fix: Wrong closure kind deduction for closures with predicates
Completes #16472, fixes#16421
The changed closure kind deduction is mostly simlar to `rustc_hir_typeck/src/closure.rs`.
Porting closure sig deduction from it seems possible too and I'm considering doing it with another PR
feat: Add "make tuple" tactic to term search
Follow up to https://github.com/rust-lang/rust-analyzer/pull/16092
Now term search also supports tuples.
```rust
let a: i32 = 1;
let b: f64 = 0.0;
let c: (i32, (f64, i32)) = todo!(); // Finds (a, (b, a))
```
In addition to new tactic that handles tuples I changed how the generics are handled.
Previously it tried all possible options from types we had in scope but now it only tries useful ones that help us directly towards the goal or at least towards calling some other function.
This changes O(2^n) to O(n^2) where n is amount of rounds which in practice allows using types that take generics for multiple rounds (previously limited to 1). Average case that also used to be exponential is now roughly linear.
This means that deeply nested generics also work.
````rust
// Finds all valid combos, including `Some(Some(Some(...)))`
let a: Option<Option<Option<bool>>> = todo!();
````
_Note that although the complexity is smaller allowing more types with generics the search overall slows down considerably. I hope it's fine tho as the autocomplete is disabled by default and for code actions it's not super slow. Might have to tweak the depth hyper parameter tho_
This resulted in a huge increase of results found (benchmarks on `ripgrep` crate):
Before
````
Tail Expr syntactic hits: 149/1692 (8%)
Tail Exprs found: 749/1692 (44%)
Term search avg time: 18ms
```
After
```
Tail Expr syntactic hits: 291/1692 (17%)
Tail Exprs found: 1253/1692 (74%)
Term search avg time: 139ms
````
Most changes are local to term search except some tuple related stuff on `hir::Type`.
performance: Speed up Method Completions By Taking Advantage of Orphan Rules
(Continues https://github.com/rust-lang/rust-analyzer/pull/16498)
This PR speeds up method completions by doing two things without regressing `analysis-stats`[^1]:
- Filter candidate traits prior to calling `iterate_path_candidates` by relying on orphan rules (see below for a slightly more in-depth explanation). When generating completions [on `slog::Logger`](5e9e59c312/common/src/ledger.rs (L78)) in `oxidecomputer/omicron` as a test, this PR halved my completion times—it's now 454ms cold and 281ms warm. Before this PR, it was 808ms cold and 579ms warm.
- Inline some of the method candidate checks into `is_valid_method_candidate` and remove some unnecessary visibility checks. This was suggested by `@Veykril` in [this comment](https://github.com/rust-lang/rust-analyzer/pull/16498#issuecomment-1929864427).
We filter candidate traits by taking advantage of orphan rules. For additional details, I'll rely on `@WaffleLapkin's` explanation [from Zulip](https://rust-lang.zulipchat.com/#narrow/stream/185405-t-compiler.2Frust-analyzer/topic/Trait.20Checking/near/420942417):
> A type `A` can only implements traits which
> 1. Have a blanket implementation (`impl<T> Trait for T {}`)
> 2. Have implementation for `A` (`impl Trait for A {}`)
>
> Blanket implementation can only exist in `Trait`'s crate. Implementation for `A` can only exist in `A`'s or `Trait`'s crate.
Big thanks to Waffle for its keen observation!
---
I think some additional improvements are possible:
- `for_trait_and_self_ty` seemingly does not distinguish between `&T`, `&mut T`, or `T`, resulting in seemingly irrelevant traits like `tokio::io::AsyncWrite` being being included for, e.g., `&slog::Logger`. I don't know they're being considered due to the [autoref/autoderef behavior](a02a219773/crates/hir-ty/src/method_resolution.rs (L945-L962)), but I wonder if it'd make sense to filter by mutability earlier and not consider trait implementations that require `&mut T` when we only have a `&T`.
- The method completions [spend a _lot_ of time in unification](https://rust-lang.zulipchat.com/#narrow/stream/185405-t-compiler.2Frust-analyzer/topic/Trait.20Checking/near/421072356), and while there might be low-hanging fruit there, it might make more sense to wait for the new trait solver in `rustc`. I dunno.
