Fix: a TODO and some clippy fixes
- fix(todo): implement IntoIterator for ArenaMap<IDX, V>
- chore: remove unused method
- fix: remove useless `return`s
- fix: various clippy lints
- fix: simplify boolean test to a single negation
MIR episode 5
This PR inits drop support (it is very broken at this stage, some things are dropped multiple time, drop scopes are wrong, ...) and adds stdout support (`println!` doesn't work since its expansion is dummy, but `stdout().write(b"hello world\n")` works if you use `RA_SYSROOT_HACK`) for interpreting. There is no useful unit test that it can interpret yet, but it is a good sign that it didn't hit a major road block yet.
In MIR lowering, it adds support for slice pattern and anonymous const blocks, and some fixes so that we can evaluate `SmolStr::new_inline` in const eval. With these changes, 57 failed mir body remains.
Support `#[macro_use(name, ...)]`
This PR adds support for another form of the `macro_use` attribute: `#[macro_use(name, ...)]` ([reference]).
Note that this form of the attribute is only applicable to extern crate decls, not to mod decls.
[reference]: https://doc.rust-lang.org/reference/macros-by-example.html#the-macro_use-attribute
Parse associated return type bounds
This PR implements parser support for associated return type bounds: `T: Foo<bar(): Send>`. This PR does not implement associated return types (`T::bar(): Send`) because it's not implemented even in rustc, and also removes `(..)`-style return type notation because it has been removed in rust-lang/rust#110203 (effectively reverting #14465).
I don't plan to proactively follow this unstable feature unless an RFC is accepted and my main motivation here is to remove no-longer-valid syntax `(..)` from our parser, nevertheless adding minimal parser support so anyone interested (as can be seen in #14465) can experiment it without rust-analyzer's syntax errors.
Expand more single ident macro calls upon item collection
Addresses https://github.com/rust-lang/rust-analyzer/pull/14781#issuecomment-1546201022
I believe this (almost) brings the number of unresolved names back to pre-#14781:
|r-a version|`analysis-stats compiler/rustc` (rust-lang/rust@69fef92ab2) |
|---|---|
|pre-#14781 (b069eb720b) | exprs: 2747778, ??ty: 122236 (4%), ?ty: 107826 (3%), !ty: 728 |
| #14781 (a7944a93a1) | exprs: 2713080, ??ty: 139651 (5%), ?ty: 114444 (4%), !ty: 730 |
| with this fix | exprs: 2747871, ??ty: 122237 (4%), ?ty: 108171 (3%), !ty: 676 |
(I haven't investigated on the increase in some numbers but hopefully not too much of a problem)
This is only a temporary solution. The core problem is that we haven't fully implemented the textual scope of legacy macros. For example, we *have been* failing to resolve `foo` in the following snippet, even before #14781 or after this patch. As noted in a FIXME, we need a way to resolve names in textual scope without eager expansion during item collection.
```rust
//- /main.rs crate:main deps:lib
lib::mk_foo!();
const A: i32 = foo!();
//^^^^^^ unresolved-macro-call
//- /lib.rs crate:lib
#[macro_export]
macro_rules! mk_foo {
() => {
macro_rules! foo { () => { 42 } }
}
}
```
We've already removed non-sysroot proc macro server, which effectively
removed support for Rust <1.64.0, so this removal of fallback path
shouldn't be problem at this point.
This function/lang_item was introduced in #104321 as a temporary workaround of future lowering.
The usage and need for it went away in #104833.
After a bootstrap update, the function itself can be removed from `std`.
More core::fmt::rt cleanup.
- Removes the `V1` suffix from the `Argument` and `Flag` types.
- Moves more of the format_args lang items into the `core::fmt::rt` module. (The only remaining lang item in `core::fmt` is `Arguments` itself, which is a public type.)
Part of https://github.com/rust-lang/rust/issues/99012
Follow-up to https://github.com/rust-lang/rust/pull/110616
fix: Fix pat fragment handling in 2021 edition
Fixes https://github.com/rust-lang/rust-analyzer/issues/9055
The fix isn't that great, but we are kind of forced to do it the quick and hacky way right now since std has changed the `matches` macro to make use of this now. And for a proper fix we need to track hygiene for identifiers which is a long way off anyways
fix: Resolve `$crate` in derive paths
Paths in derive meta item list may contain any kind of paths, including those that start with `$crate` generated by macros. We need to take hygiene into account when we lower paths in the list.
