Handle multiple error fix suggestions carefuly
The existing code seems to assume that substitutions spans are disjoint,
which is not always the case.
In the example:
pub trait AAAA {}
pub trait B {}
pub trait C {}
pub type T<P: AAAA + B + C> = P;
, we get three substituions starting from ':' and ending respectively at
the end of each trait token.
With the former offset calculation, this would cause `underline_start` to
eventually become negative before being converted to `usize`...
The new version may report erroneous results for non perfectly overlapping
substitutions but I don't know if such examples exist. Alternatively, we
could detect these cases and trim out overlapping substitutions.
Fixes#67690
Silence `TooGeneric` error
This error may be produced during intermediate failed attempts at evaluation of a generic const, which may nevertheless succeed later.
Fixes https://github.com/rust-lang/rust/issues/66962.
r? @eddyb
Fix ICE involving calling `Instance.ty` during const evaluation
Fixes#67639
`Instance.ty` assumes that we are in a fully monomorphic context (e.g.
codegen), and can therefore use an empty `ParamEnv` when performing
normalization. Howver, the MIR constant evaluator code ends up calling
`Instance.ty` as a result of us attemptign to 'speculatively'
const-evaluate generic functions during const propagation.
As a result,
we may end up with projections involving type parameters
(e.g. <T as MyTrait>::Bar>) in the type we are trying to normalize.
Normalization expects us to have proper predicates in the `ParamEnv` for
such projections, and will ICE if we don't.
This commit adds a new method `Instance.ty_env`, which takes a
`ParamEnv` for use during normalization. The MIR const-evaluator code is
changed to use this method, passing in the proper `ParamEnv` for the
context at hand.
buffer lexer errors in rustdoc syntax checking
The code isn't ideal (I really would like to display the errors inline), but this at least gets us to where we were before #63017.
Fixes#67639
`Instance.ty` assumes that we are in a fully monomorphic context (e.g.
codegen), and can therefore use an empty `ParamEnv` when performing
normalization. Howver, the MIR constant evaluator code ends up calling
`Instance.ty` as a result of us attemptign to 'speculatively'
const-evaluate generic functions during const propagation.
As a result,
we may end up with projections involving type parameters
(e.g. <T as MyTrait>::Bar>) in the type we are trying to normalize.
Normalization expects us to have proper predicates in the `ParamEnv` for
such projections, and will ICE if we don't.
This commit adds a new method `Instance.ty_env`, which takes a
`ParamEnv` for use during normalization. The MIR const-evaluator code is
changed to use this method, passing in the proper `ParamEnv` for the
context at hand.
rustdoc: Avoid panic when parsing codeblocks for playground links
`make_test` is also called when parsing codeblocks for the playground links so it should handle unwinds from the parser internally.
Fixes#63016
r? @GuillaumeGomez
The existing code seems to assume that substitutions spans are disjoint,
which is not always the case.
In the example:
pub trait AAAA {}
pub trait B {}
pub trait C {}
pub type T<P: AAAA + B + C> = P;
, we get three substituions starting from ':' and ending respectively at
the end of each trait token.
With the former offset calculation, this would cause `underline_start` to
eventually become negative before being converted to `usize`...
The new version may report erroneous results for non perfectly overlapping
substitutions but I don't know if such examples exist. Alternatively, we
could detect these cases and trim out overlapping substitutions.
Introduce an option for disabling deduplication of diagnostics
With the intent of using it in UI tests (https://github.com/rust-lang/rust/pull/67122).
The option is boolean (`-Z deduplicate-diagnostics=yes/no`) and can be specified multiple times with later values overriding earlier values (`-Z deduplicate-diagnostics=no -Z deduplicate-diagnostics=yes` == `-Z deduplicate-diagnostics=yes`), so it can be set in a hierarchical way, e.g. UI testing infra may disable the deduplication by default with specific tests being able to enable it back.
* Removes unnecessary feature flag from track_caller test.
* Tests of panic internals no longer need to explicitly construct Location.
* Add #![warn(const_err)] to retain-never-const per @oli-obk.
* Add track_caller test with diverging function.
tweak wording of mismatched delimiter errors
This PR improves the wording of the "incorrect delimiter" error messages. Here's a quick rationale:
- *"un-closed" -> "unclosed"*: "unclosed" is valid English, so there's no need to hyphenate the prefix. This should be pretty uncontroversial, I think.
- *"close delimiter" -> "closing delimiter"*: In my anecdotal experience, I've always heard "closing delimiter" or "closing parenthesis". In addition, the codebase already uses this terminology in comments and function names more than "close delimiter", which could indicate that it's more intuitive.
- "incorrect delimiter" -> "mismatched delimiter": "Incorrect delimiter" is vague; why is it incorrect? "mismatched" clearly indicates why the delimiter is causing the error.
r? @estebank
Implement uncommon_codepoints lint.
Part of #55467 . The checks of `$crate` and `{{root}}` are very unfortunate. But i'm not sure where they belongs to.
Nix reexports from `rustc_span` in `syntax`
Remove reexports `syntax::{source_map, symbol, edition}` and use `rustc_span` paths directly.
r? @petrochenkov
Add a test for #37333
The test checks that we reuse the CGU of a crate when the implementation
details of an `extern crate` have changed.
Signed-off-by: Michal Terepeta <michal.terepeta@gmail.com>
Suggest calling method when first argument is `self`
Closes: #66782
I've explored different approaches for this MR but I think the most straightforward is the best one.
I've tried to find out if the methods for given type exist (to maybe have a better suggestion), but we don't collect them anywhere and collecting them is quite problematic. Moreover, collecting all the methods would require rewriting big part of the code and also could potentially include performance degradation, which I don't think is necessary for this simple case.
- Use the span of the predicate
- Use the def's description instead of "struct/enum" (notably incorrect
for unions)
- Align formatting with other error messages
Attempt to fix intermittent failures of pgo-branch-weights test.
This PR tries to fix the intermittent failures of the pgo-branch-weights test (https://github.com/rust-lang/rust/issues/67746). The failing instances show no `!prof` annotations in LLVM IR. One possible cause is that the instrumented binary did not record anything. This is something I've occasionally seen happen for similarly small programs when using GNU ld as linker. The linker would not properly append the instruction counter sections, leading to most counters being dropped. This PR makes the test use the Gold linker instead.
It also makes each command exit immediately on failure so we can pinpoint the failure source better, should there still be a problem.
r? @Mark-Simulacrum
Rollup of 10 pull requests
Successful merges:
- #67450 (Allow for setting a ThinLTO import limit during bootstrap)
- #67595 (Suggest adding a lifetime constraint for opaque type)
- #67636 (allow rustfmt key in [build] section)
- #67736 (Less-than is asymmetric, not antisymmetric)
- #67762 (Add missing links for insecure_time)
- #67783 (Warn for bindings named same as variants when matching against a borrow)
- #67796 (Ensure that we process projections during MIR inlining)
- #67807 (Use drop instead of the toilet closure `|_| ()`)
- #67816 (Clean up err codes)
- #67825 (Minor: change take() docs grammar to match other docs)
Failed merges:
r? @ghost
