Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 19:00:01 -0400] rev 153
Added finderrors.c to revision control.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 16:37:54 -0400] rev 152
Relative addressing work.
Parsing and state appears to work now, D3D profile appears to output correctly.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 13:50:26 -0400] rev 151
Cleaned up tests for shader type and version.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 13:32:05 -0400] rev 150
Fixed bogus instruction token count errors.
Hardcoded instruction token count is wrong when relative addressing is used,
so you have to do it by what they reported vs. what we actually parsed, not
what a specific instruction should generically be.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 04:59:27 -0400] rev 149
Commented out apparently-bogus validity check.
MOD_SATURATE is showing up in pixel shaders generated from Microsoft's HLSL
compiler, so either I misread the spec when I thought this modifier was
vertex-shader-specific, or the spec is wrong. Will check later.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 03:58:57 -0400] rev 148
Hopefully sorted out the reported uniform/attribute mess.
This removes output registers from the attribute list, since those aren't
something the application should bind. Also, the bulk of this work is setting
up a new namespace for samplers.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 02:29:41 -0400] rev 147
Removed explicit numbers on enums in mojoshader.h.
Why was I doing that, again?
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 02:26:26 -0400] rev 146
Export sampler information in public API.
Ryan C. Gordon <icculus@icculus.org> [Sat, 19 Apr 2008 01:31:47 -0400] rev 145
Don't create unnecessary attributes.
Pixel shaders and no-item lists shouldn't call Malloc().
Ryan C. Gordon <icculus@icculus.org> [Fri, 18 Apr 2008 22:00:19 -0400] rev 144
Fixed TEX argument parsing.
Apparently it doesn't use the same args as TEXCOORD.