This issue was identified while investigating #14607, though this patch doesn't fix the broken trait resolution.
internal: Report macro definition errors on the definition
We still report them on the call site as well for the time being, and the diagnostic doesn't know where the error in the definition comes from, but that can be done later on
Add doc-alias based completion
Closes#14406.
I adapted the parsing code from the CfgExpr parsing code, maybe there's a better abstraction for both, or attribute parsing in general. It also includes `doc(hidden)`-parsing, which means it could replace the other function.
There are a few tests for parsing.
`process_all_names` changed the most, I added some docs there to explain what happens.
Many call sites just pass an empy vec to `add_path_resolution`'s `doc_aliases`, since either it doesn't make sense to pass anything (e.g. visibility completion) or I don't know where to get them from. Shouldn't really matter, as it will just not show aliases if the vec is empty and we can extend alias completion in these cases later.
I added two tests in `special.rs` for struct name completion (which was the main thing I wanted). I also tried function and field names, but these don't work yet. I want to add those in a follow-up PR.
Normalize associated types in paths in expressions
Part of #14393
When we resolve paths in expressions (either path expressions or paths in struct expressions), there's a need of projection normalization, which `TyLoweringContext` cannot do on its own. We've been properly applying normalization for paths in struct expressions without type anchor, but not for others:
```rust
enum E {
S { v: i32 }
Empty,
}
impl Foo for Bar {
type Assoc = E;
fn foo() {
let _ = Self::Assoc::S { v: 42 }; // path in struct expr without type anchor; we already support this
let _ = <Self>::Assoc::S { v: 42 }; // path in struct expr with type anchor; resolves with this PR
let _ = Self::Assoc::Empty; // path expr; resolves with this PR
}
}
```
With this PR we correctly resolve the whole path, but we need some more tweaks in HIR and/or IDE layers to properly resolve a qualifier (prefix) of such paths and provide IDE features that are pointed out in #14393 to be currently broken.
internal: Add config to specifiy lru capacities for all queries
Might help figuring out what queries should be limited by LRU by default, as currently we only limit `parse`, `parse_macro_expansion` and `macro_expand`.
internal: Only intern blocks that declare items
We only used `BlockId` for the block defmap, so this is wasted memory. Lowering for non item declaring blocks is also cheaper now as we no longer have to fully lower a block that defines not items.
Handle trait alias definitions
Part of #2773
This PR adds a bunch of structs and enum variants for trait aliases. Trait aliases should be handled as an independent item because they are semantically distinct from traits.
I basically started by adding `TraitAlias{Id, Loc}` to `hir_def::item_tree` and iterated adding necessary stuffs until compiler stopped complaining what's missing. Let me know if there's still anything I need to add.
I'm opening up this PR for early review and stuff. I'm planning to add tests for IDE functionalities in this PR, but not type-related support, for which I put FIXME notes.
Fix associated item visibility in block-local impls
Fixes#14046
When we're resolving visibility of block-local items...
> `self` normally refers to the containing non-block module, and `super` to its parent (etc.). However, visibilities must only refer to a module in the DefMap they're written in, so we restrict them when that happens. ([link])
...unless we're resolving visibility of associated items in block-local impls, because that impl is semantically "hoisted" to the nearest (non-block) module. With this PR, we skip the adjustment for such items.
Since visibility representation of those items is modified, this PR also adjusts visibility rendering in `HirDisplay`.
[link]: a6603fc21d/crates/hir-def/src/nameres/path_resolution.rs (L101-L103)
Fix: Run doctests for structs with lifetime parameters from IDE
Fixes#14142: Doctests can't be triggered for structs with lifetimes
This MR adds lifetime parameters to the structs path for runnables so that they can be triggered from an IDE as well.
This is my first MR for rust-analyzer, please let me know if I should change something, either in code or the description here.
Beginning of MIR
This pull request introduces the initial implementation of MIR lowering and interpreting in Rust Analyzer.
