Comment 13 for bug 326372

Launchpad Janitor (janitor) wrote :

This bug was fixed in the package postgresql-8.3 - 8.3.6-0ubuntu8.04

postgresql-8.3 (8.3.6-0ubuntu8.04) hardy-proposed; urgency=low

  * New upstream bug fix release: (LP: #326372)
    - Make "DISCARD ALL" release advisory locks, in addition to
      everything it already did. This was decided to be the most
      appropriate behavior. This could affect existing applications,
    - Fix whole-index GiST scans to work correctly. This error could
      cause rows to be lost if a table is clustered on a GiST index.
    - Fix crash of xmlconcat(NULL).
    - Fix possible crash in ispell dictionary if high-bit-set characters
      are used as flags. This is known to be done by one widely
      available Norwegian dictionary, and the same condition may exist
      in others. (Closes: #513580)
    - Fix misordering of pg_dump output for composite types. The most
      likely problem was for user-defined operator classes to be
      dumped after indexes or views that needed them.
    - Improve handling of URLs in headline() function.
    - Improve handling of overlength headlines in headline() function.
    - Prevent possible Assert failure or misconversion if an encoding
      conversion is created with the wrong conversion function for the
      specified pair of encodings.
    - Fix possible Assert failure if a statement executed in PL/pgSQL is
      rewritten into another kind of statement, for example if an
      "INSERT" is rewritten into an "UPDATE".
    - Ensure that a snapshot is available to datatype input functions.
      This primarily affects domains that are declared with CHECK
      constraints involving user-defined stable or immutable functions.
      Such functions typically fail if no snapshot has been set.
    - Make it safer for SPI-using functions to be used within datatype
      I/O; in particular, to be used in domain check constraints.
    - Avoid unnecessary locking of small tables in "VACUUM".
    - Fix a problem that sometimes kept "ALTER TABLE ENABLE/DISABLE RULE"
      from being recognized by active sessions.
    - Fix a problem that made UPDATE RETURNING tableoid return zero
      instead of the correct OID.
    - Allow functions declared as taking ANYARRAY to work on the
      pg_statistic columns of that type.
      This used to work, but was unintentionally broken in 8.3.
    - Fix planner misestimation of selectivity when transitive equality
      is applied to an outer-join clause.
      This could result in bad plans for queries like ... from a left
      join b on a.a1 = b.b1 where a.a1 = 42 ...
    - Improve optimizer's handling of long IN lists. This change
      avoids wasting large amounts of time on such lists when
      constraint exclusion is enabled.
    - Prevent synchronous scan during GIN index build. Because GIN is
      optimized for inserting tuples in increasing TID order, choosing
      to use a synchronous scan could slow the build by a factor of
      three or more.
    - Ensure that the contents of a holdable cursor don't depend on the
      contents of TOAST tables. Previously, large field values in a
      cursor result might be represented as TOAST pointers, which
      would fail if the referenced table got dropped before the cursor
      is read, or if the large value is deleted and then vacuumed
      away. This cannot happen with an ordinary cursor, but it could
      with a cursor that is held past its creating transaction.
    - Fix memory leak when a set-returning function is terminated without
      reading its whole result.
    - Fix encoding conversion problems in XML functions when the database
      encoding isn't UTF-8.
    - Fix "contrib/dblink"'s dblink_get_result(text,bool) function.
    - Fix possible garbage output from "contrib/sslinfo" functions.
    - Fix incorrect behavior of "contrib/tsearch2" compatibility trigger
      when it's fired more than once in a command.
    - Fix possible mis-signaling in autovacuum.
    - Fix ecpg's handling of varchar structs.
    - Make all documentation reference pgsql-bugs and/or pgsql-hackers as
      appropriate, instead of the now-decommissioned pgsql-ports and
      pgsql-patches mailing lists.

 -- Martin Pitt <email address hidden> Fri, 06 Feb 2009 23:51:09 +0100