Permalink
Switch branches/tags
Commits on Nov 8, 2010
  1. Revert "Fix test_io's test_sh_used_appropriately to use a more consis…

    headius committed Nov 8, 2010
    …tent ps output (cures failure on Linux)."
    
    This reverts commit 8b9bba7.
  2. Revert "One more fix to get test_io sh test working with a headless m…

    headius committed Nov 8, 2010
    …achine (ps -a only shows processes with a controlling terminal)"
    
    This reverts commit 80ab0e9.
  3. Bump for 1.5.4

    enebo committed Nov 8, 2010
Commits on Nov 6, 2010
  1. Fix for JRUBY-5122: Webrick socket.readline causes 100% cpu usage

    nahi committed Nov 6, 2010
    Do not set NONBLOCK bit to sockets accepted by WEBrick server to avoid
    busy loop.  Busy loop bug in RubyIO caused by NONBLOCK bit is fixed in
    master but it's rather big change for stable branch.  So we decided to
    just remove NONBLOCK bit at jruby-1_5 as a one-off change for this
    branch.
    
    As far as I understand, NONBLOCK bit is added to avoid interpreter block
    by write system call.([ruby-dev:26477])  Ruby level IO#write is
    originally a blocking call but IO#write of CRuby 1.8 might block
    *interpreter* by calling write system call. With NONBLOCK bit, io_fwrite
    in io.c does retry until the whole given buffer is written.
    
    In JRuby, which runs native threads, it won't block by write so we don't
    need to set NONBLOCK bit here.
    
    This change might affect to applications which expect sockets accepted
    by WEBrick to be non-blocking.  Almost all IO methods for reading are
    blocking method so it won't happen I believe.
Commits on Nov 5, 2010
  1. Merge --profile support to JRuby 1.5.

    headius committed Nov 5, 2010
    Squashed commit of the following hashes (master) with 1.5-appropriate cleanups:
    
    (in reverse order of application, i.e. first commit is at the bottom)
    65cf879
    4ddda10
    ae16db0
    75528e2
    96dacb9
    e0459e8
    6b9842b
    26be5ea
    8e299ce
  2. One more fix to get test_io sh test working with a headless machine (…

    headius committed Nov 5, 2010
    …ps -a only shows processes with a controlling terminal)
  3. Fix test_io's test_sh_used_appropriately to use a more consistent ps …

    headius committed Nov 5, 2010
    …output (cures failure on Linux).
Commits on Oct 30, 2010
  1. Test case for JRUBY-5110.

    BanzaiMan committed Oct 30, 2010
Commits on Oct 29, 2010
  1. Use the appropriate lookup method for finding ENV["PATH"] on all plat…

    BanzaiMan committed Oct 29, 2010
    …forms. This should fix JRUBY-5110 completely. (Let the third time be a charm!)
Commits on Oct 27, 2010
  1. Fix JRUBY-4766: java.lang.VerifyError: org/jruby/ext/posix/LinuxLibC$…

    BanzaiMan committed with headius Oct 13, 2010
    …jaffl$0
    
    Updated jaffl.jar to 0.5.4.
Commits on Oct 23, 2010
  1. Cucumber Japanese example raises exception on JRuby

    Naoto "Kevin" IMAI TOYODA committed with yokolet Oct 20, 2010
    It is caused by ByteArrayLexerSource's bug:
    ByteArrayLexerSource#read() must return non-negative value (like 0xe3)
    except when detecting EOF, but it can returns negative value (like
    0xffffffe3) which causes EncodingException.
    
    Resolution: mask ByteArrayLexerSource#read() with 0xff except EOF.
    
    Signed-off-by: Yoko Harada <yokolet@gmail.com>
Commits on Oct 20, 2010
  1. Fix for JRUBY-5132: java.awt.Component.instance_of?() expects 2 args

    headius committed Oct 20, 2010
    Added instance_of? as a reserved name, so it's not overwritten by Java methods.
  2. JRUBY-5064: ChannelStream#read() should return an unsigned value.

    nahi committed with headius Sep 7, 2010
    For InputStream compatibility. Reading int from buffered bytes works but
    reading from unbeffered bytes did not work.
  3. JRUBY-5123: Fixed 2 issues for unmarshalling TYPE_IVAR data.

    nahi committed with headius Oct 6, 2010
    In Ruby marshal format, TYPE_IVAR('I') is used when;
    1) any ivar is added to built-in object (String, etc), or
    2) ivar is added by Class#_dump (old API for customized marshal data)
       for serializing ivars.
    JRuby < 1.6 could not handle TYPE_IVAR correctly when it's used
    recursively. Fixed by introducing MarshalState for recursive state mgmt.
Commits on Oct 12, 2010
  1. Redo the fix of JRUBY-5110 a little more carefully.

    BanzaiMan committed Oct 12, 2010
    It turns out that, on Windows, ENV_JAVA should be case-sensitive, while ENV is not. They are both case-sensitive on all other platforms. So, subclass org.jruby.RubyGlobal.StringOnlyRubyHash for case-insensitive lookup, and instantiate ENV as an instance of that class. ENV_JAVA remains the same.
Commits on Oct 7, 2010
  1. Fix JRUBY-5110: kernel.system not working in 1.5.3 on Windows

    BanzaiMan committed Oct 1, 2010
    See also: JRUBY-5098, JRUBY-4834.
Commits on Sep 27, 2010
  1. JRUBY-5011 (jrubyc --java): 'cannot find symbol' if function def has …

    enebo committed Sep 27, 2010
    …more than 3 arguments
  2. JRUBY-4937: Add RubyFile.getUmaskSafe() to cache/sync and avoid troun…

    dekellum committed with headius Jul 15, 2010
    …cing umask
    
    Signed-off-by: Hiro Asari <asari.ruby@gmail.com>
  3. Bump versions to 1.5.3

    enebo committed Sep 27, 2010
Commits on Sep 25, 2010
Commits on Sep 21, 2010
Commits on Sep 18, 2010
  1. Fix for JRUBY-4944

    yokolet committed Sep 18, 2010
Commits on Sep 17, 2010
  1. Fix JRUBY-4948.

    yokolet committed Sep 17, 2010
  2. Add a fix for JRUBY-5076

    olabini committed with nahi Sep 13, 2010
Commits on Sep 16, 2010
  1. Fix for JRUBY-5093:

    headius committed Sep 16, 2010
    * Re-poll for thread events before locking the ThreadLibrary-related lock. This helps prevent (but does not guarantee) any timeout event from getting "lost" between a poll and the eventual locking.
    * Move RubyThread.interrupt out of the RubyThread.receiveMail synchronized section, since it doesn't need synchronization and it may need to acquire a lock to notify the currentWaitObject. Locking against RubyThread and then currentWaitObject lead to deadlocks in our builds (and probably in real code too).
  2. Hack to work around Scala shipping classes that break Class.getDeclar…

    headius committed Sep 16, 2010
    …edClasses. JRUBY-5092.