The implementation of MIR has potential to bring several benefits:
- Executing a unit test without compiling it: This is my main goal. It can be useful for quickly testing code changes and print-debugging unit tests without the need for a full compilation (ideally in almost zero time, similar to languages like python and js). There is a probability that it goes nowhere, it might become slower than rustc, or it might need some unreasonable amount of memory, or we may fail to support a common pattern/function that make it unusable for most of the codes.
- Constant evaluation: MIR allows for easier and more correct constant evaluation, on par with rustc. If r-a wants to fully support the type system, it needs full const eval, which means arbitrary code execution, which needs MIR or something similar.
- Supporting more diagnostics: MIR can be used to detect errors, most famously borrow checker and lifetime errors, but also mutability errors and uninitialized variables, which can be difficult/impossible to detect in HIR.
- Lowering closures: With MIR we can find out closure capture modes, which is useful in detecting if a closure implements the `FnMut` or `Fn` traits, and calculating its size and data layout.
But the current PR implements no diagnostics and doesn't support closures. About const eval, I removed the old const eval code and it now uses the mir interpreter. Everything that is supported in stable rustc is either implemented or is super easy to implement. About interpreting unit tests, I added an experimental config, disabled by default, that shows a `pass` or `fail` on hover of unit tests (ideally it should be a button similar to `Run test` button, but I didn't figured out how to add them). Currently, no real world test works, due to missing features including closures, heap allocation, `dyn Trait` and ... so at this point it is only useful for me selecting what to implement next.
The implementation of MIR is based on the design of rustc, the data structures are almost copy paste (so it should be easy to migrate it to a possible future stable-mir), but the lowering and interpreting code is from me.
fix: Don't expand macros in the same expansion tree after overflow
This patch fixes 2 bugs:
- In `Expander::enter_expand_id()` (and in code paths it's called), we never check whether we've reached the recursion limit. Although it hasn't been reported as far as I'm aware, this may cause hangs or stack overflows if some malformed attribute macro is used on associated items.
- We keep expansion even when recursion limit is reached. Take the following for example:
```rust
macro_rules! foo { () => {{ foo!(); foo!(); }} }
fn main() { foo!(); }
```
We keep expanding the first `foo!()` in each expansion and would reach the limit at some point, *after which* we would try expanding the second `foo!()` in each expansion until it hits the limit again. This will (by default) lead to ~2^128 expansions.
This is essentially what's happening in #14074. Unlike rustc, we don't just stop expanding macros when we fail as long as it produces some tokens so that we can provide completions and other services in incomplete macro calls.
This patch provides a method that takes care of recursion depths (`Expander::within_limit()`) and stops macro expansions in the whole macro expansion tree once it detects recursion depth overflow. To be honest, I'm not really satisfied with this fix because it can still be used in unintended ways to bypass overflow checks, and I'm still seeking ways such that misuses are caught by the compiler by leveraging types or something.
Fixes#14074
Support multi-character punct tokens in MBE
Fixes#11497
In the context of MBE, consecutive puncts are parsed as multi-character punct tokens whenever possible. For example, `:::` is parsed as ``[Punct(`::`), Punct(`:`)]`` and shouldn't get matched to patterns like `: : :` or `: ::`.
We have implemented this behavior only for when we match puncts against `tt` fragments, but not when we match puncts literally. This PR extracts the multi-character punct handling procedure into a separate method and extends its support for literal matching.
For good measure, this PR adds support for `<-` token, which is still [considered as one token in rustc](e396186407/compiler/rustc_ast/src/token.rs (L249)) despite the placement syntax having been removed.
Use `rustc_safe_intrinsic` attribute to check for intrinsic safety
Instead of maintaining a list that is poorly kept in sync we can just use the attribute.
This will make new RA versions unusable with old toolchains that don't have the attribute yet. Should we keep maintaining the list as a fallback or just don't care?
This makes code more readale and concise,
moving all format arguments like `format!("{}", foo)`
into the more compact `format!("{foo}")` form.
The change was automatically created with, so there are far less change
of an accidental typo.
```
cargo clippy --fix -- -A clippy::all -W clippy::uninlined_format_args
```
Seems like these can be safely fixed. With one, I was particularly
surprised -- `Some(pats) => &**pats,` in body.rs?
