Remove MSVC's wmemchr declaration

[tesuji]

No description provided.

[rust-highfive]

r? @gnzlbg

(rust_highfive has picked a reviewer for you, use r? to override)

[tesuji]

r? @JohnTitor

[tesuji]

Style (lint) check failed

+ ./style src
src/windows/mod.rs:383 - use cfg_if! and submodules instead of #[cfg]

What should I do? There are no way to link to wmemchr on MSVC.

[igrep]

I guess you should move the function you modified to https://github.com/rust-lang/libc/blob/master/src/windows/gnu/mod.rs
src/windows/mod.rs should contain only functions common in the Windows targets.

[tesuji]

Thanks. Guess it should be fine now.

[JohnTitor]

Hm, it'd be great if CI could detect this absence. cc @retep998 are we correct?

[retep998]

I'm surprised CI hadn't detected this. The change looks correct to me.

[JohnTitor]

Thanks, @retep998!

I'm surprised CI hadn't detected this.

Yeah, I'll take a look later.

@lzutao r=me once code is formatted, line length limit here is tight a bit.
Also, thanks @igrep for the comment.

[tesuji]

CI could detect this absence

I think that the extern block relates to linkage, which happen during linking stage and after codegen.
IMO we cannot test this case unless we have tests for every libc function.

[JohnTitor]

IMO we cannot test this case unless we have tests for every libc function.

Well CI should do that with ctest crate iirc?

[JohnTitor]

Anyway I'm going to merge this, thanks all!

[tesuji]

Just curious: So we don't use bors anymore.

[JohnTitor]

Just curious: So we don't use bors anymore.

Yeah, Alex wants to abandon the use of bors here, I think we'll do so if there are no issues for a while.