Commits on Mar 3, 2017
Commits on Feb 14, 2017
  1. Fix an issue with deepseq on the builtin RUN process.

    References #15.
    committed Feb 14, 2017
Commits on Dec 20, 2016
Commits on Dec 19, 2016
Commits on Oct 8, 2016
  1. Fix a panic where extensions was returning the incorrect set.

    This was only triggered in very odd cases when dot-values were heavily
    committed Oct 8, 2016
Commits on May 19, 2016
  1. Merge pull request #13 from sashabu/master

    Change ProfilerThunks.hs to use Template Haskell to make it easier to…
    committed May 19, 2016
  2. Change ProfilerThunks.hs to use Template Haskell to make it easier to…

    … increase the number of thunks.
    sashabu committed May 19, 2016
Commits on Mar 30, 2016
  1. Fix scripts that were erroneously passing the type checker.

    This fixes a problem with patterns such as x.y <- e passing the type
    checker where the type of e is less general than the inferred type of
    x.y. For example, x.y was previously being inferred as of type a.b but
    e could be of type Event.
    This is fixed by treating such patterns in the same way as we treat
    function calls.
    committed Mar 30, 2016
Commits on Feb 24, 2016
  1. If file contents are provided, don't use the real disk store.

    Previously, if a file was not found in the provided file map, we would
    fallback and use the actual disk. This was not the intended behaviour.
    committed Feb 24, 2016
Commits on Feb 16, 2016
  1. Fix a bug with the optimised Dot applier.

    This was erroneously optimising polymorphic functions, which clearly it
    cannot do since it does not know how these values will be instantiated
    at runtime.
    committed Feb 16, 2016
Commits on Dec 17, 2015
  1. Fix type checking of mutually recursive datatypes.

    This could sometimes erroneously fail if the types were not comparable
    for equality. This is because of the way we checked if the datatypes
    were comparable for equality -- whenever type checking a single
    definition, we would speculatively mark it as comparable for equality,
    and then backtrack if necessary. However, this would not work for
    mutually recursive datatypes, and we would always fail to identify them
    as comparable for equality.
    This commit modifies this to mark all mutually-recursive datatypes as
    comparable for equality, and then backtracks en-masse if this fails.
    committed Dec 17, 2015
Commits on Jul 21, 2015
  1. Allow variants on outputs for tests.

    This should fix the long running failure we have seen.
    committed Jul 21, 2015
Commits on Jul 20, 2015
  1. Remove a missing function.

    committed Jul 20, 2015
  2. Remove trimming support.

    After the recent changes to the evaluator trimming values to remove
    functions and processes actually slows down evaluation.
    committed Jul 20, 2015
  3. Remove unused function.

    committed Jul 20, 2015
Commits on Jun 28, 2015
  1. Remove some import warnings.

    committed Jun 28, 2015
  2. Attempt to fix Travis build.

    committed Jun 28, 2015
  3. Revert "Make the implementation of / and % match FDR2."

    This reverts commit 32601b8.
    committed Jun 28, 2015
Commits on Jun 27, 2015
  1. Quote the GHC version.

    committed Jun 27, 2015
  2. Bump GHC version.

    committed Jun 27, 2015
Commits on Jun 24, 2015
  1. Merge pull request #11 from sashabu/master

    Fix some typos.
    committed Jun 24, 2015
Commits on Jun 23, 2015
  1. Fix some typos.

    sashabu committed Jun 23, 2015
Commits on Jun 12, 2015
  1. Fix a test.

    I think this is now correct.
    committed Jun 12, 2015
Commits on Jun 11, 2015
  1. Updates to support GHC 7.10.

    committed Jun 11, 2015
Commits on Jun 10, 2015
  1. Use a single boolean true/false value.

    This reduces memory usage on extremely large scripts, and improves
    performance moderately.
    committed Jun 10, 2015
  2. Ban double patterns within prefix expressions.

    They make it incredibly difficult to compute the set of values that
    are allowed.
    committed Jun 10, 2015