5015: Account for updated module ids when determining whether a resolution is changed r=matklad a=Nashenas88
Fixes#4943
5027: Make Debug less verbose for VfsPath and use Display in analysis-stats r=matklad a=lnicola
5028: Remove namedExports config r=matklad a=lnicola
Fixes a warning:
```
(!) Plugin commonjs: The namedExports option from "@rollup/plugin-commonjs" is deprecated. Named exports are now handled automatically.
```
Co-authored-by: Paul Daniel Faria <Nashenas88@users.noreply.github.com>
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
4940: Add support for marking doctest items as distinct from normal code r=ltentrup a=Nashenas88
This adds `HighlightTag::Generic | HighlightModifier::Injected` as the default highlight for all elements within a doctest. Please feel free to suggest that a new tag be created or a different one used.
Fixes#4929Fixes#4939
Co-authored-by: Paul Daniel Faria <Nashenas88@users.noreply.github.com>
Co-authored-by: Paul Daniel Faria <nashenas88@users.noreply.github.com>
4937: Allow overriding rust-analyzer display version r=matklad a=oxalica
The build script invokes `git` for version information which is displayed when rust-analyzer is called with `--version`. But in build environment without `git` or when the source code is not a git repo, there's no way to manually specify the version information.
This patch respects environment variable ~`REV`~ `RUST_ANALYZER_REV` in compile time for overriding.
Related: https://github.com/NixOS/nixpkgs/pull/90976
Co-authored-by: oxalica <oxalicc@pm.me>
4903: Add highlighting support for doc comments r=matklad a=Nashenas88
The language server protocol includes a semantic modifier for documentation. This change exports that modifier for doc comments so users can choose to highlight them differently compared to regular comments.
Example:
<img width="375" alt="Screen Shot 2020-06-16 at 10 34 14 AM" src="https://user-images.githubusercontent.com/1673130/84788271-f6599580-afbc-11ea-96e5-7a0215da620b.png">
CC @woody77
Co-authored-by: Paul Daniel Faria <Nashenas88@users.noreply.github.com>
4934: Remove special casing for library symbols r=matklad a=matklad
We might as well handle them internally, via queries.
I am not sure, but it looks like the current LibraryData setup might
even predate salsa? It's not really needed and creates a bunch of
complexity.
bors r+
🤖
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
We might as well handle them internally, via queries.
I am not sure, but it looks like the current LibraryData setup might
even predate salsa? It's not really needed and creates a bunch of
complexity.
4927: Better encapsulate reverse-mapping of files to cargo targets r=matklad a=matklad
We need to find a better way to do it...
CrateGraph by itself is fine, CargoWorkspace as well, but the mapping
between the two seems arbitrary...
bors r+
🤖
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
4913: Remove debugging code for incremental sync r=matklad a=lnicola
4915: Inspect markdown code fences to determine whether to apply syntax highlighting r=matklad a=ltentrup
Fixes#4904
4916: Warnings as hint or info r=matklad a=GabbeV
Fixes#4229
This PR is my second attempt at providing a solution to the above issue. My last PR(#4721) had to be rolled back(#4862) due to it overriding behavior many users expected. This PR solves a broader problem while trying to minimize surprises for the users.
### Problem description
The underlying problem this PR tries to solve is the mismatch between [Rustc lint levels](https://doc.rust-lang.org/rustc/lints/levels.html) and [LSP diagnostic severity](https://microsoft.github.io/language-server-protocol/specification#diagnostic). Rustc currently doesn't have a lint level less severe than warning forcing the user to disable warnings if they think they get to noisy. LSP however provides two severitys below warning, information and hint. This allows editors like VSCode to provide more fine grained control over how prominently to show different diagnostics.
Info severity shows a blue squiggly underline in code and can be filtered separately from errors and warnings in the problems panel.
Hint severity doesn't show up in the problems panel at all and only show three dots under the affected code or just faded text if the diagnostic also has the unnecessary tag.
### Solution
The solution provided by this PR allows the user to configure lists of of warnings to report as info severity and hint severity respectively. I purposefully only convert warnings and not errors as i believe it's a good idea to have the editor show the same severity as the compiler as much as possible.
### Open questions
#### Discoverability
How do we teach this to new and existing users? Should a section be added to the user manual? If so where and what should it say?
