4227: Report invalid, nested, multi-segment crate-paths r=matklad a=djrenren
There was a bug in the previous path-validating code that didn't detect multi-segment paths that started with `crate`.
```rust
// Successfully reported
use foo::{crate};
// BUG: was not being reported
use foo::{crate::bar};
```
This was due to my confusion about path-associativity. That is, the path with no qualifier is the innermost path, not the outermost. I've updated the code with a lot of comments to explain what's going on.
This bug was discovered when I found an erroneous `ok` test which I reported here:
https://github.com/rust-analyzer/rust-analyzer/issues/4226
This test now fails and has been modified, hopefully in the spirit of the original test, to be correct. Sorry about submitting the bug in the first place!
Co-authored-by: John Renner <john@jrenner.net>
4178: Validate the location of `crate` in paths r=matklad a=djrenren
**This solution does not fully handle `use` statements. See below**
This pull requests implements simple validation of usages of the `crate` keyword in `Path`s. Specifically it validates that:
- If a `PathSegment` is starts with the `crate` keyword, it is also the first segment of the `Path`
- All other usages of `crate` in `Path`s are considered errors.
This aligns with `rustc`'s rules. Unlike rustc this implementation does not issue a special error message in the case of `::crate` but it does catch the error.
Furthermore, this change does not cover all error cases. Specifically the following is not caught:
```rust
use foo::{crate}
```
This is because this check is context sensitive. From an AST perspective, `crate` is the root of the `Path`. Only by inspecting the full `UseItem` do we see that it is not in fact the root. This problem becomes worse because `UseTree`s are allowed to be arbitrarily nested:
```rust
use {crate, {{crate, foo::{crate}}}
```
So this is a hard problem to solve without essentially a breadth-first search. In a traditional compiler, I'd say this error is most easily found during the AST -> HIR conversion pass but within rust-analyzer I'm not sure where it belongs.
Under the implementation in this PR, such errors are ignored so we're *more correct* just not *entirely correct*.
Co-authored-by: John Renner <john@jrenner.net>
4119: Cache proc-macro dlls r=matklad a=edwin0cheng
This PR try to fix a deadlock in proc-macro srv by not unloading dlls.
Currently we load and unload dlls for each request, however rustc TLS is leaky , such that if we do it a lot of times, all TLS index will be consumed and it will be deadlocked inside panic (it is because panic itself is using TLS too).
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4204: Use specific pattern when translating if-let-else to match r=matklad a=matklad
We *probably* should actually use the same machinery here, as we do
for fill match arms, but just special-casing options and results seems
to be a good first step.
bors r+
🤖
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
We *probably* should actually use the same machinery here, as we do
for fill match arms, but just special-casing options and results seems
to be a good first step.
4161: lsp-types 0.74 r=kjeremy a=kjeremy
* Fixes a bunch of param types to take partial progress into account.
* Will allow us to support insert/replace text in completions
Co-authored-by: kjeremy <kjeremy@gmail.com>
4193: Make it impossible to forget to add a semantic token type / modifier r=kjeremy a=matklad
bors r+
🤖
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
4173: Use core instead of std for builtin derive macros r=edwin0cheng a=edwin0cheng
Fixed#4087.
We can't use `$crate` here right now because :
1. We have to able to detect `macro` 2.0 in collecting phase for finding `rustc_builtin_macro` attrs.
2. And we have to make hygiene works for builtin derive macro.
r= @flodiebold
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
So e.g. if we have `fn foo<T: SomeTrait<u32>>() -> T::Item`, we want to lower
that to `<T as SomeTrait<u32>>::Item` and not `<T as SomeTrait<_>>::Item`.
4113: Support returning non-hierarchical symbols r=matklad a=kjeremy
If `hierarchicalDocumentSymbolSupport` is not true in the client capabilites
then it does not support the `DocumentSymbol[]` return type from the
`textDocument/documentSymbol` request and we must fall back to `SymbolInformation[]`.
This is one of the few requests that use the client capabilities to
differentiate between return types and could cause problems for clients.
See https://github.com/microsoft/language-server-protocol/pull/538#issuecomment-442510767 for more context.
