cleanup: remove zero-offset GEP
This GEP would've been used to change the pointer type in the past, but after opaque pointers it's a no-op. I missed removing this in #105545.
Split out from #121577.
Fix linting paths with qself in `unused_qualifications`
Fixes#121999
`resolve_qpath` ends up being called again with `qself` set to `None` to check trait items from fully qualified paths. To avoid this the lint is moved to a place that accounts for this already
96561a8fd1/compiler/rustc_resolve/src/late.rs (L4074-L4088)
r? `````@petrochenkov`````
Removing absolute path in proc-macro
With rust 1.75 the absolute build path name is embedding into proc-macro (.rustc section) and which causes reproducibility issues.
Detailed issue description is here - https://github.com/rust-lang/rust/issues/120825#issuecomment-1964307219
With this change the 'absolute path' changed back to '/rust/$hash' format as in earlier revisions.
Rework `untranslatable_diagnostic` lint
Currently it only checks calls to functions marked with `#[rustc_lint_diagnostics]`. This PR changes it to check calls to any function with an `impl Into<{D,Subd}iagnosticMessage>` parameter. This greatly improves its coverage and doesn't rely on people remembering to add `#[rustc_lint_diagnostics]`. It also lets us add `#[rustc_lint_diagnostics]` to a number of functions that don't have an `impl Into<{D,Subd}iagnosticMessage>`, such as `Diag::span`.
r? ``@davidtwco``
avoid overlapping privacy suggestion for single nested imports
Fixes#114884
This PR aims to avoid confusion inside braces for import suggestions.
r? ``@petrochenkov``
Change the documented implicit value of `-C instrument-coverage` to `=yes`
The option-value parser for `-Cinstrument-coverage=` currently accepts the following stable values:
- `all` (implicit value of plain `-Cinstrument-coverage`)
- `yes`, `y`, `on`, `true` (undocumented aliases for `all`)
- `off` (default; same as not specifying `-Cinstrument-coverage`)
- `no`, `n`, `false`, `0` (undocumented aliases for `off`)
I'd like to rearrange and re-document the stable values as follows:
- `no` (default; same as not specifying `-Cinstrument-coverage`)
- `n`, `off`, `false` (documented aliases for `no`)
- `0` (undocumented alias for `no`)
- `yes` (implicit value of plain `-Cinstrument-coverage`)
- `y`, `on`, `true` (documented aliases for `yes`)
- `all` (documented as *currently* an alias for `yes` that may change; discouraged but not deprecated)
The main changes being:
- Documented default value changes from `off` to `no`
- Documented implicit value changes from `all` to `yes`
- Other boolean aliases (`n`, `off`, `false`, `y`, `on`, `true`) are explicitly documented
- `all` remains currently an alias for `yes`, but is explicitly documented as being able to change in the future
- `0` remains an undocumented but stable alias for `no`
- The actual behaviour of coverage instrumentation does not change
# Why?
The choice of `all` as the implicit value only really makes sense in the context of the unstable `except-unused-functions` and `except-unused-generics` values. That arrangement was fine for an unstable flag, but it's confusing for a stable flag whose only other stable value is `off`, and will only become more confusing if we eventually want to stabilize other fine-grained coverage option values.
(Currently I'm not aware of any plans to stabilize other coverage option values, but that's why I think now is a fine time to make this change, well before anyone actually has to care about it.)
For example, if we ever add support for opt-in instrumentation of things that are *not* instrumented by `-Cinstrument-coverage` by default, it will be very strange for the `all` value to not actually instrument all things that we know how to instrument.
# Compatibility impact
Because this is not a functional change, there is no immediate compatibility impact. However, changing the documented semantics of `all` opens up the possibility of future changes that could be considered retroactively breaking.
I don't think this is going to be a big deal in practice, for a few reasons:
- The exact behaviour of coverage instrumentation is allowed to change, so changing the behaviour of `all` is not a *stability-breaking* change, as long as it still exists and does something reasonable.
- `-Cinstrument-coverage` is mainly used by tools or scripts that can be easily updated if necessary. It's unusual for users to pass the flag directly, because processing the profiler output is complicated enough that tools/scripts tend to be necessary anyway.
- Most people who are using coverage are probably relying on `-Cinstrument-coverage` rather than explicitly passing `-Cinstrument-coverage=all`, so the number of users actually affected by this change is likely to be low, and plausibly zero.
bootstrap/libtest: print test name eagerly on failure even with `verbose-tests=false` / `--quiet`
Previously, libtest would wait until all tests finished running to print the progress, which made it
annoying to run many tests at once (since you don't know which have failed). Change it to print the
names as soon as they fail.
This makes it much easier to know which test failed without having to wait for compiletest to completely finish running. Before:
```
Testing stage0 compiletest suite=ui mode=ui (x86_64-unknown-linux-gnu)
running 15274 tests
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 88/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 176/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 264/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 352/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 440/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 528/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiFFiiiiiii
...
```
After:
```
Testing stage0 compiletest suite=ui mode=ui (x86_64-unknown-linux-gnu)
running 15274 tests
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 88/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 176/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 264/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 352/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 440/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 528/15274
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
[ui] tests/ui/associated-type-bounds/implied-in-supertrait.rs ... F
[ui] tests/ui/associated-type-bounds/return-type-notation/basic.rs#next_with ... F
iiiiiiiiiiiii
...
