Permalink
Commits on Jun 3, 2013
  1. Handle signed size parameter of J# constructor correctly

    This fixes bug #68.
    committed Jun 3, 2013
Commits on May 2, 2013
  1. Bump cabal version to 1.2.0.8.

    atomb committed May 2, 2013
Commits on May 1, 2013
  1. Fix linking with the sse2 flag unset.

    The patch adding the sse2 flag allowed hashable to build on systems
    without the appropriate SSE symbols available, but didn't allow linking,
    as the resulting code still referenced one of the internal symbols from
    the SSE2 implementation. This patch fixes that.
    atomb committed May 1, 2013
Commits on Apr 29, 2013
  1. Bump version number to 1.2.0.7

    tibbe committed Apr 29, 2013
Commits on Apr 24, 2013
Commits on Apr 21, 2013
  1. Bump version number to 1.2.0.6

    committed Apr 21, 2013
Commits on Apr 18, 2013
  1. Add a regression test for the SSE2 bug (gh-60) (gh-60)

    --HG--
    rename : tests/Properties.hs => tests/Main.hs
    committed Apr 18, 2013
Commits on Jan 11, 2013
  1. Bump version number to 1.2.0.5

    tibbe committed Jan 11, 2013
  2. Yet more portability goo

    committed Jan 11, 2013
  3. Drop some trailing whitespace that crept in

    --HG--
    extra : rebase_source : b58566d6984d7c66e66aaebfee571ae677019739
    committed Jan 11, 2013
  4. SSE: force the stack to 16-byte alignment everywhere.

    This fixes gh-52.
    
    --HG--
    extra : rebase_source : b503958047fd9a91d87068f611ea9fb93c60178f
    committed Jan 11, 2013
  5. Disable Haskell SipHash code, for portability

    I never ported it to big-endian systems, and it prevents the library
    from working on PPC.
    
    This partly addresses gh-47.
    committed Jan 11, 2013
  6. When building with -ffixed-salt, don't pull in any PRNG code

    This is necessary to get us working with GHCi 7.6.1 on Windows x64.
    
    See gh-46 and http://hackage.haskell.org/trac/ghc/ticket/7568 for the
    unhappy details.
    committed Jan 11, 2013
Commits on Jan 10, 2013
  1. Use a non-exported class method for safe, reliable String hashing

    Previously, we had used rewrite rules, which was unreliable: they
    would only fire if a client library was compiled with optimisation,
    an unsafe assumption to make.
    committed Jan 10, 2013
  2. Merge the SSE build fix

    committed Jan 10, 2013
  3. Help gcc to find some static loop unrolling opportunities

    (I was surprised it wasn't smart enough already!)
    
    --HG--
    extra : rebase_source : 47e57aecc86eda911460d7cf260025dd61135019
    committed Jan 10, 2013
  4. SSE2: use unaligned intrinsics for loads

    This seems to work around what I believe is a nasty bug in ghci's runtime
    linker, in which it does not honour data alignment requirements.
    
    --HG--
    extra : rebase_source : 9607086ed0da4f9d25d4d6641c0f08af9018ecee
    committed Jan 10, 2013
Commits on Jan 9, 2013
  1. Bump version number to 1.2.0.4

    tibbe committed Jan 9, 2013
Commits on Jan 8, 2013
  1. Update comment to match code

    tibbe committed Jan 8, 2013
  2. Work around ghci runtime linker being broken

    It never calls static initializers.
    
    This fixes gh-52.
    committed Jan 8, 2013
Commits on Jan 5, 2013
Commits on Dec 18, 2012
Commits on Dec 17, 2012
Commits on Dec 14, 2012
  1. Bump version number to 1.2.0.2

    tibbe committed Dec 14, 2012
  2. Merge

    committed Dec 14, 2012
  3. Bump version number to 1.2.0.1

    tibbe committed Dec 14, 2012
  4. Forgot an "else" when choosing which SSE variant to use

    This had no effect on correctness, just performance.
    committed Dec 14, 2012
Commits on Dec 13, 2012