Note: we meed to skip the trivia filter to make sure that
`covers!(call_info_bad_offset)` succeeds otherwise we exit call_info
too early.
Also the test doesn't pass: `FnCallNode::with_node` always detects
a MacroCall.
2562: Fix NavigationTarget ranges r=matklad a=edwin0cheng
Fix the issue described in https://github.com/rust-analyzer/rust-analyzer/pull/2544#issuecomment-565572553
This PR change the order for finding `full_range` of `focus_range` in following orders:
1. map both ranges to macro_call
2. map focus range to a token inside macro call, and full range to the whole of macro call
3. map both ranges to the whole of macro call
And fix the corresponding tests and make these tests easily to follow.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
When calling a function, argument-position impl Trait is transparent; same for
return-position impl Trait when inside the function. So in these cases, we need
to represent that type not by `Ty::Opaque`, but by a type variable that can be
unified with whatever flows into there.
2559: Add some granularity to syntax highlighting. r=matklad a=omerbenamram
Hi,
I wanted to start using `rust-analyzer` a bit more frequently - one of the main blockers for me so far was the highlighting.
I just discovered it's possible to override the default colors with `ralsp.<something>` setting without waiting for #2061!
However, the current implementation was lumping a bunch of different tokens into `type` and `literal`.
The golden standard IMO is what Clion is currently doing (and is my current daily driver for rust).
Clion allows users to control the coloring for specific literal kinds, and the default is to distinguish between them (numerics get a different color from strings, and special colors for bytestrings).
I've also splitted the builtin types, which are also allowed to be highlighted speratly.
My goal is to match the default experience I'm getting with clion.
The only blockers now I think is that `rust-analyzer` doesn't corrently infer types in some situations, so the highlighting information is incorrect in those cases.
This is what it looks like so far (with colors overriden to match clion's theme):
If there are any other changes you feel is necessary let me know.
I did leave the default colors to match the current behavior, since I'm not familiar with the colors for this theme, I added some random (different) colors in the test to check that it indeed was working.
Co-authored-by: Omer Ben-Amram <omerbenamram@gmail.com>
2552: fix goto definition when inbetween tokens r=matklad a=succcubbus
fixes both goto_definition and goto_type_definition.
before, when running goto between some non-trivia token and an
identifier, goto would be attempted for the non-trivia token.
but this does not make sense for e.g. L_PAREN or COLONCOLON only for
IDENTs.
this resulted in goto actions not working when running them on the first
character of some identifier e.g. for `module::<|>method()` or
`method(<|>parameter)`.
now only IDENTs will be searched for in goto actions, though i'm not sure
if this is correct or if goto should also work for some other token types.
Co-authored-by: succcubbus <16743652+succcubbus@users.noreply.github.com>
2550: Infer - and ! using std::ops::{Neg, Not} r=flodiebold a=kiljacken
Found some low hanging fruit while taking a cursory look at index inferring.
Co-authored-by: Emil Lauridsen <mine809@gmail.com>
fixes both goto_definition and goto_type_definition.
before, when running goto between some non-trivia token and an
identifier, goto would be attempted for the non-trivia token.
but this does not make sense for e.g. L_PAREN or COLONCOLON only for
IDENTs. now only IDENTs will be searched for in goto actions.
2544: Map first and last tokens in original_range r=matklad a=edwin0cheng
This PR try to fix the first part of the `original_range` : Try to map the first token and last token of a `SyntaxNode` , If success, return the union range of mapped tokens.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
Tuple in type annotation expands correctly;
Expansion will prefer the following delimiter when possible.
New regression tests added to verify the consistency between tuple expansion in type annotation and tuple expansion in rvalue.
2500: Fix format_args expansion & go to definition r=matklad a=flodiebold
The expansion of format_args wasn't yet correct enough to type-check. Also make macros in statement position expand to expressions for now, since it's not handled correctly in HIR lowering yet. This finally fixes go to definition within print macros, I think 🙂
2505: Remove more dead code r=matklad a=matklad
2506: Remove one more Ty r=matklad a=matklad
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
2501: Fix coercion from &Foo to an inference variable in a reference r=matklad a=flodiebold
We didn't try to unify within the reference, but we should.
2502: Delay legacy macro expansion r=matklad a=edwin0cheng
This PR make the following changes:
* Delay legacy macro expansion such that we concentrated all item collecting macro expansion in one place.
* Add `MacroDirective` to replace 3-tuples
* After this refactoring, no macro is expanded recursively, hence we can remove the `MacroStackMonitor` and we handle the expansion limit by the fix-point loop count.
