Commits

Show all
Author Commit Message Labels Comments Date
Ronald Oussoren
Release 1.4 is imminent.
Tags
pyobjc-1.4
Branches
pyobjc-ancient
Ronald Oussoren
Remove news message about running Tiger binaries on Panther. I haven't done enough testing to claim this in the release notes.
Branches
pyobjc-ancient
Ronald Oussoren
Revert lame test
Branches
pyobjc-ancient
Ronald Oussoren
Reorganize initialization to ensure that the python type NSPoint is defined when we create a number of contant points (otherwise they'll be added as tuples)
Branches
pyobjc-ancient
Ronald Oussoren
Fix bug #1372474, a longstanding copy/paste problem
Branches
pyobjc-ancient
Ronald Oussoren
Some additional unittests that were added while looking for the OutlineEdit problem. All of them pass :-(
Branches
pyobjc-ancient
Ronald Oussoren
A minor thinko, Py_ARG_SIZE_T defitions in the python >=2.5 and others branches were switched.
Branches
pyobjc-ancient
Ronald Oussoren
Fix building of mpkgs (bdist_mpkg), bug due to restructing of source-deps
Branches
pyobjc-ancient
Ronald Oussoren
Rename the toplevel directories inside source-deps to avoid warnings with python2.5
Branches
pyobjc-ancient
Ronald Oussoren
Improve py2.5 compatibility layer.
Branches
pyobjc-ancient
Ronald Oussoren
Revert py2.5 related change to setup.py, the change didn't actually work correctly :-/
Branches
pyobjc-ancient
Ronald Oussoren
Python2.5 related cleanups
Branches
pyobjc-ancient
Ronald Oussoren
Python2.5 related cleanups
Branches
pyobjc-ancient
Ronald Oussoren
Python2.5 related cleanups
Branches
pyobjc-ancient
Ronald Oussoren
Default return was inside a switch statement instead of outside of it.
Branches
pyobjc-ancient
Ronald Oussoren
A fist cut at teaching PyObjC to do weaklinking, that is run on Panther even when build on Tiger.
Branches
pyobjc-ancient
Ronald Oussoren
Attempt to bridge more types
Branches
pyobjc-ancient
Ronald Oussoren
Recognize more toll-free bridged types
Branches
pyobjc-ancient
Ronald Oussoren
- Don't generate the static class-list, machines should be fast enough nowadays to no longer require this hack - When running 'python setup.py' test actually test the just build copy instead of the system copy.
Branches
pyobjc-ancient
Ronald Oussoren
Some more signatures have been annotated
Branches
pyobjc-ancient
Ronald Oussoren
Missing file from merger with ctype's copy of libffi
Branches
pyobjc-ancient
Ronald Oussoren
- Remove libffi code for platforms that we don't support, that is anything but x86* and ppc* - Partial merge with the copy of libffi that is used by ctypes. I haven't merged x86 support but kept our copy of that, the version of libffi used by ctypes is different enough from our copy that merging my darwin/x86 patches into that version of ffi.c would be more work that I want to do right now.
Branches
pyobjc-ancient
Ronald Oussoren
Remove an #error line that was introduced during debugging some time ago
Branches
pyobjc-ancient
Ronald Oussoren
Workaround for a buglet in pickle. This also ensures that only the relevant state gets placed into a pickle.
Branches
pyobjc-ancient
Ronald Oussoren
Make sure the AUTO_UNIVERSAL feature actually works with the current version of Xcode
Branches
pyobjc-ancient
Ronald Oussoren
Calls to class-methods from ObjC didn't work due to a missing registration call in objc_class.__new__.
Branches
pyobjc-ancient
Ronald Oussoren
Avoid #importing deprecated headers.
Branches
pyobjc-ancient
Ronald Oussoren
- Fix a number of py2.5 issues in test cases - disable osxcasts on py2.5, Carbon.CF is still broken - "Private" classes are no longer imported when using objc.loadBundle - "Private" methods are no longer visible in the class __dict__ (and hence not for introspection tools). "Private" names are names that start with an underscore. This removes a large fraction of undocumented API's from our view, making introspection more useable.
Branches
pyobjc-ancient
Ronald Oussoren
And the last one now passes too: the embedded alignment of large basic types (double, long double, ...) is 4 if they are the second or later element of a struct, except for long double whose alignment is always 16 (contrary to what is suggested by the ABI docs). This is on PPC.
Branches
pyobjc-ancient
Ronald Oussoren
Libffi tests now pass on powerpc, except for cls_align_longdouble (and all regular tests still pass to)
Branches
pyobjc-ancient
  1. Prev
  2. Next