[Merged by Bors] - Shaders can now have #else ifdef chains

[torsteingrindvik]

Objective

Currently, shaders may only have syntax such as

#ifdef FOO
// foo code
#else
#ifdef BAR
// bar code
#else
#ifdef BAZ
// baz code
#else
// fallback code
#endif
#endif
#endif

This is hard to read and follow.
Add a way to allow writing #else ifdef DEFINE to reduce the number of scopes introduced and to increase readability.

Solution

Refactor the current preprocessing a bit and add logic to allow #else ifdef DEFINE.
This includes per-scope tracking of whether a branch has been accepted.
Add a few tests for this feature.

With these changes we may now write:

#ifdef FOO
// foo code
#else ifdef BAR
// bar code
#else ifdef BAZ
// baz code
#else
// fallback code
#endif

instead.

---

Changelog

• Add #else ifdef to shader preprocessing.

[torsteingrindvik]

Trying to puzzle this out threw my head for a spin.
I'm not entirely confident this is free of errors, but the current tests and new ones are passing so the basics should be in place.

I think a follow up PR (or PRs) could add more features:

• Allowing #else ifndef in the chains
• Allow value operations (==, !=, < etc) in chains too.

[superdump]

Clean and clear. Looks good to me.

[superdump]

@cart do you have any opinion about #else ifdef FOO? It's a departure from C preprocessor #elseif but I always found the discrepancy of #if/#ifdef yet #elseif to be slightly weird, and #elseifdef feels like too much of a mouthful to me.

[ErickMVdO]

Hi, @oCaioOliveira and I are getting to know the project now and decided to review this PR to get a better understanding of the code. After doing some research, we found the match constructor, which would be equivalent to the switch in C. Would this not be more appropriate to reduce the number of introduced scopes?

[torsteingrindvik]

Hi, @oCaioOliveira and I are getting to know the project now and decided to review this PR to get a better understanding of the code. After doing some research, we found the match constructor, which would be equivalent to the switch in C. Would this not be more appropriate to reduce the number of introduced scopes?

I'm not sure I'm following.
The scopes refer to the logical scopes in the parsing of the shader being preprocessed.
For example, given the line #ifdef FOO, a scope is introduced. That scope survives until the paired #endif is encountered.

This PR allows using #else ifdef instead of adding extra #ifdefs, hence reducing the number of needed scopes.

So I'm not sure where match fits in here- I think you are talking about the keyword in Rust? Do you mean to add it as a keyword in the shader preprocessor (for use in wgsl)? Or using match in the shader preprocessing Rust code as a refactor?

[robtfm]

I'm not sure where match fits in here

agreed, i think match/switch/case might be useful for the #if x == 1 ... #else if x == 2 ... #endif type of elsifs, but it doesn't make sense for #else ifdef.

i checked through the scope logic here and it looks good to me.

[superdump]

bors r+

[bors]

Pull request successfully merged into main.

Build succeeded:

• build-and-install-on-iOS
• build-android
• build (macos-latest)
• build (ubuntu-latest)
• build-wasm
• build (windows-latest)
• build-without-default-features (bevy)
• build-without-default-features (bevy_ecs)
• build-without-default-features (bevy_reflect)
• check-compiles
• check-doc
• check-missing-examples-in-docs
• ci
• markdownlint
• msrv
• run-examples
• run-examples-on-wasm
• run-examples-on-windows-dx12