
From v0.4 to v0.5 (in progress):
-----------------

* Robert J Ohannessian:
 - More makefile fixes.
 - Added an 16bpp MMX and C color fader.
 - Added an 15bpp MMX and C color fader.
 - Added an 32bpp MMX and C color fader.
 - Added a new example (exfade) to demonstrate fading effect.
 - Added a 16bpp SSE, MMX and C sub blender.
 - Added a 15bpp SSE, MMX and C sub blender.
 - Added a 32bpp SSE, MMX and C sub blender.
 - Added a new example (exsub) to demonstrate color subtraction effect.
 - Fixed bugs that prevented compiling on DJGPP and Linux


From v0.2 to v0.4:
-----------------

* Robert J Ohannessian:
 - Fixed a potential bug in the examples. Thanks to Vincent Penquerc'h
   for pointing it out!
 - Fixes to the makefile's MMX and SSE tests.
 - Removed SSE code in MMX blenders (left there by mistake)
 - Added a 32bpp SSE, MMX and C add blender.
 - Added a 32bpp SSE, MMX and C trans blender.
 - Added a 16bpp SSE, MMX and C primitive add blender.
 - Added a 15bpp SSE, MMX and C primitive add blender.
 - Added a 32bpp SSE, MMX and C primitive add blender.
 - Added a 16bpp SSE, MMX and C primitive trans blender.
 - Added a 15bpp SSE, MMX and C primitive trans blender.
 - Added a 32bpp SSE, MMX and C primitive trans blender.
 - Added option to set depth of examples from a config file
 - Added an MMX and SSE 2x stretcher. The C code falls back to Allegro.
 - Added a DJGPP makefile
 - Various bugs fixed.
 - Makefiles now have an uninstall target.
 - Makefiles are also more expressive as to the status of compilation.


From v0.1 to v0.2:
-----------------

* Robert J Ohannessian:
 - Added 15bpp supoprt do fblend_trans() and fblend_add()
 - Optimized the common cases for both functions: 100% add and 50% trans.
 - Changed API slightly.
 - Made the trans function revert to masked_blit if factor is set to max.
 - More speedups
 - Trans code supports blocks of any widths
 
* Peter Wang:
 - Added Unix makefile
 - Fixed some assembly code which broke under Unix.
