Shift from framework to library

[vorg]

So.. this approach is good on paper but causes many problems in practice for development of pex-renderer. What about moving to more of a library than framework approach with pex-shaders? So pure utils only and all the setup and wrapper code would move back to pex-renderer? Basically all chunks with main() in them... that would allow us develop new features in pex-renderer and move out to pex-shaders when stable.

[vorg]

@dmnsgn any thoughts on this? My initial idea for pex-shaders was more of Lygia than removing every single line of GLSL from pex-renderer. Ideally this lib would have only pure functions and everything requiring uniforms or setup would be left in pex-renderer for faster iteration and prototyping. So pex-shaders is more like glsl-utils than pex-renderer-glsl

[dmnsgn]

Philosophically, I agree. But:

• Many chunks are tightly coupled with their uniforms and their #defines
• I would do that refactor while removing GLSL 100 as well
• I would group chunks into subfolders
• Standard vert/frag will have a big git diff so that would be a "rip-off the bandaid" point (ie. git history will be meh but it has to happen at some point)
• pure functions and passing variables around might not be best for performance

On the plus side:

• pure = good
• will be easier for wgsl
• might help with tree shaking (given some extra work in pex-renderer)

[vorg]

Standard vert/frag will have a big git diff so that would be a "rip-off the bandaid" point (ie. git history will be meh but it has to happen at some point

Because of change to GLSL 300?