This commit:
* Adds a 'compare_ff' function for absolute 'almost equal' comparison of floats.
* Makes 'compare_vxvx' functions use that new 'compare_ff' one.
* Adds a 'compare_ff_relative' function for secured ulp-based relative comparison of floats.
* Adds matching 'compare_vxvx_relative' functions.
* Adds some basic tests for compare_ff_relative.
See https://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/
Note that we could replace our python/mathutils' EXPP_FloatsAreEqual() by BLI's compare_ff_relative
(using a very small absolute max_diff), but these do not have exact same behavior...
Left a comment there for now, we can do it later if/when we are sure it won't break anything!
For some reason SCons defines _DEBUG, not DEBUG as mathutils was
expecting it to be.
Made it so mathutils checks for NDEBUG which mimics BLI_assert
define.
Changes for VC2013
Now, I can build Blender with VC2013 with Cycles, Collada, OpenExr,OpenImageIO disabled. Also, you need VC2008 sp1 installed to make old libs compatible.
The problem was that vertex colors only have 8 bits of precision, and integer
division always rounds down, so after some color blending iterations everything
gets darker. Instead use integer division that behaves like round() instead of
floor() for blending operations.
This is useful for gcc which does not define sqrtf/powf/... functions with preprocessor and therefore always used sqrt/pow/...
Float functions are generally 20-50% faster than their equivalents for double type.
* MEM_CacheLimitier - Size type to int conversion, should be safe for now (doing my best Bill Gates 640k impression)
* OpenNL CMakeLists.txt - MSVC and GCC have slightly different ways to remove definitions (DEBUG) without the compiler complaining
* BLI_math inlines - The include guard name and inline option macro name should be different. Suppressed warning about not exporting any symbols from inline math library
* BLI string / utf8 - Fixed some inconsistencies between declarations and definitions
* nodes - node_composite_util is apparently not used unless you enable the legacy compositor, so it should not be compiled in that case.
Leaving out changes to BLI_fileops for now, need to do more testing.
Its unlikely you want to do short -> int, int -> float etc, conversion during swapping (if its needed we could have a non type checking macro).
Double that the optimized assembler outbut using SWAP() remains unchanged from before.
This exposed quite a few places where redundant type conversion was going on.
Also remove curve.c's swapdata() and replace its use with swap_v3_v3()