[^1]: The filtering occurs outside of typechecking, after all.
fix: Don't panic on synthetic syntax in inference diagnostics
Temporary fix for https://github.com/rust-lang/rust-analyzer/issues/16682
We ought to rethink how we attach diagnostics to things, as IDs don't work for `format_args` like that!
fix: panic when inlining callsites inside macros' parameters
Close#16660, #12429, #10695.
When `inline_into_callers` encounters callsites in macros parameters, it can lead to panics. Since there is no perfect way to handle macros, this PR directly filters out these cases.
`get_path_for_executable` will now first check `$CARGO_HOME` before falling back to searching `$PATH`.
rustup is the recommended way to manage rust toolchains, therefore should be picked before the
system toolchain.
fix: Recompiles due to RUSTC_BOOTSTRAP
Some packages (e.g. thiserror) force a recompile if the value of the `RUSTC_BOOTSTRAP` env var changes. RA sets the variable to 1 in order to enable rustc / cargo unstable options. This causes flapping recompiles when building outside of RA.
Fixes#15057
internal: Optimize salsa memory usage
Reduces memory on self by ~20mb for me, there is a few more mb to save here if we made LRU caching opt-in, as currently every entry in a memoized query will store an `AtomicUsize` for the LRU.
`cargo rustc -- <args>` first builds dependencies then calls `rustc <args>` for the current package. Here, we don't want to build dependencies, we just want to call `rustc --print`. An unstable `cargo rustc` `--print` command bypasses building dependencies first. This speeds up execution of this code path and ensures RA doesn't recompile dependencies with the `RUSTC_BOOTSRAP=1` env var flag set.
Note that we must pass `-Z unstable-options` twice, first to enable the `cargo` unstable `--print` flag, then later to enable the unstable `rustc` `target-spec-json` print request.
Some packages (e.g. thiserror) force a recompile if the value of the `RUSTC_BOOTSTRAP` env var changes. RA sets the variable to 1 in order to enable rustc / cargo unstable options it uses. This causes flapping recompiles when building outside of RA.
As of Cargo 1.75 the `--keep-going` flag is stable. This change uses the flag without `RUSTC_BOOTSTRAP` if the Cargo version is >= 1.75, and drops `--keep-going` otherwise. This fixes build script recompilation.
internal: Attempt to add a timeout to rustc-tests
Looks like some new test is stuck, this might help with figuring that out, though it unfortunately won't if its a chalk hang (which is the most likely)
fix: server hanging up on build script task
This should fix https://github.com/rust-lang/rust-analyzer/issues/16614, can't say for certain since it might be not 100% reproducible... We really need to replace the current workspace fetching logic, it is completely unreadable and incredibly difficult to follow. I don't really understand how the server even got to hang here honestly (I would expect it to loop re-fetching build scripts, but not hang).
Setup infra for handling auto trait bounds disabled due to perf problems
This patch updates some of the partially-implemented functions of `ChalkContext as RustIrDatabase`, namely `adt_datum()` and `impl_provided_for()`. With those, we can now correctly work with auto trait bounds and distinguish methods based on them.
Resolves#7856 (the second code; the first one is resolved by #13074)
**IMPORTANT**: I don't think we want to merge this until #7637 is resolved. Currently this patch introduces A LOT of unknown types and type mismtaches as shown below. This is because we cannot resolve items like `hashbrown::HashMap` in `std` modules, leading to auto trait bounds on them and their dependents unprovable.
|crate (from `rustc-perf@c52ee6` except for r-a)|e3dc5a588f07d6f1d3a0f33051d4af26190abe9e|HEAD of this branch|
|---|---|---|
|rust-analyzer @ e3dc5a588f |exprs: 417528, ??ty: 907 (0%), ?ty: 114 (0%), !ty: 1|exprs: 417528, ??ty: 1704 (0%), ?ty: 403 (0%), !ty: 20|
|ripgrep|exprs: 62120, ??ty: 2 (0%), ?ty: 0 (0%), !ty: 0|exprs: 62120, ??ty: 132 (0%), ?ty: 58 (0%), !ty: 11|
|webrender/webrender|exprs: 94355, ??ty: 49 (0%), ?ty: 16 (0%), !ty: 2|exprs: 94355, ??ty: 429 (0%), ?ty: 130 (0%), !ty: 7|
|diesel|exprs: 132591, ??ty: 401 (0%), ?ty: 5129 (3%), !ty: 31|exprs: 132591, ??ty: 401 (0%), ?ty: 5129 (3%), !ty: 31|
feat!: create alias when renaming an import.
