741 Commits

Author SHA1 Message Date
Noah Lev
4943688e9d Fix from rebase
I changed the test functions to be `pub` rather than called from a
`main` function too, for easier future modification of tests.
2022-03-27 13:54:34 -07:00
bors
ab0c2e18dc Auto merge of #94495 - estebank:missing-closing-gt, r=jackh726
Provide suggestion for missing `>` in a type parameter list

When encountering an inproperly terminated type parameter list, provide
a suggestion to close it after the last non-constraint type parameter
that was successfully parsed.

Fix #94058.
2022-03-27 18:55:58 +00:00
Esteban Kuber
157c67b7a8 Handle , to ; substitution in arg params 2022-03-27 06:05:18 +00:00
Esteban Kuber
6874bd27f5 Provide suggestion for missing > in a type parameter list
When encountering an inproperly terminated type parameter list, provide
a suggestion to close it after the last non-constraint type parameter
that was successfully parsed.

Fix #94058.
2022-03-27 02:50:04 +00:00
Aaron Hill
8035796b9a
Stablize const_extern_fn for "Rust" and "C"
All other ABIs are left unstable for now.

cc #64926
2022-03-26 18:23:54 -04:00
bors
c74925438c Auto merge of #95149 - cjgillot:once-diag, r=estebank
Remove `Session::one_time_diagnostic`

This is untracked mutable state, which modified the behaviour of queries.
It was used for 2 things: some full-blown errors, but mostly for lint declaration notes ("the lint level is defined here" notes).

It is replaced by the diagnostic deduplication infra which already exists in the diagnostic emitter.
A new diagnostic level `OnceNote` is introduced specifically for lint notes, to deduplicate subdiagnostics.

As a drive-by, diagnostic emission takes a `&mut` to allow dropping the `SubDiagnostic`s.
2022-03-26 00:54:54 +00:00
Michael Howell
2a7837262f diagnostics: correct generic bounds with doubled colon
Fixes #95208
2022-03-25 13:57:05 -07:00
Michael Goulet
91ac9cf595 suggest wrapping single-expr blocks in square brackets 2022-03-24 21:40:20 -07:00
Noah Lev
86220d6e51 Fix rustfix panic on test
`run-rustfix` applies all suggestions regardless of their Applicability.
There's a flag, `rustfix-only-machine-applicable`, that does what it
says, but then the produced `.fixed` file would have invalid code from
the suggestions that weren't applied. So, I moved the cases of postfix
increment, in which case multiple suggestions are given, to the
`-notfixed` test, which does not run rustfix.

I also changed the Applicability to Unspecified since MaybeIncorrect
requires that the code be valid, even if it's incorrect.
2022-03-23 22:31:57 -07:00
Noah Lev
725cde42d5 Use multipart_suggestions
This records that the suggestions are mutually-exclusive (i.e., only one
should be applied).
2022-03-23 22:31:57 -07:00
Noah Lev
ef74796178 Change temporary variable name if it would conflict 2022-03-23 22:31:57 -07:00
Noah Lev
95960b7d54 Make standalone an enum 2022-03-23 22:31:57 -07:00
Noah Lev
4212835d99 Add heuristic to avoid treating x + +2 as increment 2022-03-23 22:31:57 -07:00
Noah Lev
29a5c363c7 Improve function names 2022-03-23 22:31:57 -07:00
Noah Lev
073010d425 Improve handling of tmp variable name conflicts 2022-03-23 22:31:57 -07:00
Noah Lev
62b8ea67b7 Emit both subexp and standalone sugg for postfix
This solves the TODO.
2022-03-23 22:31:57 -07:00
Noah Lev
7287f929b9 Emit structured suggestions for field accesses too 2022-03-23 22:31:57 -07:00
Noah Lev
67a9adbb54 Refactor, handle fields better, add field tests 2022-03-23 22:31:57 -07:00
Noah Lev
80e57e223e Reduce rightward drift 2022-03-23 22:31:57 -07:00
Noah Lev
d915606d50 Remove error recovery todos 2022-03-23 22:31:57 -07:00
Noah Lev
c9cc43aa66 Move increment checks to improve errors 2022-03-23 22:31:57 -07:00
Camelid
5d9cd4b851 Suggest i += 1 when we see i++ or ++i 2022-03-23 22:31:57 -07:00
Dylan DPC
0254928213
Rollup merge of #95212 - TaKO8Ki:replace-this-clone-with-this-create-snapshot-for-diagnostic, r=Dylan-DPC
Replace `this.clone()` with `this.create_snapshot_for_diagnostic()`

Use [`create_snapshot_for_diagnostic`](cd11905716/compiler/rustc_parse/src/parser/diagnostics.rs (L214-L223)) I implemented in https://github.com/rust-lang/rust/pull/94731 instead of `this.clone()` to avoid duplicate errors about unclosed delims being emitted when the `Parser` is dropped.
2022-03-23 03:05:32 +01:00
bors
5f37001055 Auto merge of #95215 - Dylan-DPC:rollup-l9f9t7l, r=Dylan-DPC
Rollup of 5 pull requests

Successful merges:

 - #95188 ([`macro-metavar-expr`] Fix generated tokens hygiene)
 - #95196 (rename LocalState::Uninitialized to Unallocated)
 - #95197 (Suggest constraining param for unary ops when missing trait impl)
 - #95200 (Cancel a not emitted error after parsing const generic args)
 - #95207 (update Termination trait docs)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup
2022-03-22 19:06:04 +00:00
Takayuki Maeda
b86a1d1c07 replace this.clone() with this.create_snapshot_for_diagnostic() 2022-03-23 01:26:34 +09:00
bors
64137f0b15 Auto merge of #94693 - nnethercote:parser-inlining, r=petrochenkov
Inline some parser functions

Some crates that do a lot of complex declarative macro expansion spend a lot of time parsing (and reparsing) tokens. These commits inline some functions for some minor speed wins.

r? `@ghost`
2022-03-22 16:24:50 +00:00
Takayuki Maeda
62ded071d5 cancel a not emitted error after parsing const generic args 2022-03-22 16:06:56 +09:00
Nicholas Nethercote
f8f1d3f00b Split TokenCursor::{next,next_desugared} into inlined and non-inlined halves. 2022-03-22 11:05:54 +11:00
Nicholas Nethercote
4e700a023c Split Parser::bump_with into inlined and non-inlined halves.
The call site within `Parser::bump` is hot.

