Am Mittwoch, 28. März 2012, 22:10:27 schrieb Charles Davis:
- HLSL compiler.
My general response to "HLSL Compiler" would be "too complex for gsoc, too much work, nope", but since I know that you worked on compilers in the past it might work for you. I cannot promise anything of course, but my guess is that it is worth writing an application.
I think three things would be important:
1) A plan how to integrate the compiler step by step. I think a general consensus is to start with a parser that can parse the entire (d3d9?) language, but I don't know what the idea is beyond that.
2) A flexible plan that allows for adjustments depending on how the project progresses, once a complete parser is done.
The good thing about the parser is that I've once written a hacky compiler as a school project, and the parser is pretty much complete for d3d9 shaders. Even if the other code is pretty ugly, the parser should be reusable.
3) A design that is flexible enough for future needs(or well, d3d10/11). I am still afraid of getting stuck in a dead end street with anything written from scratch, but since all other options failed I guess that's the only way left.
Maybe Henri and Matteo can weight in with their opinions as well.
Cheers, Stefan