LDFLAGS mapping breaks on msvc

[lu-zero]

As seen on here/ci:

stderr=```cl : Command line error D8021 : invalid numeric argument '/Wl,-LD:\a\cargo-c\cargo-c\example-project\target\.\release'

[Hywan]

Hello,

Thanks for the issue. I've no idea yet how MSVC is supposed to handle LDFLAGS. I'll check. Is it blocking for you?

[lu-zero]

I just passed everything through CFLAGS.

[Hywan]

OK. So I've time to find a proper solution, which starts by having a Windows VM somewhere :-D!

[Tim-Evans-Seequent]

For reference, the correct way to pass arguments to the linker via MSVC is via "/link". All arguments after /link are passed to the linker.

https://learn.microsoft.com/en-us/cpp/build/reference/compiler-command-line-syntax?view=msvc-170

[camio]

It looks like this issue was addressed by the wasmer project by forking this repository into https://github.com/wasmerio/inline-c. The package to use on cargo is https://crates.io/crates/wasmer-inline-c/0.1.2

[Hywan]

Be careful. wasmer-inline-c was a fork of this project, that is pointing to https://github.com/fschutt/wasmer-inline-c-rs, which is now dead. The first release was on September 2022. The attributions have even been corrected to reflect it was a fork.

Wasmer is trying to steal this project from me, and has created another fork https://github.com/wasmerio/inline-c (which apparently is a copy of https://github.com/fschutt/wasmer-inline-c-rs).

All this is unclear and I'm exchanging emails with Syrus Akbary (CEO of Wasmer) to try to protect my work. As soon as I've pointed him out that https://github.com/fschutt/wasmer-inline-c-rs was dead, he has created https://github.com/wasmerio/inline-c.

I don't want this fight. Syrus is claiming the ownership after 3-4 years. This is non-sense…

[Hywan]

It's even better than that.

So Syrus has put https://github.com/fschutt/wasmer-inline-c-rs back online at https://github.com/wasmerio/inline-c. It's interesting to read the commit history.

• First commit, wasmerio/inline-c@1edacf7, we see that fschutt (who was working at Wasmer at that time) has stolen my code. This first commit copies the inline-c-macro code, even with my name as the author (wasmerio/inline-c@1edacf7#diff-8f8d4d6e0fe204c781061cda8ccf902914990001e641e61ca2c7c181e386d4dbR6-R8). This first commit copies the inline-c code (the run.rs and assert.rs modules have been inlined, that's mostly it, wasmerio/inline-c@1edacf7#diff-b1a35a68f14e696205874893c07fd24fdb88882b47c23cc0e0c80a30c7d53759).
• Second commit (2 days later), wasmerio/inline-c@ef0fe1c. fschutt has clarified the author to restore my name,
• Third commit, wasmerio/inline-c@0f0b3a6, nothing was compiling so stuff had to be fixed.

Note that the LICENSE.md file has not been copied, here is my LICENSE, https://github.com/Hywan/inline-c-rs/blob/master/LICENSE.md.

Was Syrus aware? Probably, as he is co-owner on crates.io, https://crates.io/crates/wasmer-inline-c:

At the time of writing (September 14th, 2024), he has not updated crates.io yet.

[camio]

The "wasmer-inline-c" fork, as far as I can tell, works only for windows and removed the ability to run inline C/C++ code. Wasmer's c-api code switches between the original inline-c and the fork based on platform detection.

Ideally there'd be one inline-c repository (Hywan/inline-c-rs) and it would have full support (execution included) for Windows. I'd be surprised if wasmer wouldn't switch to such a thing if that were in place. Would you accept a patch adding Windows support? I'm confident I could do it. We could also add a GitHub workflow that will verify this works as part of CI.

[Hywan]

I would definitely be happy with a patch for Windows support yeah. Feel free to provide one :-).