Also add an inline annotation to `Parser::next_tok`. It was already
being inlined by the compiler; this just makes sure that continues.
2022-03-22 11:01:53 +11:00
bors
051d1176b7 Auto merge of #95104 - compiler-errors:remove-ascription, r=davidtwco
suggest removing type ascription in bad parsing position

Not sure how to test this with the non-nightly suggestion. Didn't add a new UI test because it already manifests in an existing UI test.

Fixes #95014
2022-03-21 02:29:55 +00:00
Matthias Krüger
63da51db93
Rollup merge of #95135 - TaKO8Ki:fix-not-emitted-unmatched-angle-bracket-error, r=Dylan-DPC
Fix a not emitted unmatched angle bracket error

closes #95128
2022-03-20 20:42:45 +01:00
Camille GILLOT
056951d628 Take &mut Diagnostic in emit_diagnostic.
Taking a Diagnostic by move would break the usual pattern
`diag.label(..).emit()`.
2022-03-20 20:36:08 +01:00
Takayuki Maeda
5c8aaedef6 fix a not emitted unmatched angle bracket error 2022-03-20 22:39:15 +09:00
Michael Goulet
3516a16eeb suggest removing type ascription in bad position 2022-03-18 21:22:26 -07:00
Takayuki Maeda
201a86046c use self.create_snapshot_for_diagnostic instead of self.clone() 2022-03-18 16:56:43 +09:00
Dylan DPC
270a41c33e
Rollup merge of #94960 - codehorseman:master, r=oli-obk
Fix many spelling mistakes

Signed-off-by: codehorseman <cricis@yeah.net>
2022-03-17 22:55:05 +01:00
Dylan DPC
5eb3433ed5
Rollup merge of #94731 - TaKO8Ki:const-generic-expr-recovery, r=davidtwco,oli-obk
Suggest adding `{ .. }` around a const function call with arguments

closes #91020
2022-03-17 22:55:04 +01:00
Dylan DPC
7e1415ea80
Rollup merge of #93222 - mark-i-m:errorreported, r=oli-obk
Make ErrorReported impossible to construct outside `rustc_errors`

There are a few places were we have to construct it, though, and a few
places that are more invasive to change. To do this, we create a
constructor with a long obvious name.

cc #69426 `@varkor` `@eddyb` `@estebank`

I actually didn't see that I was assigned to this issue until now...
2022-03-17 22:55:00 +01:00
Dylan DPC
c8133f6f5e
Rollup merge of #91133 - terrarier2111:unsafe-diagnostic, r=jackh726
Improve `unsafe` diagnostic

This fixes: https://github.com/rust-lang/rust/issues/90880
I didn't use the exact proposed messages though.
2022-03-17 22:54:59 +01:00
mark
bb8d4307eb rustc_error: make ErrorReported impossible to construct
There are a few places were we have to construct it, though, and a few
places that are more invasive to change. To do this, we create a
constructor with a long obvious name.
2022-03-16 10:35:24 -05:00
codehorseman
01dbfb3eb2 resolve the conflict in compiler/rustc_session/src/parse.rs
Signed-off-by: codehorseman <cricis@yeah.net>
2022-03-16 20:12:30 +08:00
David Tolnay
f427698c03
Parse inner attributes on inline const block 2022-03-15 17:56:59 -07:00
Takayuki Maeda
896b113ec3 use format_args_capture in some parts of rustc_parse 2022-03-15 19:17:29 +09:00
bors
95561b336c Auto merge of #94584 - pnkfelix:inject-use-suggestion-sites, r=ekuber
More robust fallback for `use` suggestion

Our old way to suggest where to add `use`s would first look for pre-existing `use`s in the relevant crate/module, and if there are *no* uses, it would fallback on trying to use another item as the basis for the suggestion.

But this was fragile, as illustrated in issue #87613

This PR instead identifies span of the first token after any inner attributes, and uses *that* as the fallback for the `use` suggestion.

Fix #87613
2022-03-15 03:56:33 +00:00
bors
f103b2969b Auto merge of #94865 - notriddle:notriddle/single-colon-path-not-const-generics, r=cjgillot
diagnostics: single colon within `<>` probably, not type ascription

Fixes #94812
2022-03-12 19:10:33 +00:00
Michael Howell
26e299a4a7 Use consistent prose for ::
Co-authored-by: Camille Gillot <gillot.camille@gmail.com>
2022-03-12 08:20:36 -07:00
Michael Howell
7e323370b3 diagnostics: single colon within <> probably, not type ascription
Fixes #94812
2022-03-11 15:35:18 -07:00
Dylan DPC
298c9a0e14
Rollup merge of #94839 - TaKO8Ki:suggest-using-double-colon-for-struct-field-type, r=cjgillot
Suggest using double colon when a struct field type include single colon

#92685
2022-03-11 20:29:47 +01:00
Takayuki Maeda
813f00dd4f fix a suggestion message 2022-03-11 21:26:06 +09:00
Takayuki Maeda
bdc3177868 suggest using double colon when using single colon in struct field type path 2022-03-11 16:15:57 +09:00