Skip to content


Subversion checkout URL

You can clone with
Download ZIP
tag: rel-1.0b2
Commits on Apr 21, 2011
  1. Adding META.json spec file

  2. Bump to next release number

Commits on Apr 8, 2011
  1. Bump revision to 1.0b1

Commits on Apr 7, 2011
  1. Uninstall the operator families

    They are created implicitly and don't depend on the types, so cascade doesn't
    propagate to them.
  2. Names for the operator classes fixed

    They don't need to be distinguished for types, and the builtin ones aren't.
  3. mpz hash compatible with int builtins, mpq hash compatible with mpz

    This allows cross-type hash joins of these types together.
    Note that cross-type hash join can't be used yet as we need to set up
    an operator family. However an eventual hash index created now wouldn't
    be broken if we want to add this feature later.
Commits on Apr 6, 2011
  1. README rewording

    Preparing for release.
  2. Copy the css in the _static dir if needed

    The pgxs makefile installs all the files into the same directory.
    Put the css back where Sphinx expects.
  3. Added AggCheckCallContext to make the extension work on PG 8.4

    PG 8.3 is missing other functions. Because current Ubuntu installs 8.4,
    supporting this version is already a good step.
  4. Added hash support for mpz and mpq types

    Note that marking HASHES the eq operator was a mistake: the planner
    tried to use the operator for hash aggregates and failed because no
    hash function was available for it.
Commits on Apr 5, 2011
  1. Documentation cleanup

    - better spacing in lists
    - better line height in samples
    - dropped empty definition lists in multiple-signature functions
    - homepage tweaks
  2. Added installation documentation

    README cleaned up with up-to-date information.
Commits on Apr 4, 2011
  1. Dropped version in the setup script name for PG before 9.1

    This drops the need to handle the version number in any file and simplifies
    the test suite setup.
Commits on Apr 3, 2011
Commits on Apr 2, 2011
  1. Improved cast of mpz to int8

    INT64_MIN is correctly converted. Fixed test for other casts: we were
    thinking that the conversion to INT??_MIN was not allowed: the tests were
    wrong instead (the unary minus is applied *after* the cast).
Commits on Mar 31, 2011
  1. Ignoring other items in the working dir

    tags file, regression database, website images.
  2. Test functions dropped

    Now that we have random number we can do better.
    Recalculated results for the sum of integer with a regular interval.
  3. Added/updated benchmark results

    Note that for some reason the SumTable benchmark has gone more than twice
    faster than the previous run. Don't know what is the difference: probably
    the first run was with PostgreSQL from package manager and tne second with
    the one compiled myself. Will try to investigate more.
Commits on Mar 30, 2011
  1. Emit the test labels in the result file

    This allows to generate variable labels.
  2. Benchmark scripts executable

Something went wrong with that request. Please try again.