Commits on Sep 26, 2011
Commits on Sep 22, 2011
  1. squelch easy temp bugs

    committed Sep 22, 2011
Commits on Sep 21, 2011
  1. comment on cause for error

    committed Sep 21, 2011
  2. handle :keep flag on argagg

    committed Sep 21, 2011
Commits on Sep 16, 2011
Commits on Sep 15, 2011
  1. remove debug statements

    committed Sep 15, 2011
Commits on Sep 13, 2011
  1. checkpoint

    committed Sep 13, 2011
Commits on Aug 6, 2011
Commits on Aug 5, 2011
  1. checkpoint

    committed Aug 5, 2011
Commits on Jul 27, 2011
Commits on Jul 15, 2011
  1. In budplot, escape graphviz node names to avoid syntax errors.

    If we don't escape graphviz node names, budplot produces an error on programs
    that contain graphviz reserved words (e.g., "node"). Fixes #88.
    neilconway committed Jul 15, 2011
  2. Tweak comment for map2pro.

    neilconway committed Jul 15, 2011
  3. Avoid use of "dump" as a collection name.

    Apparently it is monkeypatched by something: "dump" is fine when the
    tc_mapvariants test is run by itself, but fails when the entire ts_bud test
    suite is run.
    neilconway committed Jul 15, 2011
  4. Monkeypatch Enumerable#pro as an alias for map.

    The map => pro rewrite is overly eager: it rewrites some cases in which map is
    applied to a Ruby Enumerable that is not a BudCollection. It would be possible
    to fix this by making the rewrite more accurate, but it seems hard to guarantee
    accuracy without being overly conservative. Instead, define Enumerable#pro as an
    alias for map.
    Fixes #225.
    neilconway committed Jul 15, 2011
Commits on Jul 11, 2011
Commits on Jul 8, 2011
  1. Add a serializer test case.

    neilconway committed Jul 8, 2011
Commits on Jul 7, 2011
  1. Clarify argmin and argmax behavior in case of ties.

    argmin and argmax might produce multiple tuples. That is:
        result <= t.argmin([], :a)
    If multiple tuples contain the minimal value of `a`, argmin will produce both
    result tuples. Document this behavior, and add a test for it. (The only real
    alternative would be to produce an error in this case -- that might also be
    worth considering, but it seems less flexible than the current behavior.)
    neilconway committed Jul 7, 2011
  2. Tweak for docs.

    neilconway committed Jul 7, 2011