Compute lint levels by definition
Lint levels are currently computed once for the whole crate. Any code that wants to emit a lint depends on this single `lint_levels(())` query. This query contains the `Span` for each attribute that participates in the lint level tree, so any code that wants to emit a lint basically depends on the spans in all files in the crate.
Contrary to hard errors, we do not clear the incremental session on lints, so this implicit world dependency pessimizes incremental reuse. (And is furthermore invisible for allowed lints.)
This PR completes https://github.com/rust-lang/rust/pull/99634 (thanks for the initial work `@fee1-dead)` and includes it in the dependency graph.
The design is based on 2 queries:
1. `lint_levels_on(HirId) -> FxHashMap<LintId, LevelAndSource>` which accesses the attributes at the given `HirId` and processes them into lint levels. The `TyCtxt` is responsible for probing the HIR tree to find the user-visible level.
2. `lint_expectations(())` which lists all the `#[expect]` attributes in the crate.
This PR also introduces the ability to reconstruct a `HirId` from a `DepNode` by encoding the local part of the `DefPathHash` and the `ItemLocalId` in the two `u64` of the fingerprint. This allows for the dep-graph to directly recompute `lint_levels_on` directly, without having to force the calling query.
Closes https://github.com/rust-lang/rust/issues/95094.
Supersedes https://github.com/rust-lang/rust/pull/99634.
Refactor macro-by-example code
I had a look at the MBE code because of #7857. I found some easy readability wins, that might also _marginally_ improve perf.
make `mk_attr_id` part of `ParseSess`
Updates #48685
The current `mk_attr_id` uses the `AtomicU32` type, which is not very efficient and adds a lot of lock contention in a parallel environment.
This PR refers to the task list in #48685, uses `mk_attr_id` as a method of the `AttrIdGenerator` struct, and adds a new field `attr_id_generator` to `ParseSess`.
`AttrIdGenerator` uses the `WorkerLocal`, which has two advantages: 1. `Cell` is more efficient than `AtomicU32`, and does not increase any lock contention. 2. We put the index of the work thread in the first few bits of the generated `AttrId`, so that the `AttrId` generated in different threads can be easily guaranteed to be unique.
cc `@cjgillot`
Update rust-analyzer documentation, mention linkedProjects
r-a uses the `rustc-dev` component from the rustup installed toolchain clippy specifies so it doesn't need to be manually installed. Also remove references to nightly r-a as the feature is long stable
I discovered `rust-analyzer.linkedProjects` recently and it has made working on the crates not referenced by the `clippy` crate so much nicer
changelog: none
Make module-style lints resilient to --remap-path-prefix
changelog: [`self_named_module_files`], [`mod_module_files`]: Make module-style lints resilient to `--remap-path-prefix`
Without this if a user has configured `--remap-path-prefix` to be used for a prefix containing the current source directory the lints would silently fail to generate a warning.
The background is transparent by default.
It was added in 5a01dbe67b43660bf1df96074f34a635aad50e56 to work around a bug
in the JavaScript syntax highlighting engine that rustdoc used at the time.
`bool_to_int_with_if` inverse case patch
Enhances `bool_to_int_with_if` such that it can also catch an inverse bool int conversion scenario, and makes the right suggestion for converting to int with a prefixed negation operator.
changelog: [`bool_to_int_with_if`]: Now correctly detects the inverse case, `if bool { 0 } else { 1 }`
Initial implementation of dyn*
This PR adds extremely basic and incomplete support for [dyn*](https://smallcultfollowing.com/babysteps//blog/2022/03/29/dyn-can-we-make-dyn-sized/). The goal is to get something in tree behind a flag to make collaboration easier, and also to make sure the implementation so far is not unreasonable. This PR does quite a few things:
* Introduce `dyn_star` feature flag
* Adds parsing for `dyn* Trait` types
* Defines `dyn* Trait` as a sized type
* Adds support for explicit casts, like `42usize as dyn* Debug`
* Including const evaluation of such casts
* Adds codegen for drop glue so things are cleaned up properly when a `dyn* Trait` object goes out of scope
* Adds codegen for method calls, at least for methods that take `&self`
Quite a bit is still missing, but this gives us a starting point. Note that this is never intended to become stable surface syntax for Rust, but rather `dyn*` is planned to be used as an implementation detail for async functions in dyn traits.
Joint work with `@nikomatsakis` and `@compiler-errors.`
r? `@bjorn3`