#### Defaults
Other languages such as TypeScript report unused code as hint by default. Should rust-analyzer similarly report some problems as hint/info by default?
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
Co-authored-by: Leander Tentrup <leander.tentrup@gmail.com>
Co-authored-by: Gabriel Valfridsson <gabriel.valfridsson@gmail.com>
Anchoring to the SourceRoot wont' work if the path is absolute:
#[path = "/tmp/foo.rs"]
mod foo;
Anchoring to a file will.
However, we *should* anchor, instead of just producing an abs path.
I can imagine a situation where, for example, rust-analyzer processes
crates from different machines (or, for example, from in-memory git
branch), where the same absolute path in different crates might refer
to different files in the end!
4784: Change management of test cfg to better support json projects r=Nashenas88 a=Nashenas88
This helps support json projects where they can decide whether to add the `test` cfg or not. One alternative is to add support for marking json project crates as a sysroot crate, and adding logic to remove the `test` cfg in those cases. In my opinion, that option gives less flexibility to json projects and leads to more functionality that needs to be maintained.
Fixes#4508
cc @woody77
Co-authored-by: Paul Daniel Faria <Nashenas88@users.noreply.github.com>
Co-authored-by: Paul Daniel Faria <nashenas88@users.noreply.github.com>
4729: Hover actions r=matklad a=vsrs
This PR adds a `hoverActions` LSP extension and a `Go to Implementations` action as an example:
4748: Add an `ImportMap` and use it to resolve item paths in `find_path` r=matklad a=jonas-schievink
Removes the "go faster" queries I added in https://github.com/rust-analyzer/rust-analyzer/pull/4501 and https://github.com/rust-analyzer/rust-analyzer/pull/4506. I've checked this PR on the rustc code base and the assists are still fast.
This should fix https://github.com/rust-analyzer/rust-analyzer/issues/4515.
Note that this does introduce a change in behavior: We now always refer to items defined in external crates using paths through the external crate. Previously we could also use a local path (if for example the extern crate was reexported locally), as seen in the changed test. If that is undesired I can fix that, but the test didn't say why the previous behavior would be preferable.
Co-authored-by: vsrs <vit@conrlab.com>
Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
Co-authored-by: Jonas Schievink <jonas.schievink@ferrous-systems.com>
4740: Remove unneeded "./" prefix affecting error messages r=kjeremy a=dtolnay
I noticed this in the error in the commit message of https://github.com/rust-analyzer/rust-analyzer/pull/4739.
Before:
```console
error[E0599]: no method named `initialize_finish` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/./src/bin/main.rs:99:16
|
99 | connection.initialize_finish(initialize_id, initialize_result)?;
| ^^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
```
After:
```console
error[E0599]: no method named `initialize_finish` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/src/bin/main.rs:99:16
|
99 | connection.initialize_finish(initialize_id, initialize_result)?;
| ^^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
```
```diff
- --> crates/rust-analyzer/./src/bin/main.rs:99:16
+ --> crates/rust-analyzer/src/bin/main.rs:99:16
```
Co-authored-by: David Tolnay <dtolnay@gmail.com>
Before:
error[E0599]: no method named `initialize_finish` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/./src/bin/main.rs:99:16
|
99 | connection.initialize_finish(initialize_id, initialize_result)?;
| ^^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
After:
error[E0599]: no method named `initialize_finish` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/src/bin/main.rs:99:16
|
99 | connection.initialize_finish(initialize_id, initialize_result)?;
| ^^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
My codebase already depended on lsp-server and introducing a dependency
on rust-analyzer failed at first because it assumes some functions that
were first present in lsp-server 0.3.2.
Without this change:
error[E0599]: no method named `initialize_start` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/./src/bin/main.rs:83:57
|
83 | let (initialize_id, initialize_params) = connection.initialize_start()?;
| ^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
error[E0599]: no method named `initialize_finish` found for struct `lsp_server::Connection` in the current scope
--> crates/rust-analyzer/./src/bin/main.rs:99:16
|
99 | connection.initialize_finish(initialize_id, initialize_result)?;
| ^^^^^^^^^^^^^^^^^ method not found in `lsp_server::Connection`
Eventually, we should support "just open random rust file" use case,
we don't really do this now, so let's avoid spending time on it until
we fix it properly.