Closes#15858
Implemented:
- [x] - Prevent using `reserved` keywords (e.g self) and `_`.
- [x] - Rename other modules that might be referencing the import.
- [x] - Fix "broken" tests.
- [ ] - Rename **only** "direct" references.
- [ ] - Test more cases.
Future possibilities:
1. Also support `extern crate <name>` syntax.
2. Allow aliasing `self` when it is inside an `UseTreeList`.
~3. If import path already has an alias, "rename" the alias.~
~[4. Create alias even if path is not the last path segment.](https://github.com/rust-lang/rust-analyzer/pull/16489#issuecomment-1930541697)~
feat: add non-exhaustive-let diagnostic
I want this to have a quickfix to add an else branch but I couldn't figure out how to do that, so here's the diagnostic on its own. It pretends a `let` is a match with one arm, and asks the match checking whether that match would be exhaustive.
Previously the pattern was checked based on its own type, but that was causing a panic in `match_check` (while processing e.g. `crates/hir/src/lib.rs`) so I changed it to use the initialiser's type instead, to align with the checking of actual match expressions. I think the panic can still happen, but I hear that `match_check` is going to be updated to a new version from rustc, so I'm posting this now in the hopes that the panic will magically go away when that happens.
test: include `rename_path_inside_use_tree`.
Keeps tracks the progress of the changes. 3 other tests broke with the changes
of this.
feat: rename all other usages within the current file.
feat: fix most of the implementation problems.
test: `rename_path_inside_use_tree` tests a more complicated scenario.
Commit 6a06f6f72 (Deduplicate reference search results, 2022-11-07) deduplicates references
within each definition.
There is an edge case when requesting references of a macro argument. Apparently, our
descend_into_macros() stanza in references.rs produces a cartesian product of
- references inside the macro times
- times references outside the macro.
Since the above deduplication only applies to the references within a single definition, we
return them all, leading to many redundant references.
Work around this by deduplicating definitions as well. Perhaps there is a better fix to not
produce this cartesian product in the first place; but I think at least for definitions the
problem would remain; a macro can contain multiple definitions of the same name, but since the
navigation target will be the unresolved location, it's the same for all of them.
We can't use unique() because we don't want to drop references that don't have a declaration
(though I dont' have an example for this case).
I discovered this working with the "bitflags" macro from the crate of the same name.
Fixes#16357
This mostly works well, and eliminates a couple of delayed bugs.
One annoying thing is that we should really also add an
`ErrorGuaranteed` to `proc_macro::bridge::LitKind::Err`. But that's
difficult because `proc_macro` doesn't have access to `ErrorGuaranteed`,
so we have to fake it.
Add completions to show only traits in trait `impl` statement
This is prerequisite PR for adding the assist mentioned in #12500
P.S: If wanted, I will add the implementation of the assist in this PR as well.
Implement `literal_from_str` for proc macro server
Closes#16233
Todos and unanswered questions:
- [x] Is this the correct approach? Can both the legacy and `rust_analyzer_span` servers depend on the `syntax` crate?
- [ ] How should we handle suffixes for string literals? It doesn't seem like `rust-analyzer` preservers suffix information after parsing.
- [x] Why are the `expect` tests failing? Specifically `test_fn_like_macro_clone_literals`
Substitute $saved_file in custom check commands
If the custom command has a $saved_file placeholder, and we know the file being saved, replace the placeholder and run a check command.
If there's a placeholder and we don't know the saved file, do nothing.
