9453: Add first-class limits. r=matklad,lnicola a=rbartlensky
Partially fixes#9286.
This introduces a new `Limits` structure which is passed as an input
to `SourceDatabase`. This makes limits accessible almost everywhere in
the code, since most places have a database in scope.
One downside of this approach is that whenever you query limits, you
essentially do an `Arc::clone` which is less than ideal.
Let me know if I missed anything, or would like me to take a different approach!
Co-authored-by: Robert Bartlensky <bartlensky.robert@gmail.com>
9652: Don't concat path in replace_qualified assist when they start with a keyword r=Veykril a=Veykril
Also keep the path if we can't find a path to the item instead of becoming non applicable.
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9637: Overhaul doc_links testing infra r=Veykril a=Veykril
and fix several issues with current implementation.
Fixes#9617
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9646: add 'use' prefix for any auto-import r=Veykril a=mahdi-frms
Fixes#9643
looks like there was an if condition that had to be removed.
Co-authored-by: mahdi-frms <mahdif1380@outlook.com>
Our project model code is rather complicated -- the logic for lowering
from `cargo metadata` to `CrateGraph` is fiddly and special-case. So
far, we survived without testing this at all, but this increasingly
seems like a poor option.
So this PR introduces a simple tests just to detect the most obvious
failures. The idea here is that, although we rely on external processes
(cargo & rustc), we are actually using their stable interfaces, so we
might just mock out the outputs.
Long term, I would like to try to virtualize IO here, so as to do such
mocking in a more principled way, but lets start simple.
Should we forgo the mocking and just call `cargo metadata` directly
perhaps? Touch question -- I personally feel that fast, in-process tests
are more important in this case than any extra assurance we get from
running the real thing.
Super-long term, we would probably want to extend our heavy tests to
cover more use-cases, but we should figure a way to do that without
slowing the tests down for everyone.
Perhaps we need two-tiered bors system, where we pull from `master` into
`release` branch only when an additional set of tests passes?
I saw reference to globs in #7755, but it doesn't look like they're
actually supported, and I had to dig through the source to discover
that the folders are relative to the workspace root. Further digging
was required to get VS Code from hanging for long periods trying to
watch giant Bazel folders that had already been excluded from Rust
Analyzer. Hopefully this tweak will save others the confusion :-)
9614: Parse input expressions for dbg! invocations in remove_dbg r=Veykril a=Veykril
Instead of inspecting the input tokentree manually, parse the input as `,` delimited expressions instead and act on that. This simplifies the assist quite a bit.
Fixes#8455
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9600: fix: Single-line and nested blocks in the `unwrap_block` assist r=Veykril a=patrick-gu
Fixes#8411
Rework the system for stripping whitespace and braces in the unwrap_block assist to allow correct unwrapping of blocks such as:
```rust
{ $0 0 }
```
into
```rust
0
```
and nested blocks, such as:
```rust
$0{
{
3
}
}
```
into
```rust
{
3
}
```
This is done by creating the `update_expr_string_with_pat` function (along with `update_expr_string` and `update_expr_string_without_newline`), which strips whitespace and braces in a way that ensures that only whitespace and a maximum of one brace are removed from the start and end of the expression string.
I have also created several tests to ensure that this functionality works correctly.
Co-authored-by: patrick-gu <55641350+patrick-gu@users.noreply.github.com>
9550: Proc macro multi abi proof of concept r=matklad a=alexjg
#8925 was irritating me so I thought I would have a bash at fixing it. What I've done here is copy the `crates/proc_macro_srv/src/proc_macro` code (which is copied from `<RUST>/library/proc_macro`) to `crates/proc_macro_srv/src/proc_macro_nightly` and the modified the nightly version to include the changes from https://github.com/rust-analyzer/rust-analyzer/pull/9047 and aeb7b183a2
This gives us the code to support both stable and nightly ABIs. Then we use the `proc_macro_api::version::read_dylib_info` to determine which version of the ABI to load when creating a `ProcMacroLibraryLibLoading` (which is now an enum).
This seems to work for me. The code could be cleaned up but I wanted to see if the approach makes sense before I spend more time on it.
I've split the change into two commits, the first is just copying and modifying the `proc_macro` crate, the second contains most of the interesting work around figuring out which ABI to use.
Co-authored-by: Alex Good <alex@memoryandthought.me>
Co-authored-by: alexjg <alex@memoryandthought.me>
In rust-analyzer, we avoid defualt impls for types which don't have
sensible, "empty" defaults. In particular, we avoid using invalid
indices for defaults and similar hacks.
