Permalink
Commits on Jun 25, 2012
  1. @headius
Commits on Jun 22, 2012
  1. @enebo
  2. @subbuss

    HEADS UP! Removed all trace of setWithinDefined and isWithinDefined.

    * Tested on rubyspecs in 1.8 and 1.9 mode -- AST, JIT, IR.  No
      change in failures after this change.
    
    * Removed this from AST interpreter, JIT, IR
    
    * setWithinDefined was set by all of these before entering defined?
      checks and cleared on exit.  isWithinDefined was only used by
      RaiseException to decide whether to set $! or not -- effectively
      $! was not being set for exceptions raised by code executing
      by the call tree rooted in the defined? checks.  However, this
      resulted in buggy behavior because this call tree could do a ton
      of work (including loading files, requires, setting up classes,
      etc.) any of which can raise exceptions and might have to be
      handled, which require access to $!.  The workaround in IR and
      AST interpreters was to set "$!" whenever an exception was
      received (effectively bypassing the condition that RaiseException
      was guarding).  But, since I removed both these hacky checks as
      part of a previous commit (20632af),
      we had code failures in Rails (bug JRUBY-6705) because of files
      not getting loaded properly. So, given that isWithinDefined is only
      used by RaiseException, which was anyway getting bypassed, there is
      no reason to keep that check around.  Once I removed that check,
      since there was no other use of isWithinDefined in the codebase,
      I also removed setWithinDefined, which led to a bunch of dead code
      which I purged.
    
    * NOTE: JIT is probably yet to remove the $! hack that I removed in
      the other commit.  It would be best to remove that hack as well
      so we know for sure that these two commits are working together
      well without bugs.
    subbuss committed Jun 22, 2012
Commits on Jun 20, 2012
  1. @headius

    Final fix to get 'ant test' passing with new Java 7 backquote.

    * Modify trace test setup to revert Object.method_added.
    headius committed Jun 20, 2012
  2. @headius
  3. @headius
Commits on Jun 17, 2012
  1. Remove stale cext JNI builds.

    Wayne Meissner committed Jun 17, 2012
Commits on Jun 16, 2012
  1. @headius
Commits on Jun 15, 2012
  1. First pass at enabling select() on both NIO and ENXIO channels at the…

    … same time. Probably contains subtle bugs, but headius can clean them up.
    Wayne Meissner committed Jun 15, 2012
  2. Always use the channel's selector provider to open a selector in Chan…

    …nelStream.ready()
    Wayne Meissner committed Jun 15, 2012
  3. Implement rb_eql()

    Wayne Meissner committed Jun 15, 2012
Commits on Jun 14, 2012
  1. @enebo
  2. @enebo

    date strftime micro-opt 2/2. Only one date format is usable per threa…

    …d so cache out formatter in thread context (+300% more past opt 1/2)
    enebo committed Jun 14, 2012
  3. @enebo

    date strftime micro-opt 1/2. Use specific formatting methods instead …

    …of Generic String.format (+30% gain on reasonable iso date format string)
    enebo committed Jun 14, 2012
  4. @enebo
  5. @enebo
Commits on Jun 13, 2012
  1. @enebo

    Errant printf removed

    enebo committed Jun 13, 2012
  2. @headius

    Fix JRUBY-6702

    File.utime does not check file permissions
    
    We were ignoring the return value from utimes. Added logic to
    check it and raise errno on -1.
    headius committed Jun 13, 2012
Commits on Jun 12, 2012
  1. @bbrowning @enebo

    Fix memory leak in FFI usage from multiple runtimes

    This fixes JRUBY-5053 and GitHub issue #203. Instead of storing the
    AbstractInvoker refmap in a static Map, the same data is now stored on
    an instance of the FFI class which holds all other per-runtime
    FFI-related data.
    bbrowning committed with enebo Jun 12, 2012
  2. @enebo

    bump versions for dev

    enebo committed Jun 12, 2012
  3. @headius

    Fix JRUBY-6706

    Collections.shuffle works incorreclty on Ruby arrays
    
    We were not implementing List.set properly; it should return the
    previous value at that index, and we were just returning the
    given value.
    headius committed Jun 12, 2012
  4. @headius
  5. @headius

    Update to jline-2.7.

    headius committed Jun 12, 2012
  6. @headius
  7. @headius

    Fix and test for #190

    Baffling that this got through, but I obviously botched the impl
    of __persistent__ and __persistent__=. They were set up as meta-
    class methods but using the instance method versions. So they
    received a RubyClass object but then the invoker has to cast to
    JavaProxy. Fixed by making them static and adjusting impl
    accordingly.
    headius committed Jun 3, 2012
  8. @jkutner @BanzaiMan

    upgraded idea project files to jdk 1.7

    Signed-off-by: Hiro Asari <asari.ruby@gmail.com>
    jkutner committed with BanzaiMan May 27, 2012
  9. @BanzaiMan

    warning: loading in progress, circular require considered harmful - r…

    …ubygems/spec_fetcher
    
    Signed-off-by: Hiro Asari <asari.ruby@gmail.com>
    committed with BanzaiMan Apr 29, 2012
  10. @alindeman @BanzaiMan

    Fixes JRUBY-6677

    * Passes the `mode` argument correctly from Kernel#open to IO#popen, so
      if a non-default mode is specified (e.g., "w"), it is honored
    
    Signed-off-by: Hiro Asari <asari.ruby@gmail.com>
    alindeman committed with BanzaiMan May 22, 2012
Commits on Jun 11, 2012
  1. @enebo
  2. @alindeman @BanzaiMan

    Fixes JRUBY-6697: --1.9 is default

    Signed-off-by: Hiro Asari <asari.ruby@gmail.com>
    alindeman committed with BanzaiMan Jun 2, 2012
  3. @BanzaiMan
  4. @BanzaiMan
Commits on Jun 8, 2012
  1. @headius

    jline 2.7 should be released soon, and at least before preview2.

    Revert "Remove JLine from being seen by maven and jarjar'ing our internal committed version to a non-conflicting package (to be reverted for next release when jline 2.7 is out)"
    
    This reverts commit 47d42eb.
    headius committed Jun 8, 2012
  2. @enebo