Sure, I think it's reasonable to expose the module signature.

Could you also explain the workflow that requires matching LLVM's file names? I personally just dump everything to a single file and then pass it to grcov.

Could you also explain the workflow that requires matching LLVM's file names? I personally just dump everything to a single file and then pass it to grcov.

I'm currently working on adding nextest support to wasm-bindgen-test-runner, which executes all tests in parallel. Therefor the runner process is spawned many times and I can't merge everything in a single file. So while there are a bunch of possible solutions to that, my overall efforts so far have been to align as close as possible to cargo test, which solves this issue by producing unique file names for every file.

So no, my workflow doesn't not require the module signature, but it would be great to have and the tradeoff seems quite small to me.

If you don't actually care about the name, just that it's unique, then there are better ways of doing this than using lprofGetLoadModuleSignature. Is there a specific reason for wanting the same names as the ones generated by LLVM?

If you don't actually care about the name, just that it's unique, then there are better ways of doing this than using lprofGetLoadModuleSignature.

Indeed, I'm aware!

Is there a specific reason for wanting the same names as the ones generated by LLVM?

wasm-bindgen is a foundational and low-level crate. I'm trying to eliminate any differences to cargo test to allow users to expect the same output as close as I can possibly get it. In this case I'm trying to emulate LLVM_PROFILE_FILE. wasm-bindgen-test-runner is able to cover all the patterns except %m. However, instead of %m generating the instrumented binary signature, I can just generate a random number or a hash of the output. I would however prefer simply aligning with LLVM if possible.

But no, there is no specific use-case that I anticipate that will require files to be named exactly the same.

I think it's still fine to expose this from minicov.