Found while looking at #144
4136: add support for cfg feature attributes on expression #4063 r=matklad a=bnjjj
close issue #4063
4141: Fix typo r=matklad a=Veetaha
4142: Remove unnecessary async from vscode language client creation r=matklad a=Veetaha
Co-authored-by: kjeremy <kjeremy@gmail.com>
Co-authored-by: Benjamin Coenen <5719034+bnjjj@users.noreply.github.com>
Co-authored-by: veetaha <veetaha2@gmail.com>
4139: Add check for rules that no repetition which could match an empty token r=matklad a=edwin0cheng
Fix#4103
for `/ui/issues/issue-57597.rs`
This is `ParseError` of the macro rules , because it is how rustc handle it :
a58b1ed44f/src/librustc_expand/mbe/macro_rules.rs (L558)
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4133: main: eagerly prime goto-definition caches r=matklad a=BurntSushi
This commit eagerly primes the caches used by goto-definition by
submitting a "phantom" goto-definition request. This is perhaps a bit
circuitous, but it does actually get the job done. The result of this
change is that once RA is finished its initial loading of a project,
goto-definition requests are instant. There don't appear to be any more
surprise latency spikes.
This _partially_ addresses #1650 in that it front-loads the latency of the
first goto-definition request, which in turn makes it more predictable and
less surprising. In particular, this addresses the use case where one opens
the text editor, starts reading code for a while, and only later issues the
first goto-definition request. Before this PR, that first goto-definition request
is guaranteed to have high latency in any reasonably sized project. But
after this PR, there's a good chance that it will now be instant.
What this _doesn't_ address is that initial loading time. In fact, it makes it
longer by adding a phantom goto-definition request to the initial startup
sequence. However, I observed that while this did make initial loading
slower, it was overall a somewhat small (but not insignificant) fraction
of initial loading time.
-----
At least, the above is what I _want_ to do. The actual change in this PR is just a proof-of-concept. I came up with after an evening of printf-debugging. Once I found the spot where this cache priming should go, I was unsure of how to generate a phantom input. So I just took an input I knew worked from my printf-debugging and hacked it in. Obviously, what I'd like to do is make this more general such that it will always work.
I don't know whether this is the "right" approach or not. My guess is that there is perhaps a cleaner solution that more directly primes whatever cache is being lazily populated rather than fudging the issue with a phantom goto-definition request.
I created this as a draft PR because I'd really like help making this general. I think whether y'all want to accept this patch is perhaps a separate question. IMO, it seems like a good idea, but to be honest, I'm happy to maintain this patch on my own since it's so trivial. But I would like to generalize it so that it will work in any project.
My thinking is that all I really need to do is find a file and a token somewhere in the loaded project, and then use that as input. But I don't quite know how to connect all the data structures to do that. Any help would be appreciated!
cc @matklad since I've been a worm in your ear about this problem. :-)
Co-authored-by: Andrew Gallant <jamslam@gmail.com>
This commit makes RA more aggressive about eagerly priming the caches.
In particular, this fixes an issue where even after RA was done priming
its caches, an initial goto-definition request would have very high
latency. This fixes that issue by requesting syntax highlighting for
everything. It is presumed that this is a tad wasteful, but not overly
so.
This commit also tweaks the logic that determines when the cache is
primed. Namely, instead of just priming it when the state is loaded
initially, we attempt to prime it whenever some state changes. This
fixes an issue where if a modification notification is seen before cache
priming is done, it would stop the cache priming early.
4134: Special case for empty comments in doc comment kind r=matklad a=edwin0cheng
Part of #4103
Fix `ui/empty/empty-comment.rs macros`
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4125: Avoid lossy OsString conversions r=matklad a=lnicola
This is a bit invasive, and perhaps for not much benefit since non-UTF-8 environment variables don't work anyway.
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
4128: Include correct item path for variant completions r=matklad a=jonas-schievink
The test would previously suggest `E::V`, which is not enough to name the variant as the enum is in a module. Now it correctly suggests the full path `m::E::V`.
Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
3998: Make add_function generate functions in other modules via qualified path r=matklad a=TimoFreiberg
Additional feature for #3639
- [x] Add tests for paths with more segments
- [x] Make generating the function in another file work
- [x] Add `pub` or `pub(crate)` to the generated function if it's generated in a different module
- [x] Make the assist jump to the edited file
- [x] Enable file support in the `check_assist` helper
4006: Syntax highlighting for format strings r=matklad a=ltentrup
I have an implementation for syntax highlighting for format string modifiers `{}`.
The first commit refactors the changes in #3826 into a separate struct.
The second commit implements the highlighting: first we check in a macro call whether the macro is a format macro from `std`. In this case, we remember the format string node. If we encounter this node during syntax highlighting, we check for the format modifiers `{}` using regular expressions.
There are a few places which I am not quite sure:
- Is the way I extract the macro names correct?
