Use --color=always in rustdoc command when stderr is tty

[tonowak]

Fixes #326

[obi1kenobi]

Should this reference the GlobalConfig struct's stderr value instead of the global stderr, just in case they are different?

This might need a bit more plumbing.

[tonowak]

But isn't anyhow::bail! using the normal stderr, and that's where this command's output will go?

[tonowak]

cargo-semver-checks/src/rustdoc_cmd.rs

Line 92 in 4de3203

anyhow::bail!(

[obi1kenobi]

That is an excellent point!

We should change that too, before a "cargo-semver-checks-as-a-library" user opens an issue about their library polluting their stderr even though they configured its stderr behavior to be something else.

I take it back: bail!() doesn't print to stderr, it's just return Err() with that error message: https://docs.rs/anyhow/latest/anyhow/macro.bail.html

[tonowak]

I see. But nevertheless it should use the GlobalConfig, right?

[tonowak]

Done in 8f7d387.

[obi1kenobi]

Yes, I think so.

…

On Mon, Feb 20, 2023, 8:29 AM Tomasz Nowak ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In src/rustdoc_cmd.rs <#360 (comment)> : > + if atty::is(atty::Stream::Stderr) { + cmd.arg("--color=always"); + } I see. But nevertheless it should use the GlobalConfig, right? — Reply to this email directly, view it on GitHub <#360 (comment)>, or unsubscribe </~https://github.com/notifications/unsubscribe-auth/AAR5MSXBSWCA67KNJJJCRTTWYNWUTANCNFSM6AAAAAAUWWGVO4> . You are receiving this because your review was requested.Message ID: ***@***.***>

[tonowak]

Why is the CI failing? I don't understand it. The compilation error references line numbers in rustdoc_cmd.rs which don't exist (the file is shorter than the line numbers).

[obi1kenobi]

Why is the CI failing? I don't understand it. The compilation error references line numbers in rustdoc_cmd.rs which don't exist (the file is shorter than the line numbers).

Possibly due to merging main into this branch first. There's some poorly-documented default behavior of GitHub Actions where PR tests implicitly run on the current branch with main merged into it, IIRC.

I just merged main and if that causes the line numbers to more-or-less match up (because I merged #375 into main first, after this CI run), that might be the cause.

[obi1kenobi]

Nice!