Rather than a "Stable" and "Nightly" ABI we instead name ABIs based on
the version of the rust compiler in which they were introduced. We place
these ABIs in a new module - `proc_macro_srv::abis` - where we also add
some mchinery to abstract over ABIs. This should make it easy to add new
ABIs at a later date as the rust compiler evolves.
9535: internal: remove proc macro management thread r=jonas-schievink a=jonas-schievink
Communication with the proc macro server process has always happened one request at a time, so the additional thread isn't really needed (it just forwarded each request, and sent back the response). This removes some indirection that was a bit hard to understand (a channel was allocated and sent over another channel to return the response).
Hope I'm not missing anything here
Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
9569: internal: Explicitly check for reference locals or fields in Name classification r=Veykril a=Veykril
Closes#7524
Inlines all the calls to reference related name classification as well as emits both goto definition targets for field shorthands.
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
This treats the consts generated by older synstructure versions like
unnamed consts. We should remove this at some point (at least after
Chalk has switched).
9567: remove unneded special case r=matklad a=matklad
bors r+
🤖
9568: feat: add 'for' postfix completion r=lnicola a=mahdi-frms
adds #9561
used ```ele``` as identifier for each element in the iteration
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
Co-authored-by: mahdi-frms <mahdif1380@outlook.com>
9553: minor: Disambiguate replace with if let assist labels r=Veykril a=Veykril
Turns out we have two assists for replacing something with `if let` constructs, so having the cursor on a `let` keyword inside a match gave you two identical assist labels which is rather confusing.
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9555: feat: Enable `auto_import` on ident patterns r=Veykril a=Veykril
Helpful for when you want to import a type in a pattern right before destructuring it.
9556: Bump deps r=lnicola a=lnicola
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
9548: add: Adding self keyword completion in complete_fn_param r=lnicola a=feniljain
Solves #9522
I haven't added Arc<self> for now as there were some conflicting opinions on it
Co-authored-by: vi_mi <fenil.jain2018@vitstudent.ac.in>
... instead of using `AliasTy`. Chalk turns the alias type into the
placeholder during unification anyway, which confuses our method
resolution logic.
Fixes#9530.
9519: Explicitly name all spawned threads r=weirdsmiley a=weirdsmiley
Fixes: [#9385](https://github.com/rust-analyzer/rust-analyzer/issues/9385)
The thread name is shown in debugger as well as panic messages and this
patch makes it easier to follow a thread instead of looking through
full backtrace, by naming all spawned threads according to their
functioning.
Co-authored-by: Manas <manas18244@iiitd.ac.in>
The thread name is shown in debugger as well as panic messages and this
patch makes it easier to follow a thread instead of looking through
full backtrace, by naming all spawned threads according to their
functioning.
9500: internal: Only inline closure, literal and local arguments when used once r=Veykril a=Veykril
See https://github.com/rust-analyzer/rust-analyzer/pull/9474#discussion_r663881507 for reasoning.
This still inlines single use closures and literals as naming these is usually not as useful. Prime examples being the Option/Result consuming functions like `map_or` etc.
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9497: Wrap inlined closures in parens when inlined in an expression in `inline_call` r=Veykril a=Veykril
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
Note that, while we don't currently have a fuzzy-matching score, it
makes sense to special-case postfix templates -- it's very annoying when
`.not()` gets sorted before `.not`. We might want to move this infra to
fuzzy matching, once we have that!
Before this PR, SourceChange used a bool and CompletionItem used an enum
to signify if edit is a snippet. It makes sense to use the same pattern
in both cases. `bool` feels simpler, as there's only one consumer of
this API, and all producers are encapsulated anyway (we check the
capability at the production site).
One source completion can produce up to two lsp completions.
Additionally, `preselct` and `sort_text` are global properties of the
whole set of completions, so the right granularity here is to convert
many completions.
As a side-benefit, we no loger allocate intermediate vec.
Moving tests to `rust-analyzer` crate allows removing walkdir dependency
from `xtask`. It does seem more reasonable to keep tidy tests outside of
the "build system" and closer to other integration tests.
* Keep codegen adjacent to the relevant crates.
* Remove codgen deps from xtask, speeding-up from-source installation.
This regresses the release process a bit, as it now needs to run the
tests (and, by extension, compile the code).
9455: feat: Handle not let if expressions in replace_if_let_with_match r=Veykril a=Veykril
Transforms bare `if cond {}` into `_ if cond` guard patterns in the match as long as at least one `if let` is in the if chain, otherwise the assist wont be applicable.