- Is the `HighlightTag::Attribute` suitable for highlighting the `{}`?
Let me know what you think, any feedback is welcome!
Co-authored-by: Timo Freiberg <timo.freiberg@gmail.com>
Co-authored-by: Leander Tentrup <leander.tentrup@gmail.com>
Co-authored-by: Leander Tentrup <ltentrup@users.noreply.github.com>
If `hierarchicalDocumentSymbolSupport` is not true in the client capabilites
then it does not support the `DocumentSymbol[]` return type from the
`textDocument/documentSymbol` request and we must fall back to `SymbolInformation[]`.
This is one of the few requests that use the client capabilities to
differentiate between return types and could cause problems for clients.
See https://github.com/microsoft/language-server-protocol/pull/538#issuecomment-442510767 for more context.
Found while looking at #144
4101: Panic proc macro srv if read request failed r=matklad a=edwin0cheng
This PR fixed a bug when the rust-analyzer is killed suddenly, the `rust-analyzer proc-macro` will become stale.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
I might be reading this wrong, but it looks like we are setting it to
essentially arbitrary string at the moment, as there are no defined
order on the items in the *set* of completions.
4106: Fix wrong substitution code r=matklad a=flodiebold
We need to shift in when we're substituting inside a binder.
This should fix#4053 (it doesn't fix the occasional overflow that also occurs on the Diesel codebase though).
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
We need to shift in when we're substituting inside a binder.
This should fix#4053 (it doesn't fix the occasional overflow that also occurs
on the Diesel codebase though).
3954: Improve autocompletion by looking on the type and name r=matklad a=bnjjj
This tweet (https://twitter.com/tjholowaychuk/status/1248918374731714560) gaves me the idea to implement that in rust-analyzer.
Basically for this first example I made some examples when we are in a function call definition. I look on the parameter list to prioritize autocompletions for the same types and if it's the same type + the same name then it's displayed first in the completion list.
So here is a draft, first step to open a discussion and know what you think about the implementation. It works (cf tests) but maybe I can make a better implementation at some places. Be careful the code needs some refactoring to be better and concise.
PS: It was lot of fun writing this haha
Co-authored-by: Benjamin Coenen <5719034+bnjjj@users.noreply.github.com>
4111: Record definitions in `extern` blocks r=jonas-schievink a=jonas-schievink
Enables completion of extern functions and statics.
Closes https://github.com/rust-analyzer/rust-analyzer/issues/3711
4112: Add Launch configuration for release build r=matklad a=jonas-schievink
The debug build takes very long until I can test anything useful, with the release build it's much quicker. Add another Run configuration for it.
Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
In general, there should be no reason to call `.to_string_lossy`.
If you want to display the path, use `.display()`.
If you want to pass the path to an OS API (like std::process::Command)
than use `PathBuf` or `OsString`.
4065: Complete unqualified enum names in patterns and expressions r=matklad a=nathanwhit
This PR implements the completion described in #4014.
The result looks like so for patterns:
<img width="542" alt="Screen Shot 2020-04-20 at 3 53 55 PM" src="https://user-images.githubusercontent.com/17734409/79794010-8f529400-831f-11ea-9673-f838aa9bc962.png">
and for `expr`s:
<img width="620" alt="Screen Shot 2020-04-21 at 3 51 24 PM" src="https://user-images.githubusercontent.com/17734409/79908784-d73ded80-83e9-11ea-991d-921f0cb27e6f.png">
I'm not confident that the completion text itself is very robust, as it will unconditionally add completions for enum variants with the form `Enum::Variant`. This means (I believe) it would still suggest `Enum::Variant` even if the local name is changed i.e. `use Enum as Foo` or the variants are brought into scope such as through `use Enum::*`.
Co-authored-by: nathanwhit <nathan.whitaker01@gmail.com>
4094: proc_macro: add ability to log to stderr and view output in vscode r=matklad a=Veetaha
r? @edwin0cheng
Co-authored-by: veetaha <veetaha2@gmail.com>
4092: feat: run ignored tests r=matklad a=hdevalke
I started making some exercices on https://exercism.io/ and a lot of test have the `#[ignore]` attribute.
The `Run Test|Debug` code lens show up, but running the test results in:
```
running 1 test
test test_one_piece ... ignored
test result: ok. 0 passed; 0 failed; 1 ignored; 0 measured; 5 filtered out
```
This pull request adds the `--ignored` flag if needed.
