Keith Winstein

For `fac`, I added `--num-outputs=4` to the `wasm2c` invocation, e.g. `../../../bin/wasm2c fac.wasm -o fac.c --num-outputs=4` and then you can compile the individual .c files with something like `clang -I../.. -c...

I think my own long-term plan is to get wasm-tools's wat2wasm and wasm2wat (https://github.com/bytecodealliance/wasm-tools/) to the point where they can substitute fully for WABT's wat2wasm/wasm2wat, and then sunset WABT. wasm-tools...

I'm a little confused -- we pass the tail-call tests, so... this suggests we need some additional testing here. How is this code-path not hit by the current tests?

I guess there are no spec tests for tail-calls that use multi-value? If so that seems like an omission we should fix. @squk , are you up for either: -...

We definitely want this fix and thank you for finding and fixing the bug! If you don't have the cycles to improve the testing, as I said we can probably...

My understanding is that #2240 (specifically the failure on aarch64) was already fixed by #2406. It seems like a minimalist option could be to enforce that every platform **either** uses...

Hmm, I'm not sure I totally follow. The BinaryReaderIR does impose limits like kMaxNestingDepth, kMaxFunctionLocals, kMaxFunctionParams, and kMaxFunctionResults, which match the values in V8 (https://chromium.googlesource.com/v8/v8/+/refs/heads/main/src/wasm/wasm-limits.h) and hopefully the other big...

FWIW, when I hit a case where the prose spec is unclear, I usually just go try it in the reference interpreter and in wasm-tools. If those two disagree, it's...

Yeah, I agree the problem is with the text parser -- these are supposed to be malformed (not just invalid). I think ideally the validator would not be getting run...

Hmm, I can't speak to the web demo, but this seems to work fine in the actual main branch of the codebase. Are you able to bisect and identify when...