As per matklad, we now pass the responsibility for finding the binary to the frontend.
Also, added caching for finding the binary path to reduce
the amount of filesystem interactions.
The line separator is moved below the function signature to split
regions between the docs. This is very similar to how IntelliJ
displays tooltips. Adding an additional separator between the module
name and function signature currently has rendering issues.
Fixes#4594
Alternative to #4615
4602: Add boolean literal semantic token type to package.json r=matklad a=lnicola
Closes#4583.
CC @GrayJack
4603: Add self keyword semantic token type r=matklad a=lnicola
Not sure if this is warranted a new token type or just a modifier.
---
CC #4583, @GrayJack
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
4571: KISS SourceChange r=matklad a=matklad
The idea behind requiring the label is a noble one, but we are not
really using it consistently anyway, and it should be easy to retrofit
later, should we need it.
bors r+
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
The idea behind requiring the label is a noble one, but we are not
really using it consistently anyway, and it should be easy to retrofit
later, should we need it.
4516: LSP: Two stage initialization r=kjeremy a=kjeremy
Fills in server information.
Derives CodeAction capabilities from the client. If code action literals
are unsupported we fall back to the "simple support" which just sends back
commands (this is already supported in our config). The difference being
that we did not adjust our server capabilities so that if the client was
checking for `CodeActionProvider: "true"` in the response that would have failed.
Part of #144Fixes#4130 (the specific case called out in that issue)
Co-authored-by: kjeremy <kjeremy@gmail.com>
This also changes our handiling of snippet edits on the client side.
`editor.insertSnippet` unfortunately forces indentation, which we
really don't want to have to deal with. So, let's just implement our
manual hacky way of dealing with a simple subset of snippets we
actually use in rust-analyzer
Fills in server information.
Derives CodeAction capabilities from the client. If code action literals
are unsupported we fall back to the "simple support" which just sends back
commands (this is already supported in our config). The difference being
that we did not adjust our server capabilities so that if the client was
checking for `CodeActionProvider: "true"` in the response that would have failed.
4448: Generate configuration for launch.json r=vsrs a=vsrs
This PR adds two new commands: `"rust-analyzer.debug"` and `"rust-analyzer.newDebugConfig"`. The former is a supplement to the existing `"rust-analyzer.run"` command and works the same way: asks for a runnable and starts new debug session. The latter allows adding a new configuration to **launch.json** (or to update an existing one).
If the new option `"rust-analyzer.debug.useLaunchJson"` is set to true then `"rust-analyzer.debug"` and Debug Lens will first look for existing debug configuration in **launch.json**. That is, it has become possible to specify startup arguments, env variables, etc.
`"rust-analyzer.debug.useLaunchJson"` is false by default, but it might be worth making true the default value. Personally I prefer true, but I'm not sure if it is good for all value.
----
I think that this PR also solves https://github.com/rust-analyzer/rust-analyzer/issues/3441.
Both methods to update launch.json mentioned in the issue do not work:
1. Menu. It is only possible to add a launch.json configuration template via a debug adapter. And anyway it's only a template and it is impossible to specify arguments from an extension.
2. DebugConfigurationProvider. The exact opposite situation: it is possible to specify all debug session settings, but it is impossible to export these settings to launch.json.
Separate `"rust-analyzer.newDebugConfig"` command looks better for me.
----
Fixes#4450Fixes#3441
Co-authored-by: vsrs <vit@conrlab.com>
Co-authored-by: vsrs <62505555+vsrs@users.noreply.github.com>
4397: Textmate cooperation r=matklad a=georgewfraser
This PR tweaks the fallback TextMate scopes to make them more consistent with the existing grammar and other languages, and edits the builtin TextMate grammar to align with semantic coloring. Before is on the left, after is on the right:
<img width="855" alt="Screen Shot 2020-05-10 at 1 45 51 PM" src="https://user-images.githubusercontent.com/1369240/81512320-a8be7e80-92d4-11ea-8940-2c03f6769015.png">
**Use keyword.other for regular keywords instead of keyword**. This is a really peculiar quirk of TextMate conventions, but virtually *all* TextMate grammars use `keyword.other` (colored blue in VSCode Dark+) for regular keywords and `keyword.control` (colored purple in VSCode Dark+) for control keywords. The TextMate scope `keyword` is colored like control keywords, not regular keywords. It may seem strange that the `keyword` scope is not the right fallback for the `keyword` semantic token, but TextMate has a long and weird history. Note how keywords change from purple back to blue (what they were before semantic coloring was added):
**(1) Use punctuation.section.embedded for format specifiers**. This aligns with how Typescript colors formatting directives:
<img width="238" alt="Screen Shot 2020-05-09 at 10 54 01 AM" src="https://user-images.githubusercontent.com/1369240/81481258-93b5f280-91e3-11ea-99c2-c6d258c5bcad.png">
**(2) Consistently use `entity.name.type.*` scopes for type names**. Avoid using `entity.name.*` which gets colored like a keyword.
