Refactor: Cleanup one part of assign_ops lint
Removes a lot of indentation and separates lint emission from lint
logic. Only touches the `hir::ExprKind::AssignOp` part of the lint.
Refactor: Extract `trait_ref_of_method` function
This pattern was used in three places after #3844, so I think it's worth moving it into `utils/mod.rs` and documenting it.
Cognitive Complexity (step 1 out of 3+): name changes
Following up on #3793
**Overall checklist:**
1. **Name changes**
2. MVP of functionality
3. Tests
After this PR, we will start working on the implementation itself.
* Ran automatic naming update
* Formalized rename of `cyclomatic_complexity` to `cognitive_complexity`
** Added the rename to `lib.rs`
** Added rename test
* Added warning for deprecated key `cyclomatic_complexity_threshold` and tests for it
* Added deprecation status for Clippy's builtin attribute
* Updated tests for new builtin attribute renaming
move lint documentation into macro invocations
This PR moves lint documentation inside `declare_clippy_lint!` macro invocations, to avoid triggering the `unused_doc_comments` lint once it's modified in rust-lang/rust#57882. This PR is necessary to unblock that work, since the large number of warnings generated in `clippy_lints` causes Travis to hit the log length limit.
This PR also updates the documentation and website generation script.
It would be nice to get a clippy update in the Rust repo once this is merged.
cc @phansch
Use the same <title> as <h1> on gh-pages
I think this makes the page easier to find using firefox's URL bar - it (afaik) searches based on the page's URL and title so if somebody doesn't have it bookmarked but remembers that the page contained "ALL lints" or something similar it's easy to go to it directly without resorting to a search engine.
Error on unknown or deprecated attributes
The reason I decided to not make this a lint is, that unknown attributes are always a hard error in rustc and also in rustfmt, which cannot be turned off by an `allow`. This could have been done by introducing a category which would be forbid-by-default. But that would have introduced a new lint category+group for just one lint and also wouldn't be consistent with rustfmt and rustc.
