[wasm2c] MSVC miscompiles for certain fp constants about wabt

[keithw]

Thanks for reporting this! Can you please test #2389 and see if it fixes the issue for you?

There should probably be a test for this situation, ideally eventually upstream in the spec tests.

I'm also nervous about MSVC's handling of wasm2c output for conversion instructions given https://github.com/WebAssembly/wabt/blob/main/src/c-writer.cc#L4851 and https://github.com/WebAssembly/wabt/blob/main/src/config.cc#L76 -- we should make sure this is also tested.

[rossberg]

There should probably be a test for this situation, ideally eventually upstream in the spec tests.

@keithw, PTAL at this spec PR.

[InflexCZE]

#2389 looks good, no unexpected overhead either
https://godbolt.org/z/xf4Wfa4Ev

1. Might be good to export with //decimal value comment at the end of the line, for users' convenience
2. Will prolly need byte re-order for little-endian

[SoniEx2]

it works fine on both endian

[keithw]

#2389 looks good, no unexpected overhead either https://godbolt.org/z/xf4Wfa4Ev

1. Might be good to export with `//decimal value` comment at the end of the line, for users' convenience

Thanks, good idea. Done in #2389.