```
cargo clippy --fix -- -A clippy::all -D clippy::explicit_auto_deref
```
I am not certain if this will improve performance,
but it seems having a .clone() without any need should be removed.
This was done with clippy, and manually reviewed:
```
cargo clippy --fix -- -A clippy::all -D clippy::redundant_clone
```
Compute data layout of types
cc #4091
Things that aren't working:
* Closures
* Generators (so no support for `Future` I think)
* Opaque types
* Type alias and associated types which may need normalization
Things that show wrong result:
* ~Enums with explicit discriminant~
* SIMD types
* ~`NonZero*` and similar standard library items which control layout with special attributes~
At the user level, I didn't put much work, since I wasn't confident about what is the best way to present this information. Currently it shows size and align for ADTs, and size, align, offset for struct fields, in the hover, similar to clangd. I used it some days and I feel I liked it, but we may consider it too noisy and move it to an assist or command.
Mega-sync from `rust-lang/rust`
This essentially implements `@oli-obk's` suggestion here https://github.com/rust-lang/rust-analyzer/pull/13459#issuecomment-1297285607, with `@eddyb's` help.
This PR is equivalent to 14 syncs (back and forth) between `rust-lang/rust` and `rust-lang/rust-analyzer`.
Working from this list (from bottom to top):
```
(x) a2a1d9954⬆️ rust-analyzer
(x) 79923c382⬆️ rust-analyzer
(x) c60b1f641⬆️ rust-analyzer
(x) 8807fc4cc⬆️ rust-analyzer
(x) a99a48e78⬆️ rust-analyzer
(x) 4f55ebbd4⬆️ rust-analyzer
(x) f5fde4df4⬆️ rust-analyzer
(x) 459bbb422⬆️ rust-analyzer
(x) 65e1dc4d9⬆️ rust-analyzer
(x) 3e358a682⬆️ rust-analyzer
(x) 31519bb39⬆️ rust-analyzer
(x) 8231fee46⬆️ rust-analyzer
(x) 22c8c9c40⬆️ rust-analyzer
(x) 9d2cb42a4⬆️ rust-analyzer
```
(This listed was assembled by doing a `git subtree push`, which made a branch, and looking at the new commits in that branch, picking only those that were `⬆️ rust-analyzer` commits)
We used the following commands to simulate merges in both directions:
```shell
TO_MERGE=22c8c9c40 # taken from the list above, bottom to top
git merge --no-edit --no-ff $TO_MERGE
git merge --no-edit --no-ff $(git -C ../rust log --pretty=format:'%cN | %s | %ad => %P' | rg -m1 -F "$(git show --no-patch --pretty=format:%ad $TO_MERGE)" | tee /dev/stderr | rg '.* => \S+ (\S+)$' --replace '$1')
```
We encountered no merge conflicts that Git wasn't able to solve by doing it this way.
Here's what the commit graph looks like (as shown in the Git Lens VSCode extension):
<img width="1345" alt="image" src="https://user-images.githubusercontent.com/7998310/203984523-7c1a690a-8224-416c-8015-ed6e49667066.png">
This PR closes#13459
## Does this unbreak `rust->ra` syncs?
Yes, here's how we tried:
In `rust-analyzer`:
* check out `subtree-fix` (this PR's branch)
* make a new branch off of it: `git checkout -b subtree-fix-merge-test`
* simulate this PR getting merged with `git merge master`
In `rust`:
* pull latest master
* make a new branch: `git checkout -b test-change`
* mess with rust-analyzer (I added a comment to `src/tools/rust-analyzer/Cargo.toml`)
* commit
* run `git subtree push -P src/tools/rust-analyzer ra-local final-sync` (this follows the [Clippy sync guide](https://doc.rust-lang.org/nightly/clippy/development/infrastructure/sync.html))
This created a `final-sync` branch in `rust-analyzer`.
