql-8.3 (8.3.4-1ubuntu1) intrepid; urgency=low

  * Upload current Debian bzr head to intrepid to get back in sync with
    Debian.
  * debian/rules: Enable hardening when we build on Ubuntu. Lenny is too close
    to release, so we do not enable that yet for Debian. Add hardening-wrapper
    and lsb-release build dependencies for this.

postgresql-8.3 (8.3.4-1) unstable; urgency=low

  * New upstream bugfix release:
    - Fix bug in btree WAL recovery code. Recovery failed if the WAL ended
      partway through a page split operation.
    - Fix potential use of wrong cutoff XID for HOT page pruning.
      This error created a risk of corruption in system catalogs that are
      consulted by "VACUUM": dead tuple versions might be removed too
      soon. The impact of this on actual database operations would be
      minimal, since the system doesn't follow MVCC rules while examining
      catalogs, but it might result in transiently wrong output from
      pg_dump or other client programs.
    - Fix potential miscalculation of datfrozenxid.
    - Fix incorrect HOT updates after pg_class is reindexed.
      Corruption of pg_class could occur if REINDEX TABLE pg_class was
      followed in the same session by an ALTER TABLE RENAME or ALTER
      TABLE SET SCHEMA command.
    - Fix missed "combo cid" case. This error made rows incorrectly invisible
      to a transaction in which they had been deleted by multiple
      subtransactions that all aborted.
    - Prevent autovacuum from crashing if the table it's currently
      checking is deleted at just the wrong time.
    - Widen local lock counters from 32 to 64 bits. This responds to reports
      that the counters could overflow in sufficiently long transactions,
      leading to unexpected "lock is already held" errors.
    - Fix possible duplicate output of tuples during a GiST index scan.
    - Regenerate foreign key checking queries from scratch when either
      table is modified. Previously, 8.3 would attempt to replan the query,
      but would work from previously generated query text. This led to
      failures if a table or column was renamed.
    - Fix missed permissions checks when a view contains a simple UNION
      ALL construct. Permissions for the referenced tables were checked
      properly, but not permissions for the view itself.
    - Add checks in executor startup to ensure that the tuples produced
      by an "INSERT" or "UPDATE" will match the target table's current
      rowtype. This situation is believed to be impossible in 8.3, but it can
      happen in prior releases, so a check seems prudent.
    - Fix possible repeated drops during "DROP OWNED". This would typically
      result in strange errors such as "cache lookup failed for relation NNN".
    - Fix several memory leaks in XML operations.
    - Fix xmlserialize() to raise error properly for unacceptable target
      data type.
    - Fix a couple of places that mis-handled multibyte characters in
      text search configuration file parsing. Certain characters occurring in
      configuration files would always cause "invalid byte sequence for
      encoding" failures.
    - Provide file name and line number location for all errors reported
      in text search configuration files.
    - Fix AT TIME ZONE to first try to interpret its timezone argument as
      a timezone abbreviation, and only try it as a full timezone name if
      that fails, rather than the other way around as formerly.
    - Fix datetime input functions to correctly detect integer overflow
      when running on a 64-bit platform.
    - Prevent integer overflows during units conversion when displaying a
      configuration parameter that has units.
    - Improve performance of writing very long log messages to syslog.
    - Allow spaces in the suffix part of an LDAP URL in "pg_hba.conf".
    - Fix bug in backwards scanning of a cursor on a SELECT DISTINCT ON
      query.
    - Fix planner bug that could improperly push down IS NULL tests below
      an outer join.
    - Fix planner bug with nested sub-select expressions.
    - Fix planner to estimate that GROUP BY expressions yielding boolean
      results always result in two groups, regardless of the expressions'
      contents.
    - Fix PL/PgSQL to not fail when a FOR loop's target variable is a
      record containing composite-type fields.
    - Fix PL/Tcl to behave correctly with Tcl 8.5, and to be more careful
      about the encoding of data sent to or from Tcl.
    - Improve performance of PQescapeBytea().
    - Fix ecpg to handle variables properly in "SET" commands.
    - Improve pg_dump and pg_restore's error reporting after failure to
      send a SQL command.
    - Fix pg_ctl to properly preserve postmaster command-line arguments
      across a restart.
    - Fix erroneous WAL file cutoff point calculation in pg_standby.
  * debian/control: Bump Standards-Version (no changes necessary).
  * debian/control: intarray contrib module was renamed to _int. (LP: #250483)

 -- Martin Pitt <email address hidden> Tue, 07 Oct 2008 14:57:53 +0200