Permalink
Commits on Jul 9, 2013
  1. @enebo

    HEH when both type and edgeType are not null we must not use object i…

    …dentity since we use strings
    enebo committed Jul 9, 2013
  2. @enebo

    Third time is a charm

    enebo committed Jul 9, 2013
  3. @enebo
  4. @enebo

    Simplify logic a little bit

    enebo committed Jul 9, 2013
  5. @mkristian

    remove native check since it has no more references - fixes also the …

    …problem with missing build_lib
    mkristian committed Jul 9, 2013
  6. @enebo

    Merge pull request #852 from prathamesh-sonpatki/vertext_specs

    Fixed a bug in checking the type of edges in EdgeTypeIterator
    enebo committed Jul 9, 2013
  7. Fixed a bug in checking the type of edges in EdgeTypeIterator

    - While checking for type of edges, equality is not sufficient as we
      are comparing two objects
    - Added specs for this change
    committed Jul 3, 2013
  8. @mkristian
  9. @mkristian
  10. @headius

    Temporarily rename problematic remove method to removeFrom.

    Final decision about how to fix this still remains with @yokolet
    
    See #858.
    headius committed Jul 9, 2013
  11. @headius

    Add threads 1-by-1 to ThreadGroup.list, to avoid adding a null.

    Because RubyThread objects are held in a weak set inside
    ThreadGroup, blindly copying them into an array *might* cause us
    to insert a null somewhere. This new logic at least ensures that
    does not happen, though it may not fix @halorgium's issue reported
    on IRC today.
    headius committed Jul 9, 2013
  12. @headius

    Enable incremental compilation.

    Don't look at me like that.
    headius committed Jul 9, 2013
  13. @headius

    Merge pull request #864 from jendave/maven-reporting

    Move Maven site reporting to parent pom. Fixed parent hierarchy.
    headius committed Jul 9, 2013
  14. @atambo
  15. @atambo
  16. @jendave
Commits on Jul 8, 2013
  1. @mkristian

    Revert "use compiler plugin 2.5.x to have proper incremental compilat…

    …ion"
    
    just breaks travis :(
    
    This reverts commit 3f93c50.
    mkristian committed Jul 8, 2013
  2. @mkristian
  3. @mkristian
  4. @headius
  5. @eregon

    always map ENOSYS to NotImplementedError

    * since this is basically what it means, but the headers still define the function
    * this is to have similar feature to configure MRI tests and rb_f_notimplement
    eregon committed Jul 8, 2013
  6. @eregon
  7. @mkristian
  8. @mkristian
  9. @mkristian
  10. @subbuss

    [IR] Updated keyword args implementation -- matches MRI behavior

    * Updated ReceiveArg cleanup and keyword args implementation.
    
    * But on a complex set of tests, IR keyword args behavior now
      matches MRI fully.
    
      See https://gist.github.com/subbuss/5945677
    
    * This is now mostly finished except for check_arity instruction
      that needs updating.
    subbuss committed Jul 8, 2013
  11. @subbuss

    [IR] First pass cleanup of ReceiveArg* instructions

    * Removed specialized 1.8 and 1.9 ReceiveOptArg and ReceiveRestArg
      instructions -- it is possible to use the same version for both
      modes. In addition, the same versions are usable in 2.0 mode which
      eliminates more specialized versions.
    
    * TODO: Fix up keyword arg instructions and interpreter to pass
      used keyword arg information into the ReceiveArg* interpretation.
    subbuss committed Jul 7, 2013
Commits on Jul 7, 2013
  1. @headius
  2. @headius
  3. @eregon
  4. @eregon
  5. @headius
  6. @headius
  7. @headius