In `rust-analyzer`:
* `git merge --no-ff final-sync` (this follows the [Clippy sync guide](https://doc.rust-lang.org/nightly/clippy/development/infrastructure/sync.html))
Now `git log` in `rust-analyzer` shows this:
```
commit 460128387e46ddfc2b95921b2d7f6e913a3d2b9f (HEAD -> subtree-fix-merge-test)
Merge: 0513fc02a 9ce6a734f
Author: Amos Wenger <amoswenger@gmail.com>
Date: Fri Nov 25 13:28:24 2022 +0100
Merge branch 'final-sync' into subtree-fix-merge-test
commit 0513fc02a08ea9de952983624bd0a00e98044b36
Merge: 38c98d1ff6918009fe
Author: Amos Wenger <amoswenger@gmail.com>
Date: Fri Nov 25 13:28:02 2022 +0100
Merge branch 'master' into subtree-fix-merge-test
commit 9ce6a734f37ef8e53689f1c6f427a9efafe846bd (final-sync)
Author: Amos Wenger <amoswenger@gmail.com>
Date: Fri Nov 25 13:26:26 2022 +0100
Mess with rust-analyzer just for fun
```
And `git diff 0513fc02a08ea9de952983624bd0a00e98044b36` shows this:
```patch
diff --git a/Cargo.toml b/Cargo.toml
index 286ef1e7d..c9e24cd19 100644
--- a/Cargo.toml
+++ b/Cargo.toml
`@@` -32,3 +32,5 `@@` debug = 0
# ungrammar = { path = "../ungrammar" }
# salsa = { path = "../salsa" }
+
+# lol, hi
```
## Does this unbreak `ra->rust` syncs?
Yes, here's how we tried.
From `rust`:
* `git checkout -b sync-from-ra`
* `git subtree pull -P src/tools/rust-analyzer ra-local subtree-fix-merge-test` (this is adapted from the [Clippy sync guide](https://doc.rust-lang.org/nightly/clippy/development/infrastructure/sync.html#performing-the-sync-from-clippy-to-rust-langrust), you would normally use `ra-upstream master` but we're simulating things here)
A commit editor pops up, there was no merge conflicts.
## How do we prevent this from happening again?
Like `@bjorn3` said in https://github.com/rust-lang/rust-analyzer/pull/13459#issuecomment-1293587848
> Whenever syncing from rust-analyzer -> rust you have to immediately sync the merge commit from rust -> rust-analyzer to prevent merge conflicts in the future.
But if we get it wrong again, at least now we have a not-so-painful way to fix it.
fix: check visibility of each path segment
Upon path resolution, we have not been checking if every def pointed to by each segment of the path is visible from the original module. This leads to incorrect import resolutions, in particular when one uses glob imports and names collide.
There is decent amount of changes in this PR because:
- some of our tests were not correct in terms of visibility
- I left several basic nameres tests as-is (with expect test updated) since I thought it would be nice to ensure we don't resolve defs that are not visible.
- `fix_visibility` assist relied on `Semantics::resolve_path()`, which uses the name resolution procedure I'm fixing and wouldn't be able to "see through" the items with strict visibility with this patch
The first commit is the gist of the fix itself.
Fixes#10991Fixes#11473Fixes#13252
Fix `tt::Punct`'s spacing calculation
Fixes#13499
We currently set a `tt::Punct`'s spacing to `Spacing::Joint` unless its next token is a trivia (i.e. whitespaces or comment). As I understand it, rustc only [sets `Spacing::Joint` if the next token is an operator](5b3e909075/compiler/rustc_parse/src/lexer/tokentrees.rs (L77-L78)) and we should follow it to guarantee the consistent behavior of proc macros.
Use $crate instead of std for panic builtin_fn_macro
This should be closer to the expected output and gets rid of a few type mismatches in rustc/library
feat: type inference for generators
This PR implements basic type inference for generator and yield expressions.
Things not included in this PR:
- Generator upvars and generator witnesses are not implemented. They are only used to determine auto trait impls, so basic type inference should be fine without them, but method resolutions with auto trait bounds may not be resolved correctly.
Open questions:
- I haven't (yet) implemented `HirDisplay` for `TyKind::Generator`, so generator types are just shown as "{{generator}}" (in tests, inlay hints, hovers, etc.), which is not really nice. How should we show them?
- I added moderate amount of stuffs to minicore. I especially didn't want to add `impl<T> Deref for &T` and `impl<T> Deref for &mut T` exclusively for tests for generators; should I move them into the test fixtures or can they be placed in minicore?
cc #4309