Co-authored-by: Hannes De Valkeneer <hannes@de-valkeneer.be>
4076: Improve remove derive attrs r=matklad a=edwin0cheng
This PR implemented a proper `remove_derive_attrs` function which **merely** remove any `#[derive(*)]` attributes.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4038: Group generated ast boilerplate apart from the interesting part r=matklad a=Veetaha
Boilerplate `AstNode` and `From` impls are moved to the end further from the interesting part in `generated.rs`
Co-authored-by: veetaha <veetaha2@gmail.com>
The fix is admittedly quit literally just papering over.
Long-term, I see two more principled approaches:
* we switch to a fully tree-based impl, without parse . to_string
step; with this approach, there shouldn't be any panics. The results
might be nonsensical, but so was the original input.
* we preserve the invariant that re-parsing constructed node is an
identity, and make all the `make_xxx` method return an `Option`.
closes#4044
Detailed changes:
1) Implement a lexer for string literals that divides the string in format specifier `{}` including the format specifier modifier.
2) Adapt syntax highlighting to add ranges for the detected sequences.
3) Add a test case for the format string syntax highlighting.
4035: Convert bool to ident instead of literal in mbe r=matklad a=edwin0cheng
Fixed#1249
Currently we treat boolean literal as `tt::Literal` , which makes parsing $lit:lit matcher easily.
However, proc-macro2 treat boolean literal as `ident` :
4173a21dc4/src/lib.rs (L939)
OT: I am quite happy we finally need to fix this bug :)
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4034: Add semantic tag for unresolved references r=matklad a=matklad
This is a quick way to implement unresolved reference diagnostics.
For example, adding to VS Code config
"editor.tokenColorCustomizationsExperimental": {
"unresolvedReference": "#FF0000"
},
will highlight all unresolved refs in red.
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
This is a quick way to implement unresolved reference diagnostics.
For example, adding to VS Code config
"editor.tokenColorCustomizationsExperimental": {
"unresolvedReference": "#FF0000"
},
will highlight all unresolved refs in red.
3894: Match check enum record r=flodiebold a=JoshMcguigan
This PR implements match statement exhaustiveness checking for record type enums.
It also make a minor addition to the test infrastructure to allow testing against a single diagnostic, so you can be sure your test is triggering (or not) whichever diagnostic you expect.
Co-authored-by: Josh Mcguigan <joshmcg88@gmail.com>
4029: Fix various proc-macro bugs r=matklad a=edwin0cheng
This PRs does the following things:
1. Fixed#4001 by splitting `LIFETIME` lexer token to two mbe tokens. It is because rustc token stream expects `LIFETIME` as a combination of punct and ident, but RA `tt:TokenTree` treats it as a single `Ident` previously.
2. Fixed#4003, by skipping `proc-macro` for completion. It is because currently we don't have `AstNode` for `proc-macro`. We would need to redesign how to implement `HasSource` for `proc-macro`.
3. Fixed a bug how empty `TokenStream` merging in `proc-macro-srv` such that no L_DOLLAR and R_DOLLAR will be emitted accidentally.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
4023: Fix another crash from wrong binders r=matklad a=flodiebold
Basically, if we had something like `dyn Trait<T>` (where `T` is a type parameter) in an impl we lowered that to `dyn Trait<^0.0>`, when it should be `dyn Trait<^1.0>` because the `dyn` introduces a new binder. With one type parameter, that's just wrong, with two, it'll lead to crashes.
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
Basically, if we had something like `dyn Trait<T>` (where `T` is a type
parameter) in an impl we lowered that to `dyn Trait<^0.0>`, when it should be
`dyn Trait<^1.0>` because the `dyn` introduces a new binder. With one type
parameter, that's just wrong, with two, it'll lead to crashes.
Fixes a lot of false type mismatches.
(And as always when touching the unification code, I have to say I'm looking
forward to replacing it by Chalk's...)
It's not entirely clear what subnode ranges should mean in the
presence of macros, so let's leave them out for now. We are not using
them heavily anyway.
3996: Fix path for proc-macro in nightly / stable release r=matklad a=edwin0cheng
I messed up that I forget we use different executable names for nightly / stable release, I changed to use the current executable name instead.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
3958: Add proc-macro related config and tests r=matklad a=edwin0cheng
This PR do the following things:
1. Add cli argument `proc-macro` for running proc-macro server.
2. Added support for proc-macro in bench and analysis-stats
3. Added typescript config for proc-macros
4. Added an heavy test for proc-macros.
To test it out:
1. run `cargo xtask install --proc-macro`
2. add `"rust-analyzer.cargo.loadOutDirsFromCheck": true"` and `"rust-analyzer.procMacro.enabled": true"` in vs code config.