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9454: feat: Empower `replace_if_let_with_match` r=Veykril a=Veykril
Now instead of only working on `if let ... {} else {}` if expressions it now works on all of them where the condition expression is the same text-wise.
This includes if let expressions without an else block, in which case a simple `_ => ()` will be generated in the resulting match but also in more complex cases where multiple `if let` expressions are chained.
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9452: feat: Add "View Crate Graph (Full)" r=jonas-schievink a=jonas-schievink
Works like "View Crate Graph", but also includes crates.io and sysroot dependencies. The resulting graph might be enormous.
Closes https://github.com/rust-analyzer/rust-analyzer/issues/8867
bors r+
Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
9450: internal: Add ModuleOrItem guess to import granularity guessing r=Veykril a=Veykril
I think this should be the last fix needed for this(🤞)
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9436: minor: Add test for macro expanded test module in runnables r=Veykril a=Veykril
Expected this to fail as thats behaving incorrectly on current nightly but I think I fixed this accidentally with https://github.com/rust-analyzer/rust-analyzer/pull/9435
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9269: Recreate status page r=lnicola a=Milo123459
I'm working on redesigning the status page.
Co-authored-by: Milo <50248166+Milo123459@users.noreply.github.com>
9423: fix: Resolve attribute paths in attribute highlighting r=Veykril a=Veykril
Attributes have a new highlighting format now, whereas the `#[` `]` tokens are now tagged with `attribute.attribute` like before, but all other idents inside token trees are now `generic.attribute`. If a path in an attribute can't be resolved it will instead get the `builtinAttribute.attribute` tags now as highlighting doesn't know about builtins like `allow` yet, so we don't want to emit unresolved references.
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9418: internal: Include `self` in usage search for modules in their definition source r=Veykril a=Veykril
bors r+
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
Macro that deep clone the tokens but otherwise preserves source
locations and hygiene info is an interesting case for IDE support. Lets
have this, although we don't actively use it at the moment.
9380: feat: Implement goto_declaration support r=matklad a=Veykril
This is just a simple implementation that falls back to `goto_definition` for everything but modules where it goes to the actual module declaration if possible.
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9353: Include extra targets when the pkg_root is not the same as the target root. r=matklad a=rezural
Fixes#7715
For example, if a sub-crate includes sets the path='../somewhere-else/lib.rs', the files will not be in pkg_root , but in the target root's parent.
It may actually be in root.parent().parent(), I'm not sure about that.
At the moment it is just a fix, are there any relevant tests that this could go in? I've got about 1 brain cell left... but im happy to add tests where appropriate.
Co-authored-by: rezural <rezural@protonmail.com>
Definition::visibility was implemented in a rather roundabout way -- by
asking the parent module about the effective visibility.
This is problematic for a couple of reasons:
* first, it doesn't work for local items
* second, asking module about visibility of a child is a linear
operation (that's a problem in itself, tracked in #9378)
Instead, lets ask the declared visibility directly, we have all the code
for it, and need only to actually us it.
9348: output to log file if RA_LOG_FILE is defined in environment r=rezural a=rezural
This adds a check for RA_LOG_FILE, and logs to that if defined. It currently overrides flags.log_file. If this is undesirable, I will add a check.
Co-authored-by: rezural <rezural@protonmail.com>
This story begins in #8384, where we added a smart test for our syntax
highting, which run the algorithm on synthetic files of varying length
in order to guesstimate if the complexity is O(N^2) or O(N)-ish.
The test turned out to be pretty effective, and flagged #9031 as a
change that makes syntax highlighting accidentally quadratic. There was
much rejoicing, for the time being.
Then, lnicola asked an ominous question[1]: "Are we sure that the time
is linear right now?"
Of course it turned out that our sophisticated non-linearity detector
*was* broken, and that our syntax highlighting *was* quadratic.
Investigating that, many brave hearts dug deeper and deeper into the
guts of rust-analyzer, only to get lost in a maze of traits delegating
to traits delegating to macros.
Eventually, matklad managed to peel off all layers of abstraction one by
one, until almost nothing was left. In fact, the issue was discovered in
the very foundation of the rust-analyzer -- in the syntax trees.
Worse, it was not a new problem, but rather a well-know, well-understood
and event (almost) well-fixed (!) performance bug.
The problem lies within `SyntaxNodePtr` type -- a light-weight "address"
of a node in a syntax tree [3]. Such pointers are used by rust-analyzer all
other the place to record relationships between IR nodes and the
original syntax.
