Switch branches/tags
Commits on Mar 4, 2012
  1. Assert stack_ after status_

    ryoqun committed Mar 4, 2012
Commits on Mar 3, 2012
  1. Fix FFI alignment

    dbussink committed Mar 3, 2012
    This needs to align by the type, not the complete element size in
    case of an array of elements of a certain type.
    Fixes rubinius#1579
  2. Updated CI tags.

    brixen committed Mar 3, 2012
  3. Fixed Module#name.

    brixen committed Feb 24, 2012
    OMG Module#name. We have moved a lot of critical functionality into
    Rubinius::Type to avoid issues with people undef'ing methods on core
    library classes. We need to create a proper Mirrors implementation.
    So some of this commit removes helper methods and under, under protective
    methods for Module#name. Further, we need the name at various points in
    the VM/C-API, so the 1.8/1.9 logic for "" vs nil for an anonymous module's
    name is now in the primitive.
  4. Rewrote Module language specs.

    brixen committed Feb 23, 2012
Commits on Mar 2, 2012
Commits on Mar 1, 2012
  1. Fix instance_eval specs

    jfirebaugh committed Mar 1, 2012
    instance_eval has always yielded the object. The behavior
    difference between Ruby versions was the return value: in
    1.9.1, it returned nil rather than the result of the block.
  2. Merge pull request rubinius#1577 from ryoqun/llvm-shutdown

    dbussink committed Mar 1, 2012
    Call llvm_shutdown()
  3. Site regen'd

    argent-smith committed Feb 10, 2012
Commits on Feb 29, 2012
  1. Use MurmurHash 3 for String hashing

    dbussink committed Feb 27, 2012
    This gives us a fast hashing algorithm with good distribution.
    It also allows for using a seed value to randomize the hash
    output so it mitigates hash collision DoS attacks.
  2. Indent scoped code of main()

    ryoqun committed Feb 29, 2012
Commits on Feb 28, 2012
  1. Fixed rubinius#1578

    hosiawak committed Feb 28, 2012
  2. Remove blank line

    ryoqun committed Feb 28, 2012
Commits on Feb 27, 2012
  1. Don't use Fiber locals if we're not trying to change the current thread

    dbussink committed Feb 27, 2012
    Celluloid actually uses this by setting up locals in another
    thread that it just started and thus exposed this issue.
  2. Add specs for setting thread locals from another thread

    dbussink committed Feb 27, 2012
    When we set a thread local from another thread, we should
    not look at whether we're running a fiber or not, but just
    set the thread local.
Commits on Feb 26, 2012
  1. Move random number generator from MRI into separate files

    dbussink committed Feb 26, 2012
    This moves the implementation taken from MRI into it's
    own files, so it isn't mixed with the Rubinius Randomizer
    object and thus defines a cleaner api between the two.
Commits on Feb 25, 2012