…the RGB24 format was just plain wrong for everything.
1) Making use of version 1.0 of Clutter/PyClutter if present. 2) Porting to the 2.6 LLVM API. 3) Moving over to using <4 x float> LLVM vectors throughout. This means we can finally drop the 'disable-mmx' option which is no-longer supported in LLVM 2.6.
…his. * Added an API to BufferSampler which allows for a zero-copy buffer assignment. This should be used with caution.
buffers. Update the test suite to test every byte-oriented buffer format we claim to support. This caused a bug-fix for the non-premultiplied ARGB and ABGR formats.
developmental. Odd patch numbers denote developmental releases, even denote actual releases.
…nt if the set is empty
… map reduce branch.
…to some locking in GLib thread pool implementation
…ows about reduce kernels.
…reeKernel in preparation for reduction kernel support
…ared by both the render and reduce engines.