2503: Code: check whether the LSP binary is in PATH r=matklad a=lnicola
I'm not really sure about the TS changes. I just made a couple of functions async and it seems to work.
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
2466: Handle partial resolve cases r=matklad a=edwin0cheng
Another try to fix#2443 :
We resolve all imports every time in `DefCollector::collect` loop even it is resolved previously.
This is because other unresolved imports and macros will bring in another `PerNs`, so we can only assume that it has been partially resolved.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
2489: Implement `format_args` r=flodiebold a=flodiebold
This fixes a huge amount of type mismatches (because every format call was a type mismatch so far); I also hoped to get go to def working within `format!` etc., and the test says it should, but in practice it still doesn't seem to...
Also remove the `len` parameter from `Name::new_inline_ascii`, which I'm assuming was only there because of `const fn` limitations?
cc @edwin0cheng
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
2484: DynMap r=matklad a=matklad
Implement a `DynMap` a semi-dynamic, semi-static map, which helps to thread heterogeneously typed info in a uniform way. Totally inspired by df3bee3038/compiler/frontend/src/org/jetbrains/kotlin/resolve/BindingContext.java.
@flodiebold wdyt? Seems like a potentially useful pattern for various source-map-like things.
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
SourceAnalyzer didn't work properly within expression macro expansions because
it didn't find the enclosing function. Fix this by going up the expansion chain
to find ancestors. This makes the test work, but apparently in real usage it's
still not working.
If we are expecting a `&Foo` and get a `&something`, when checking the
`something`, we are *expecting* a `Foo`, but we shouldn't try to unify whatever
we get with that expectation, because it could actually be a `&Foo`, and `&&Foo`
coerces to `&Foo`. So this fixes quite a few false type mismatches.
2479: Add expansion infrastructure for derive macros r=matklad a=flodiebold
I thought I'd experiment a bit with attribute macro/derive expansion, and here's what I've got so far. It has dummy implementations of the Copy / Clone derives, to show that the approach works; it doesn't add any attribute macro support, but I think that fits into the architecture.
Basically, during raw item collection, we look at the attributes and generate macro calls for them if necessary. Currently I only do this for derives, and just add the derive macro calls as separate calls next to the item. I think for derives, it's important that they don't obscure the actual item, since they can't actually change it (e.g. sending the item token tree through macro expansion unnecessarily might make completion within it more complicated).
Attribute macros would have to be recognized at that stage and replace the item (i.e., the raw item collector will just emit an attribute macro call, and not the item). I think when we implement this, we should try to recognize known inert attributes, so that we don't do macro expansion unnecessarily; anything that isn't known needs to be treated as a possible attribute macro call (since the raw item collector can't resolve the macro yet).
There's basically no name resolution for attribute macros implemented, I just hardcoded the built-in derives. In the future, the built-ins should work within the normal name resolution infrastructure; the problem there is that the builtin stubs in `std` use macros 2.0, which we don't support yet (and adding support is outside the scope of this).
One aspect that I don't really have a solution for, but I don't know how important it is, is removing the attribute itself from its input. I'm pretty sure rustc leaves out the attribute macro from the input, but to do that, we'd have to create a completely new syntax node. I guess we could do it when / after converting to a token tree.
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
The stand-alone `unify` requires that the type doesn't contain any type
variables. So we can't share the code here for now (without more refactoring)...
2465: Extract built-in trait implementations to separate module r=matklad a=flodiebold
This untangles the builtin logic from the Chalk translation.
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
2018: assists: add assist for custom implementation for derived trait r=matklad a=paulolieuthier
Please, tell me if something could be more idiomatic or efficient.
Fixes#1256.
Co-authored-by: Paulo Lieuthier <paulolieuthier@gmail.com>
2455: Add BuiltinShadowMode r=flodiebold a=edwin0cheng
This PR try to fix#1905 by introduce an `BuiltinShadowMode` in name resolving functions.
cc @flodiebold
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
2451: Use env_logger instead of flexi_logger r=matklad a=AlexanderEkdahl
This fixes https://github.com/rust-analyzer/rust-analyzer/issues/2335
- By default only `error` will be printed. From what I can tell this matches the current behaviour. Configured through `RUST_LOG`.
- I looked through the optional dependencies for `env_logger`and I have only enabled `human_time`. Without this feature no timestamp will be shown for log messages.
- `RA_LOG_DIR` feature is removed
This PR adds 2 new dependencies(`env_logger` and `human_time`) and removes 6 dependencies.