**(3) Use Property instead of Member for fields**. Property and Member are very similar, but if you look at the TextMate fallback scopes, it's clear that Member is intended for function-like-things (methods?) and Property is intended for variable-like-things.
**(4) Color `for` as a regular keyword when it's part of `impl Trait for Struct`**.
**(5) Use `variable.other.constant` for constants instead of `entity.name.constant`**. In the latest VSCode insiders, variable.other.constant has a subtly different color that differentiates constants from ordinary variables. It looks close to the green of types but it's not the same---it's a new color recently added to take advantage of semantic coloring.
I also made some minor changes that make the TextMate scopes better match the semantic scopes. The effect of this for the user is you observe less of a change when semantic coloring "activates". You can see the changes I made relative to the built-in TextMate grammar here:
a91d15c80c..97428b6d52 (diff-6966c729b862f79f79bf7258eb3e0885)
Co-authored-by: George Fraser <george@fivetran.com>
4394: Simplify r=matklad a=Veetaha
4414: Highlighting improvements r=matklad a=matthewjasper
- `static mut`s are highlighted as `mutable`.
- The name of the macro declared by `macro_rules!` is now highlighted.
Co-authored-by: veetaha <veetaha2@gmail.com>
Co-authored-by: Matthew Jasper <mjjasper1@gmail.com>
This starts the transition to a new method of documenting the cfgs that are
enabled for a given crate in the json file. This is changing from a list
of atoms and a dict of key:value pairs, to a list of strings that is
equivalent to that returned by `rustc --print cfg ..`, and parsed in the
same manner by rust-analyzer.
This is the first of two changes, which adds the new field that contains
the list of strings. Next change will complete the transition and remove
the previous fields.
4166: Defining a default target to support cross-compilation targets r=matklad a=FuriouZz
Related to #4163
Co-authored-by: Christophe MASSOLIN <christophe.massolin@gmail.com>
4207: Add unwrap block assist #4156 r=matklad a=bnjjj
close issue #4156
4253: Remove `workspaceLoaded` setting r=matklad a=eminence
The `workspaceLoaded` notification setting was originally designed to
control the display of a popup message that said:
"workspace loaded, {} rust packages"
This popup was removed and replaced by a much sleeker message in the
VSCode status bar that provides a real-time status while loading:
rust-analyzer: {}/{} packages
This was done as part of #3587
The change in this PR simply renames this setting from `workspaceLoaded` to
`progress` to better describe what it actually controls. At the moment,
the only type of progress message that is controlled by this setting is
the initial load messages, but in theory other messages could also be
controlled by this setting.
Reviewer notes:
* If we didn't like the idea of causing minor breaking to user's config, we could keep the setting name as `workspaceLoaded`
* I think we can now close both #2719 and #3176 since the notification dialog in question no longer exists (actually I think you can close those issues even if you reject this PR 😄 )
Co-authored-by: Benjamin Coenen <5719034+bnjjj@users.noreply.github.com>
Co-authored-by: Andrew Chin <achin@eminence32.net>
The `workspaceLoaded` notification setting was originally designed to
control the display of a popup message that said:
"workspace loaded, {} rust packages"
This popup was removed and replaced by a much sleeker message in the
VSCode status bar that provides a real-time status while loading:
rust-analyzer: {}/{} packages
This was done as part of #3587
The new status-bar indicator is unobtrusive and shouldn't need to be
disabled. So this setting is removed.
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>
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>
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.
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>
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.
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>
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`.
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>
