On Wed Apr 3 14:29:38 2024 +0000, David Heidelberg wrote:
@stefan What would be next thing to move this forward?
Hi David (and Aleksey),
I am sorry for the late reply - I was mostly away over easter.
The concerns regarding the the clipped draws and WARP failures (See my last comment) and opaque hardcoded result values (Zeb's comments) still apply I think. It seems to me the big tests[] table was written by listing all possible permutations of input values, hardcoding the output you get without thinking *why* the resulting colors are what they are.
The answer to the WARP failures might be to just accept them as broken(), although in this case please add a comment about what WARP does wrong and why we want to test the conditions where WARP misbehaves regardless. If possible, please try to calculate the expected fog values similarly to what the shader implementation should do or at least add a comment explaining the results.