Don't merge trait_impls_in_deps results
Closes https://github.com/rust-lang/rust-analyzer/issues/9167
We effectively keep a lot of copies of `TraitImpl` results in memory by merging them into the result of this query. This wastes a lot of memory unnecessarily. The change here brings memory usage of self down by 50mb (at a slight cost of trait solving unfortunately, though I'm hopeful it was just noice)
Refactor symbol index
Closes https://github.com/rust-lang/rust-analyzer/issues/14677
instead of eagerly fetching the source data in symbol index we do it lazily now, this shouldn't make it much more expensive as we had to parse the source most of the time anyways even after fetching.
fix: ide: exclude sized in go-to actions in hover
fixes#13163
i opted to just simply omit `Sized` entirely from go-to actions, as opposed to including it if even someone writes an explicit `T: Sized`, as i think a go-to on Sized is of dubious value practically.
feat: Highlight closure captures when cursor is on pipe or move keyword
This runs into the same issue on vscode as exit points for `->`, where highlights are only triggered on identifiers, https://github.com/rust-lang/rust-analyzer/issues/9395
Though putting the cursor on `move` should at least work.
chore: rust-analyzer: refactor notification handlers
Fixes the FIXME in `on_notification`.
```rust
// FIXME: Move these implementations out into a module similar to on_request
```
No code has changed, this just moves stuff around.
Don't validate constants in const propagation
Validation is neither necessary nor desirable.
The constant validation is already omitted at mir-opt-level >= 3, so there there are not changes in MIR test output (the propagation of invalid constants is covered by an existing test in tests/mir-opt/const_prop/invalid_constant.rs).
MIR episode 4
In lowering, it now supports overloaded and arith assignment binary operators, statics. and constants in patterns. There is now 252 functions that we fail to emit mir for them, and the majority of them are due type mismatches or other deep and unrelated issues (but it isn't done yet, for example slice patterns and destructing assignment is not implemented yet).
In evaluating, it now can evaluate associated constants in traits (so now typenum's `U5::ToConst` should work), allocator functions, atomic intrinsics, and some more things. It also provides a (hacky) basis for making progress in #14275. I also added a `Interpret` code lens to `Run` and `Debug` when the experimental `interpret tests` is enabled (previously it showed result in hover, which was unusable even for debugging)
Changes in unrelated files are:
* Changes substitutions of closures, now it includes parent substs ~~before~~ after `sig_ty`.
* ~~A salsa input for retrieving the path of a file id, used in emitting stack trace for interpret result.~~
* Normalizing associated types in layout computing
Parallelize initial Rust download in bootstrap
Parallelize the initial download of Rust in `bootstrap.py`
`time ./x.py --help` after `rm -r build`
Before: 33s
After: 27s
`inline(always)` for `lt`/`le`/`ge`/`gt` on integers and floats
I happened to notice one of these not getting inlined as part of `Range::next` in <https://rust.godbolt.org/z/4WKWWxj1G>
```rust
bb1: {
StorageLive(_5);
_6 = &mut _4;
StorageLive(_21);
StorageLive(_14);
StorageLive(_15);
_15 = &((*_6).0: usize);
StorageLive(_16);
_16 = &((*_6).1: usize);
_14 = <usize as PartialOrd>::lt(move _15, move _16) -> bb7;
}
```
So since a call for something that's just one instruction is never the right choice, `#[inline(always)]` seems appropriate, like we have it on things like the rotate methods on integers.
Remove `QueryEngine` trait
This removes the `QueryEngine` trait and `Queries` from `rustc_query_impl` and replaced them with function pointers and fields in `QuerySystem`. As a side effect `OnDiskCache` is moved back into `rustc_middle` and the `OnDiskCache` trait is also removed.
This has a couple of benefits.
- `TyCtxt` is used in the query system instead of the removed `QueryCtxt` which is larger.
- Function pointers are more flexible to work with. A variant of https://github.com/rust-lang/rust/pull/107802 is included which avoids the double indirection. For https://github.com/rust-lang/rust/pull/108938 we can name entry point `__rust_end_short_backtrace` to avoid some overhead. For https://github.com/rust-lang/rust/pull/108062 it avoids the duplicate `QueryEngine` structs.
- `QueryContext` now implements `DepContext` which avoids many `dep_context()` calls in `rustc_query_system`.
- The `rustc_driver` size is reduced by 0.33%, hopefully that means some bootstrap improvements.
- This avoids the unsafe code around the `QueryEngine` trait.
r? `@cjgillot`