Co-authored-by: Alexander Ekdahl <alexander@ekdahl.io>
2418: Hide MacroCallLoc outside hir_expand r=matklad a=edwin0cheng
This PR refactor `MacroCallLoc` such that it be hided to become implementation details of hir_expand.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
Although structs and unions have the same syntax and differ only in
the keyword, re-using the single syntax node for both of them leads to
confusion in practice, and propagates further down the hir in an
upleasent way.
Moreover, static and consts also share syntax, but we use different
nodes for them.
2388: Show missing struct fields in the error message r=matklad a=Frizi
This provides the most interesting information about the "missing structure fields" error directly to the user.
Co-authored-by: Frizi <frizi09@gmail.com>
2362: Expand compile_error! r=edwin0cheng a=kjeremy
Does not validate that the input is a string literal. I thought that I could `match_ast!` against the `macro_args` but that did not work. Even if it had I am not sure which error would be appropriate.
Co-authored-by: Jeremy Kolb <kjeremy@gmail.com>
2392: Fix panic during the expansion of `column!` r=edwin0cheng a=marcogroppo
Fixes#2379. Well, this isn't the "proper" fix but it doesn't hurt, IMHO.
The problem is that `to_col_number`, called by `column_expand`, receives a position number that isn't included in the text range of the file. My (very limited) understanding is that the text is the one of the original file, while `pos` is relative to something else, probably the text of the macro. Notice that in this case the `column!` expansion seems to be triggered by `assert_eq!`, so we're in the middle of another expansion. This PR simply avoids the panic by checking the length of the text.
r? @edwin0cheng
Co-authored-by: Marco Groppo <marco.groppo@gmail.com>
2396: Switch to variant-granularity field type inference r=flodiebold a=matklad
r? @flodiebold
Previously, we had a `ty` query for each field. This PR switcthes to a query per struct, which returns an `ArenaMap` with `Ty`s.
I don't know which approach is better. What is bugging me about the original approach is that, if we do all queries on the "leaf" defs, in practice we get a ton of queries which repeatedly reach into the parent definition to compute module, resolver, etc. This *seems* wasteful (but I don't think this is really what causes any perf problems for us).
At the same time, I've been looking at Kotlin, and they seem to use the general pattern of analyzing the *parent* definition, and storing info about children into a `BindingContext`.
I don't really which way is preferable. I think I want to try this approach, where query granularity generally mirrors the data granularity. The primary motivation for me here is probably just hope that we can avoid adding a ton of helpers to a `StructField`, and maybe in general avoid the need to switch to a global `StructField`, using `LocalStructFieldId` most of the time internally.
For external API (ie, for `ra_ide_api`), I think we should continue with fine-grained `StructField::ty` approach, which internally fetches the table for the whole struct and indexes into it.
In terms of actual memory savings, the results are as follows:
```
This PR:
142kb FieldTypesQuery (deps)
38kb FieldTypesQuery
Status Quo:
208kb TypeForFieldQuery (deps)
18kb TypeForFieldQuery
```
Note how the table itself occupies more than twice as much space! I don't have an explanation for this: a plausible hypothesis is that single-field structs are very common and for them the table is a pessimisation.
THere's noticiable wallclock time difference.
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
2381: Add proc-macro crate type handling r=JasperDeSutter a=JasperDeSutter
Resolves the libproc_macro crate in crates that are the proc-macro type.
This doesn't seem the ideal implementation though, since the compiler still requires you to write `extern crate proc_macro;` (even in 2018 edition).
Co-authored-by: JasperDeSutter <jasper.desutter@gmail.com>
2383: Add alloc to the crate graph r=matklad a=marcogroppo
`alloc` has been added to the crate graph.
Completions work, but they are available even when the user has **not** declared an `extern crate alloc`. Is this the correct approach?
Fixes#2376.
Co-authored-by: Marco Groppo <marco.groppo@gmail.com>
2365: Make expand-macro more flexible r=matklad a=edwin0cheng
Due to lack of implementation or other types of errors, some macros do not expand correctly in the current situation. The PR attempts to make `expand-macro` more flexible in error situations by ignoring internal failed macro expansion.
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
2343: implement assist invert_if r=matklad a=bravomikekilo
fix [issue 2219 invert if condition](https://github.com/rust-analyzer/rust-analyzer/issues/2219)
I put the assist cursor range to `if` of the if expression, because both condition and body will be replaced. Is there any way to replace them without cover the cursor position?
@matklad
Co-authored-by: bravomikekilo <bmk1221@126.com>
