Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Commits on Dec 15, 2012
  1. @TooTallNate

    v1.2.1

    TooTallNate authored
  2. @TooTallNate
Commits on Dec 11, 2012
  1. @TooTallNate

    libffi: define the 'target_arch' default variable at the top

    TooTallNate authored
    Otherwise an error gets thrown when trying to build standalone on Windows.
    
    Reported on the libffi mailing list:
      http://sourceware.org/ml/libffi-discuss/2012/msg00291.html
Commits on Dec 10, 2012
  1. @TooTallNate

    add .jshintrc file

    TooTallNate authored
Commits on Oct 13, 2012
  1. @TooTallNate

    v1.2.0

    TooTallNate authored
Commits on Oct 5, 2012
  1. @TooTallNate
  2. @TooTallNate
  3. @TooTallNate
  4. @TooTallNate
  5. @TooTallNate

    callback_info: return after throwing an exception to JS

    TooTallNate authored
    You're not supposed to do anything after throwing...
  6. @TooTallNate

    test: make the async cb call invoke it's own copy of the function poi…

    TooTallNate authored
    …nter
    
    Not really sure if this helps anything, but it seemed like a good idea...
  7. @TooTallNate
  8. @TooTallNate
Commits on Oct 4, 2012
  1. @TooTallNate
  2. @TooTallNate

    ffi: the dlsym() constants need to be Buffer instances

    TooTallNate authored
    Since internally they are a "void *" type, and these get passed as the "handle"
    argument to dlsym().
  3. @TooTallNate
  4. @TooTallNate
  5. @TooTallNate

    dynamic_library: use RTLD_LAZY by default

    TooTallNate authored
    You can still override it manually, but this seems to be the recommended
    default mode. From OS X's `man dlopen`:
    
    "RTLD_LAZY is normally preferred, for reasons of efficiency.  However,
    RTLD_NOW is useful to ensure that any undefined symbols are discovered during
    the call to dlopen()."
    
    This should at least prevent the "undefined symbol" errors in #80.
Commits on Oct 2, 2012
  1. @TooTallNate
  2. @TooTallNate
  3. @TooTallNate

    function: finish up the "function" type for now

    TooTallNate authored
    This doesn't support things like variadic functions, but let's cross that bridge
    when we get there...
  4. @TooTallNate

    callback: remove "POINTER_SIZE" variable

    TooTallNate authored
    it's stupid...
Commits on Oct 1, 2012
  1. @TooTallNate

    libffi: fix unused variable warnings

    TooTallNate authored
    ../deps/libffi/src/dlmalloc.c:3391:30: warning: expression result unused [-Wunused-value]
      set_segment_flags(&m->seg, mmapped);
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~
  2. @TooTallNate
Commits on Sep 29, 2012
  1. @TooTallNate

    test: put the sync callback GC test in its own test

    TooTallNate authored
    Rather than the "async" one
Commits on Sep 28, 2012
  1. @TooTallNate

    don't call the "ref()" function on passed in arguments

    TooTallNate authored
    This is a violation of the "type" interface as it bypasses the "set()"
    function of the "type" in most cases.
  2. @TooTallNate

    DynamicLibrary: remove manual string to buffer conversion

    TooTallNate authored
    The types on these functions has been changed to "string", so this is no
    longer necessary...
  3. @TooTallNate
  4. @TooTallNate
  5. @TooTallNate
  6. @TooTallNate
  7. @TooTallNate

    test: remove printf call

    TooTallNate authored
  8. @TooTallNate
  9. @TooTallNate
  10. @TooTallNate

    type: rename the "t" variable to "ffi_type"

    TooTallNate authored
    just a better name...
Something went wrong with that request. Please try again.