Stream: t-compiler/wg-rls-2.0

Topic: Failed CI in PR


Paul Faria (Aug 08 2020 at 03:06, on Zulip):

Any ideas why the latest CI run in https://github.com/rust-analyzer/rust-analyzer/pull/4743 failed? My guess is because no tests ran in the mac target. I'm not able to trigger a re-run though, and I'm not sure why it would fail. The last push was a rebase to fix merge conflicts.

Laurențiu Nicola (Aug 08 2020 at 04:00, on Zulip):

thread 'test_format_document' panicked at 'called Result::unwrap() on an Err value: Os { code: 66, kind: Other, message: "Directory not empty" }', crates/rust-analyzer/tests/heavy_tests/testdir.rs:44:9

Laurențiu Nicola (Aug 08 2020 at 04:00, on Zulip):

Hmm

Laurențiu Nicola (Aug 08 2020 at 04:02, on Zulip):

So we've recently replaced tempdir with a simpler implementation: https://github.com/rust-analyzer/rust-analyzer/pull/5510/files, but I'd expect it to be more problematic on Windows than Mac OS

Laurențiu Nicola (Aug 08 2020 at 04:07, on Zulip):

I don't know why https://github.com/rust-analyzer/rust-analyzer/blob/master/crates/rust-analyzer/tests/heavy_tests/testdir.rs#L44 would panic. Maybe a new file showed up while it was removing them?

Laurențiu Nicola (Aug 08 2020 at 04:08, on Zulip):

It could also be a bug in the server or a weird interaction between tests, but I don't think it's related to your changes

Laurențiu Nicola (Aug 08 2020 at 04:09, on Zulip):

I'd try to push a dummy commit to trigger CI again, and maybe run the tests locally with RUN_SLOW_TESTS=1, but I don't think it's that easy to reproduce

Paul Faria (Aug 08 2020 at 15:05, on Zulip):

I'll rebase on master and push again

Paul Faria (Aug 08 2020 at 15:38, on Zulip):

Looks like that fixed it :)

Last update: Sep 27 2020 at 14:00UTC