hmm I'll have to try the test suite locally, but are you sure it hangs and doesn't just take forever (it rebuilds a lot of things on the first run)

Thanks for starting CI:)

are you sure it hangs and doesn't just take forever (it rebuilds a lot of things on the first run)

I mean my laptop is admittedly quite slow, but AFAICT it isn't the rebuilding part that fails, just the integration test in particular:

     Running tests/integration.rs (target/debug/deps/integration-a917ca08f79bf777)
tests/integrations/basic/Cargo.toml ⠤
tests/integrations/basic-bin/Cargo.toml ⠤
tests/integrations/basic-fail-mode/Cargo.toml ⠠
tests/integrations/cargo-run/Cargo.toml ⠠
tests/integrations/ui_test_dep_bug/Cargo.toml ⠠
tests/integrations/basic-fail/Cargo.toml ⠠
tests/integrations/dep-fail/Cargo.toml ⠠ 

Oh, it did move after a while! The thing is that it completely maxed out all the CPUs, which is why I assumed that it got into an infinite loop somewhere

EDIT: Okay, yeah, the following runs are much smoother

Yea it rebuilds the crate for each of those tests

I also got a bunch of seemingly unrelated changes when running cargo test -- -- --bless; I removed them, I think the committed changes should be the correct ones

EDIT: oh no, apparently some unrelated changes did end up sneaking into Cargo.stdout

Maybe I need to do a version bump again of the rust version we use, let's see what CI says

That does sound right -- many of the changes seem to come from improved rustc diagnostics

EDIT: Ahh, right! That's because I have Cargo installed as a free-standing program (so not from rustup), which is why rust-toolchain.toml isn't respected and I end up running the latest stable instead of 1.80.1. Apologies for the confusion

Great! Thanks for the fix

Sorry, I'd realized that my test case doesn't actually test what it was supposed to test -- I wanted to make it emit there were 2 unmatched diagnostics, but for some reason, it emits there was 1 unmatched diagnostic twice instead. Would you happen to know why that would be?.. I'd then open a follow-up PR to fix up the test

Hmm I saw the test changes for 1 unmatched diagnostic and forgot to check that there's still a case for 2+