Improve accuracy of mut_key

[kneasle]

Fixes #6745.

Whilst writing the tests for this, I noticed what I believe is a false negative (the code in @xFrednet's comment doesn't trigger the lint). Currently the tests contain a case for this (which is blatantly ignored), but I'm not at all sure how to implement this (since the lint currently behaves completely differently for ADTs). I'm not sure what should be done - on the one hand the extra test cases are misleading, but on the other hand they don't cause much harm and would save effort for anyone fixing that false negative.

---

changelog: Improve accuracy of clippy::mutable_key_type.

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @camsteffen (or someone else) soon.

Please see the contribution instructions for more information.

[camsteffen]

Whilst writing the tests for this, I noticed what I believe is a false negative (the code in @xFrednet's comment doesn't trigger the lint). Currently the tests contain a case for this (which is blatantly ignored), but I'm not at all sure how to implement this (since the lint currently behaves completely differently for ADTs). I'm not sure what should be done - on the one hand the extra test cases are misleading, but on the other hand they don't cause much harm and would save effort for anyone fixing that false negative.

My personal opinion is that this case is out of scope for the lint. I think if a user defines a type with a pointer and a custom impl Hash, they are on their own.

[camsteffen]

A few nits and this is nearly ready. Please update the changelog in your post to reflect added changes.

[kneasle]

Wait hold on what's going on with the unit tests? It seems to be that my local build of clippy is producing different test results to the build in the PR checker. So when I cargo dev bless my local copy it causes the PR check to fail. @camsteffen what d'ya reckon I should do? I can update my PR branch to be what the test runner wants (which will let this PR be merged), but I feel like I can't be alone in having this discrepancy.

[camsteffen]

Your test output is not including changes from recent PRs. Try rebasing, don't run with TESTNAME, and maybe try cargo clean. Or just don't check in unrelated changes.

[kneasle]

Aha I've figured out what's going - I appear to not have target-dir not set in <clippy>/.cargo/config. So it was compiling my new changes into ~/.build/rust but the test runner was still reading the old version from <clippy>/test.

[kneasle]

Yes finally! The tests are passing at last.

[camsteffen]

Please squash some commits and I'll merge.

[kneasle]

Please squash some commits and I'll merge.

Rebased down to two commits :).

[camsteffen]

Thanks! @bors r+

[bors]

📌 Commit b2ffb28 has been approved by camsteffen

[bors]

⌛ Testing commit b2ffb28 with merge 746a005...

[bors]

☀️ Test successful - checks-action_dev_test, checks-action_remark_test, checks-action_test
Approved by: camsteffen
Pushing 746a005 to master...

[kneasle]

🎉🎉🎉🎉