This is a simplified version of #15381, which I hope is easier to review.
feat: Introduce term search to rust-analyzer
# Introduce term search to `rust-analyzer`
_I've marked this as draft as there might be some shortcomings, please point them out so I can fix them. Otherwise I think it is kind of ready as I think I'll rather introduce extra functionality in follow up PRs._
Term search (or I guess expression search for rust) is a technique to generate code by basically making the types match.
Consider the following program
```rust
fn wrap(arg: i32) -> Option<i32> {
todo!();
}
```
From the types of values in scope and constructors of `Option`, we can produce the expected result of wrapping the argument in `Option`
Dependently typed languages such as `Idris2` and `Agda` have similar tools to help with proofs, but this can be also used in everyday development as a "auto-complete".
# Demo videos
https://github.com/rust-lang/rust-analyzer/assets/19900308/7b68a1b7-7dba-4e31-9221-6c7485e77d88https://github.com/rust-lang/rust-analyzer/assets/19900308/0fae530a-aabb-4b28-af71-e19f8d3d64b2
# What does it currently do
- It works well with locals, free functions, type constructors and non-static impl methods that take items by value.
- Works with functions/methods that take shared references, but not with unique references (very conservative).
- Can handle projections to struct fields (eg. `foo.bar.baz`) but this might me more conservative than it has to be to avoid conflicting with borrow checker
- Should create only valid programs (no type / borrow checking errors). Tested with `rust-analyzer analysis-stats /path/to/ripgrep/Cargo.toml --run-term-search --validate-term-search` (basically running `cargo check` on all of the generated programs and only error seems to be due to type inference which is more of issue of testing method.
# Performace / fitness
```txt
ripgrep (latest)
Tail Expr syntactic hits: 130/1692 (7%)
Tail Exprs found: 523/1692 (30%)
Term search avg time: 9ms
Term search: 15.64s, 97ginstr, 8mb
rust-analyzer (on this branch)
Tail Expr syntactic hits: 804/13860 (5%)
Tail Exprs found: 6757/13860 (48%)
Term search avg time: 78ms
Term search: 1088.23s, 6765ginstr, 98mb
```
Highly generic code seems to blow up the search space so currently the amount of generics allowed is functions/methods is limited down to 0 (1 didn't give much improvement and 2 is already like 0.5+s search time)
# Plans for the future (not in this PR)
- ``~~Add impl methods that do not take `self` type (should be quite straight forward)~~ Done
- Be smarter (aka less restrictive) about borrow checking - this seems quite hard but since the current approach is rather naive I think some easy improvement is available.
- ``~~See if it works as a autocomplete while typing~~ Done
_Feel free to ask questions / point of shortcoming either here or on Zulip, I'll be happy to address them. I'm doing this as part of my MSc thesis so I'll be working on it till summer anyway 😄_
If the custom command has a $saved_file placeholder, and we know the
file being saved, replace the placeholder and then run a check command.
If there's a placeholder and we don't know the saved file, do nothing.
feat: ignored and disabled macro expansion
Supersedes #15117, I was having some conflicts after a rebase and since I didn't remember much of it I started clean instead.
The end result is pretty much the same as the linked PR, but instead of proc macro lookups, I marked the expanders that explicitly cannot be expanded and we shouldn't even attempt to do so.
## Unresolved questions
- [ ] I introduced a `DISABLED_ID` next to `DUMMY_ID` in `hir-expand`'s `ProcMacroExpander`, that is effectively exactly the same thing with slightly different semantics, dummy macros are not (yet) expanded probably due to errors, while not expanding disabled macros is part of the usual flow. I'm not sure if it's the right way to handle this, I also thought of just adding a flag instead of replacing the macro ID, so that the disabled macro can still be expanded for any reason if needed.
internal: tool discovery prefers sysroot tools
Fixes https://github.com/rust-lang/rust-analyzer/issues/15927, Fixes https://github.com/rust-lang/rust-analyzer/issues/16523
After this PR we will look for `cargo` and `rustc` in the sysroot if it was succesfully loaded instead of using the current lookup scheme. This should be more correct than the current approach as that relies on the working directory of the server binary or loade workspace, meaning it can behave a bit odd wrt overrides.
Additionally, rust-project.json projects now get the target data layout set so there should be better const eval support now.
