Migrate test suite to use run-pass stdout/stderr testing

[Mark-Simulacrum]

Where possible, we should prefer to use the functionality added by #63825 instead of std::process and the like. At least one such test is known, but it's possible more can be found.

[Mark-Simulacrum]

I'll try to get mentoring instructions up for the actual move but the first step here is to look for all run-pass tests (// run-pass in src/test/ui I believe) and verify we're ignoring cloudabi, emscripten, and sgx in all of them. This is pretty annoying to do manually I suspect but I'd recommend comparing the results of rg -o ignore-\w+ src/test/ui and rg -c run-pass src/test/ui (exact commands may need to differ)

[RalfJung]

verify we're ignoring cloudabi, emscripten, and sgx in all of them

It does not at all seem like we do:

$ rg run-pass src/test/ui | wc -l
3200
$ rg ignore-cloudabi src/test/ui | wc -l
82

[RalfJung]

I think that file changed in #63825 was ignored just because it uses process -- but just running the test seems to work fine on those platforms. I suppose CI has some way to cross-run them and compiletest uses that?

It does not use process any more so probably it can be unignored.

[Mark-Simulacrum]

Ah, so the comment is just misleading in that this is not "no processes" but rather no "std::process"?

[RalfJung]

That's how I initially read that comment, and I still do, yes.

[Mark-Simulacrum]

Aha, okay, then probably just misread on my part. Closing since we probably don't actually want to do anything here.

[RalfJung]

Well I think we want to remove those ignore from the test that got changed in #63825.

[Mark-Simulacrum]

I guess so. I also don't care too much, it seems fine to ignore on those platforms too. But reopening.

[RalfJung]

Well, that PR hasn't landed yet, so you could just r- and ask the author to remove the ignore?

I can't even open a PR to fix this until that PR landed.^^

[nathanwhit]

(Author of #63825) Since the PR is still open, and I went ahead and re-removed the ignores and pushed again, so that should be resolved.

[RalfJung]

Thanks a lot!

[RalfJung]

@Mark-Simulacrum I only just saw you changed the bug title... I guess I should reopen as you repurposed the issue? Or would it make more sense to open a new one, as the existing discussion here will be rather confusing?

[Mark-Simulacrum]

New issue is fine, I don't care too much realistically :)