Internally, the pointer to a syntax node is represented by node's range.
To "dereference" the pointer, you traverse the syntax tree from the
root, looking for the node with the right range. The inner loop of this
search is finding a node's child whose range contains the specified
range. This inner loop was implemented by naive linear search over all
the children. For wide trees, dereferencing a single `SyntaxNodePtr` was
linear. The problem with wide trees though is that they contain a lot of
nodes! And dereferencing pointers to all the nodes is quadratic in the
size of the file!
The solution to this problem is to speed up the children search --
rather than doing a linear lookup, we can use binary search to locate
the child with the desired interval.
Doing this optimization was one of the motivations (or rather, side
effects) of #6857. That's why `rowan` grew the useful
`child_or_token_at_range` method which does exactly this binary search.
But looks like we've never actually switch to this method? Oups.
Lesson learned: do not leave broken windows in the fundamental infra.
Otherwise, you'll have to repeatedly re-investigate the issue, by
digging from the top of the Everest down to the foundation!
[1]: https://rust-lang.zulipchat.com/#narrow/stream/185405-t-compiler.2Frust-analyzer/topic/.60syntax_highlighting_not_quadratic.60.20failure/near/240811501
[2]: https://rust-lang.zulipchat.com/#narrow/stream/185405-t-compiler.2Frust-analyzer/topic/Syntax.20highlighting.20is.20quadratic
[3]: https://rust-lang.zulipchat.com/#narrow/stream/185405-t-compiler.2Frust-analyzer/topic/Syntax.20highlighting.20is.20quadratic/near/243412392
The completion of cfg will look at the enabled cfg keys when
performing completion.
It will also look crate features when completing a feature cfg
option. A fixed list of known values for some cfg options are
provided.
For unknown keys it will look at the enabled values for that cfg key,
which means that completion will only show enabled options for those.
9264: feat: Make documentation on hover configurable r=Veykril a=Veykril
This also implements deprecation support for config options as this renames `hoverActions_linksInHover` to `hover_linksInHover`.
Fixes#9232
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9227: Add a config setting to disable the 'test' cfg in specified crates r=matklad a=lf-
If you are opening libcore from rust-lang/rust as opposed to e.g.
goto definition from some other crate which would use the sysroot
instance of libcore, a `#![cfg(not(test))]` would previously have made
all the code excluded from the module tree, breaking the editor
experience.
Core does not need to ever be edited with `#[cfg(test)]` enabled,
as the tests are in another crate.
This PR puts in a slight hack that checks for the crate name "core" and
turns off `#[cfg(test)]` for that crate.
Fixes#9203Fixes#9226
Co-authored-by: Jade <software@lfcode.ca>
If you are opening libcore from rust-lang/rust as opposed to e.g.
goto definition from some other crate which would use the sysroot
instance of libcore, a `#![cfg(not(test))]` would previously have made
all the code excluded from the module tree, breaking the editor
experience.
This puts in a slight hack that checks for the crate name "core" and
turns off `#[cfg(test)]`.
9334: feat: Allow to disable import insertion on single path glob imports r=Veykril a=Veykril
On by default as I feel like this is something the majority would prefer.
Closes#8490
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9321: Inline generics in const and function trait completions r=Veykril a=RDambrosio016
This PR does a couple of things:
- moves path_transform from ide_assists to ide_db to be shared by both assists and completions
- when completing a const or a function for a trait, it will "inline" any generics in those associated items instead
of leaving the generic's name. For example:
```rust
trait Foo<T> {
const BAR: T;
fn foo() -> T;
}
struct Bar;
impl Foo<u32> for Bar {
// autocompletes to this
fn foo() -> u32;
// and not this (old)
fn foo() -> T;
// also works for associated consts and where clauses
const BAR: u32 = /* */
}
```
Currently this does not work for const generics, because `PathTransform` does not seem to account for them. If this should work on const generics too, `PathTransform` will need to be changed. However, it is uncommon to implement a trait only for a single const value, so this isnt a huge concern.
Co-authored-by: rdambrosio <rdambrosio016@gmail.com>
9258: minor: Give `ImportPrefix` variants better config names r=matklad a=Veykril
I feel like `crate` and `self` work better than `by_crate` and `by_self`. The only reason for the current names were that `Self` doesn't work for the variant name on the rust side so I forgot about setting proper config names on serde layer.
Co-authored-by: Lukas Wirth <lukastw97@gmail.com>
9294: internal: introduce minicore -- a subset of libcore for testing r=matklad a=matklad
Clearly, we need one more fixed point iteration loop!
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