[Edit] Change to use `rust-analyzer proc-macro` for running proc-macro standalone process.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
3979: fix missing match arm false positive for enum with no variants r=flodiebold a=JoshMcguigan
fixes#3974
Co-authored-by: Josh Mcguigan <joshmcg88@gmail.com>
3966: Add support for bounds on associated types in trait definitions r=matklad a=flodiebold
E.g.
```rust
trait Trait {
type Item: SomeOtherTrait;
}
```
Note that these don't simply desugar to where clauses; as I understand it, where clauses have to be proved by the *user* of the trait, but these bounds are proved by the *implementor*. (Also, where clauses on associated types are unstable.)
(Another one from my recursive solver branch...)
3968: Remove format from syntax_bridge hot path r=matklad a=edwin0cheng
Although only around 1% speed up by running:
```
Measure-Command {start-process .\target\release\rust-analyzer "analysis-stats -q ." -NoNewWindow -wait}
```
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
3964: Nicer Chalk debug logs r=matklad a=flodiebold
I'm looking at a lot of Chalk debug logs at the moment, so here's a few changes to make them slightly nicer...
3965: Implement inline associated type bounds r=matklad a=flodiebold
Like `Iterator<Item: SomeTrait>`.
This is an unstable feature, but it's used in the standard library e.g. in the definition of Flatten, so we can't get away with not implementing it :)
(This is cherry-picked from my recursive solver branch, where it works better, but I did manage to write a test that works with the current Chalk solver as well...)
3967: Handle `Self::Type` in trait definitions when referring to own associated type r=matklad a=flodiebold
It was implemented for other generic parameters for the trait, but not for `Self`.
(Last one off my recursive solver branch 😄 )
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
3969: Change add_function assist to use todo!() instead of unimplemented!() r=matklad a=TimoFreiberg
In the spirit of #3935
Co-authored-by: Timo Freiberg <timo.freiberg@gmail.com>
3971: add diagnostics subcommand to rust-analyzer CLI r=JoshMcguigan a=JoshMcguigan
This PR adds a `diagnostics` subcommand to the rust-analyzer CLI. The intent is to detect all diagnostics on a workspace. It returns a non-zero status code if any error diagnostics are detected. Ideally I'd like to run this in CI against the rust analyzer project as a guard against false positives.
```
$ cargo run --release --bin rust-analyzer -- diagnostics .
```
Questions for reviewers:
1. Is this the proper way to get all diagnostics for a workspace? It seems there are at least a few ways this can be done, and I'm not sure if this is the most appropriate mechanism to do this.
2. It currently prints out the relative file path as it is collecting diagnostics, but it doesn't print the crate name. Since the file name is relative to the crate there can be repeated names, so it would be nice to print some identifier for the crate as well, but it wasn't clear to me how best to accomplish this.
Co-authored-by: Josh Mcguigan <joshmcg88@gmail.com>
3961: Fix double comma when merge imports on second line r=edwin0cheng a=IceSentry
This fixes the bug when merging imports from the second line when it already has a comma it would previously insert a comma.
There's probably a better way to check for a COMMA.
This also ends up with a weird indentation, but rust-fmt can easily deal with it so I'm not sure how to resolve that.
Closes#3832
Co-authored-by: IceSentry <c.giguere42@gmail.com>
3960: ellipsis in tuple patterns r=JoshMcguigan a=JoshMcguigan
This PR lowers ellipsis in tuple patterns. It fixes a bug in the way ellipsis were previously lowered (by replacing the ellipsis with a single `Pat::Wild` no matter how many items the `..` was taking the place of).
It also uses this new information to properly handle `..` in tuple struct patterns when perform match statement exhaustiveness checks.
While this PR provides the building blocks for match statement exhaustiveness checks for tuples, there are some additional challenges there, so that is still unimplemented (unlike tuple structs).
Co-authored-by: Josh Mcguigan <joshmcg88@gmail.com>
E.g.
```
trait Trait {
type Item: SomeOtherTrait;
}
```
Note that these don't simply desugar to where clauses; as I understand it, where
clauses have to be proved by the *user* of the trait, but these bounds are proved
by the *implementor*. (Also, where clauses on associated types are unstable.)
Like `Iterator<Item: SomeTrait>`.
This is an unstable feature, but it's used in the standard library e.g. in the
definition of Flatten, so we can't get away with not implementing it :)
