Permalink
Commits on Feb 9, 2011
  1. Merge branch 'upstream'

    jgoerzen committed Feb 9, 2011
  2. Fix abstime parsing

    lpsmith committed with jgoerzen Feb 9, 2011
    In PostgreSQL 9.0,  when PGgetvalue (from libpq) is called on column
    of type ab stime,  the string returned will contain a timezone.  I don't
    know if this is true for older PostgreSQL versions.   Also,  abstime is
    kind of obsolete.
  3. Fix abstime parsing

    lpsmith committed Feb 9, 2011
    In PostgreSQL 9.0,  when PGgetvalue (from libpq) is called on column
    of type ab stime,  the string returned will contain a timezone.  I don't
    know if this is true for older PostgreSQL versions.   Also,  abstime is
    kind of obsolete.
Commits on Jan 24, 2011
  1. Merge branch 'upstream'

    jgoerzen committed Jan 24, 2011
  2. Bump version number

    jgoerzen committed Jan 24, 2011
  3. Merge branch 'upstream'

    jgoerzen committed Jan 24, 2011
Commits on Jan 23, 2011
  1. Improve error message handling

    lpsmith committed Jan 23, 2011
    It would be nice to get a PG_DIAG_SQLSTATE-style error code in the case when
    the result pointer is null,  but how to accomplish this with libpq is not
    obvious, if it's even possible at all.
Commits on Jan 21, 2011
Commits on Jan 10, 2011
  1. Merge branch 'upstream'

    jgoerzen committed Jan 10, 2011
  2. Attempt to deal with \0 in ByteStrings going to the DB

    jgoerzen committed Jan 10, 2011
    to help handle binary data.
    
    issue #1
Commits on Dec 20, 2010
  1. Prepping 2.2.3.2

    jgoerzen committed Dec 20, 2010
Commits on Sep 16, 2010
  1. Merge branch 'upstream'

    jgoerzen committed Sep 16, 2010
  2. Mark the FFI calls pqconnectdb, pqexecParams and pqexec as 'safe' to …

    drpowell committed with jgoerzen Sep 12, 2010
    …allow concurrent access to the database.
    
    Previously all FFI calls were marked 'unsafe' which meant any long running db
    query would "stop the world".  That is, the whole haskell runtime system is
    blocked until the database query completes.  This is undesirable in many
    multi-threaded programs, and especially so in "server" programs where it is
    also necessary to be able to have multiple active database queries.
    
    - Added an MVar to the Conn type.  This is used to ensure each Connection is
      only used with one FFI at a time.  Multiple database queries may be performed
      using separate Connections.
    
    - In the case where the local libpq is not compiled to be thread-safe, issue a
      warning to stderr, and use a global MVar to allow only a single FFI call to
      be active at once.
Commits on Sep 12, 2010
  1. Mark the FFI calls pqconnectdb, pqexecParams and pqexec as 'safe' to …

    drpowell committed Sep 12, 2010
    …allow concurrent access to the database.
    
    Previously all FFI calls were marked 'unsafe' which meant any long running db
    query would "stop the world".  That is, the whole haskell runtime system is
    blocked until the database query completes.  This is undesirable in many
    multi-threaded programs, and especially so in "server" programs where it is
    also necessary to be able to have multiple active database queries.
    
    - Added an MVar to the Conn type.  This is used to ensure each Connection is
      only used with one FFI at a time.  Multiple database queries may be performed
      using separate Connections.
    
    - In the case where the local libpq is not compiled to be thread-safe, issue a
      warning to stderr, and use a global MVar to allow only a single FFI call to
      be active at once.
Commits on Apr 24, 2010
  1. Merge branch 'upstream'

    jgoerzen committed Apr 24, 2010
  2. Tightening deps for new tests

    jgoerzen committed Apr 24, 2010
Commits on Apr 16, 2010
  1. Relax the system_schema exclusion

    jgoerzen committed Apr 16, 2010
    From: MightyByte (Doug Beardsley)
    The database I was using for
    reference had system_schema tables that made me think they were
    Postgres-specific, but really were a part of the database.  So the
    last AND clause for "system_schema" should be removed.  It looks like
    the "pg_catalog" and "information_schema" schemas really are for
    internal Postgres stuff, but you might want to double check me to make
    sure.
Commits on Feb 23, 2010
  1. update changelog

    jgoerzen committed Feb 23, 2010
  2. Merge branch 'upstream'

    jgoerzen committed Feb 23, 2010
  3. Update cabal version spec for GHC 6.12

    jgoerzen committed Feb 23, 2010
    Fixes #201
    
    GHC 6.12 introduced an incompatibility; Setup.hs can now work with it
    or with 6.10 and earlier, and not with both.  Updating the .cabal file
    to note the Cabal version required.
Commits on Feb 22, 2010
  1. Add closes: #570880 to changelog.

    Marco Túlio Gontijo e Silva committed Feb 22, 2010
Commits on Feb 17, 2010
  1. Update changelog

    jgoerzen committed Feb 17, 2010
  2. Updating control for GHC 6.12

    jgoerzen committed Feb 17, 2010
  3. Merge branch 'upstream'

    jgoerzen committed Feb 17, 2010
  4. Bump version to 2.2.3.0

    jgoerzen committed Feb 17, 2010
  5. Merge branch 'upstream'

    jgoerzen committed Feb 17, 2010