Josh Triplett

Could you post a transcript of the test failure (including with nocapture so we can see the actual test output)?

So, reading that test case, I don't think that's a test that should pass in general, even if the platform doesn't actually reserve such filenames. The logic that checks for...

> If I remember right, the code for publishing packages tries to prevent you from publishing packages that couldn't be unpacked on all OSes supported by Cargo, so it makes...

Per discussion on Zulip, this approach seems fine to me, as long as it doesn't actually use the same code in both this check and the code being tested (which...

Seems reasonable to me. Does anyone see an issue with adding that dependency to the test support crate?

@ehuss Good point; done.

@ehuss We've talked about, for other reasons, separating compilation and linking: https://github.com/rust-lang/rust/issues/64191 If we do that in rustc, cargo should use that, and could then reflect that in its output.

I *do* think it makes sense to keep these separate. I like the idea of having `--out-dir` to put the final build artifacts and not all the intermediate bits.

@Byron I answered the metadata test question in the corresponding linked issue. On further consideration, I don't think you need to put any effort at all into reducing the number...

> The package name is used instead of the name in the manifest, the hashes end up being the same. Ah, I see! I think, for artifact dependencies, it may...