Open rafaelbeckel opened 1 year ago
I don't think it would be out of scope, it would be helpful to have more readable output especially when debugging issues.
However, I think it will not be easy to improve the status quo.
What are the steps you think of that would get us closer to generating a more human readable output?
I couldn't take a close look at that yet, but I was roughly thinking about implementing it as a post-processing step. This way, I'd avoid messing up with the internals, and it would be relatively easy to implement.
I'm using
naga
CLI to convert a few.glsl
shaders to.wgsl
with the intention of making.wgsl
my new source code.However, the output contains a lot of intermediate variables named _e{N} for each time a source variable is used, which makes the code bloated and unreadable.
I assume
naga
was designed as a backend tool and is not supposed to produce human-readable code, so these variables are there to produce reliable Spir-V bytecode from any source language.However, it would be nice to introduce an option to set the output to be as human-readable and clean as possible. I was looking for something like:
naga in.frag out.wgsl --clean
(or maybe--format=human
).Is there a way to do that? If not, would you guys accept a PR introducing this option, or is it way out of scope?