```
This serves a similar use case to the existing RUSTC_TEST_FAIL_FAST, but is on by default and as a result much more discoverable. We should consider unifying RUSTC_TEST_FAIL_FAST with the `--no-fail-fast` flag in the future for consistency and discoverability.
This commit is extracted from #122036 and adds a new directive to the
`compiletest` test runner, `//@ needs-threads`. This is intended to
capture the need that a target must implement threading to execute a
specific test, typically one that uses `std::thread`. This is primarily
done for WebAssembly targets which currently do not have threads by
default. This enables transitioning a lot of `//@ ignore-wasm*`-style
ignores into a more self-documenting `//@ needs-threads` directive.
Additionally the `wasm32-wasi-preview1-threads` target, for example,
does actually have threads, but isn't tested in CI at this time. This
change enables running these tests for that target, but not other wasm
targets.
Replace libLLVM symlink with linker script
It turns out that the libLLVM-N.so -> libLLVM.so.N.1 symlink is also needed when projects like miri link against librustc_driver.so. As such, we have to distribute it in real rustup components like rustc-dev, rather than only for download-ci-llvm.
To avoid actually distributing symlinks (which are not supported or not fully supported by the rustup infrastructure) replace it with a linker script that does the same thing instead.
Fixes https://github.com/rust-lang/rust/issues/121889.
r? `@cuviper`
stricter hidden type wf-check [based on #115008]
Original work by `@aliemjay` in #115008. A huge thanks to them for originally figuring out this approach ❤️
Fixes https://github.com/rust-lang/rust/issues/114728
Fixes https://github.com/rust-lang/rust/issues/114572
Instead of adding the `WellFormed` obligations when relating opaque types, we now always emit such an obligation when defining the hidden type.
This causes nested opaque types which aren't wf to error, see the comment below for the described impact. I believe this change to be desirable as it significantly reduces complexity by removing special-cases.
It also caused an issue with RPITIT: in defaulted trait methods, we add a `Projection(synthetic_assoc, rpit_of_trait_method)` clause to the `param_env`. This clause is not added to the `ParamEnv` of the nested coroutines. This caused a normalization failure in `fn check_coroutine_obligations` with the new solver. I fixed that by using the env of the typeck root instead.
r? `@oli-obk`
Embed length of offset/position into Span tag byte
This cuts the average bytes/relative span from 3.5 to 3.2 on libcore, ultimately saving ~400kb of data.
C++ style guides I am aware of recommend specifically preferring = syntax
for any classes with fairly obvious constructors[^0] that do not perform
any complicated logic in their constructor. I contend that all constructors
that the `rustc_llvm` code uses qualify. This has only become more common
since C++ 17 guaranteed many cases of copy initialization elision.
The other detail is that I tried to ask another contributor with
infinitely more C++ experience than me (i.e. any) what this constructor
syntax was, and they thought it was a macro. I know of no other language
that has adopted this same syntax. As the rustc codebase features many
contributors experienced in many other languages, using a less...
unique... style has many other benefits in making this code more
lucid and maintainable, which is something it direly needs.
[^0]: e.g. https://abseil.io/tips/88
This makes it easier to see that the call to `in_scope` returns both the then
block and the else block. The rather confusing `unpack!` step is confined to
its own separate line.
(This patch reindents several lines, so using "ignore whitespace" is
recommended in order to focus on the actual changes.)
Prior to the previous commit, `#[rust_lint_diagnostics]` attributes
could only be used on methods with an `impl Into<{D,Subd}iagMessage>`
parameter. But there are many other nearby diagnostic methods (e.g.
`Diag::span`) that don't take such a parameter and should have the
attribute.
This commit adds the missing attribute to these `Diag` methods. This
requires adding some missing
`#[allow(rustc::diagnostic_outside_of_impl)]` markers at call sites to
these methods.
Currently it only checks calls to functions marked with
`#[rustc_lint_diagnostics]`. This commit changes it to check calls to
any function with an `impl Into<{D,Subd}iagMessage>` parameter. This
greatly improves its coverage and doesn't rely on people remembering to
add `#[rustc_lint_diagnostics]`.
The commit also adds `#[allow(rustc::untranslatable_diagnostic)`]
attributes to places that need it that are caught by the improved lint.
These places that might be easy to convert to translatable diagnostics.
Finally, it also:
- Expands and corrects some comments.
- Does some minor formatting improvements.
- Adds missing `DecorateLint` cases to
`tests/ui-fulldeps/internal-lints/diagnostics.rs`.
Rollup of 9 pull requests
Successful merges:
- #121065 (Add basic i18n guidance for `Display`)
- #121744 (Stop using Bubble in coherence and instead emulate it with an intercrate check)
- #121829 (Dummy tweaks (attempt 2))
- #121857 (Implement async closure signature deduction)
- #121894 (const_eval_select: make it safe but be careful with what we expose on stable for now)
- #122014 (Change some attributes to only_local.)
- #122016 (will_wake tests fail on Miri and that is expected)
- #122018 (only set noalias on Box with the global allocator)
- #122028 (Remove some dead code)
r? `@ghost`
`@rustbot` modify labels: rollup
This always produces zero offset, regardless of what the struct layout
is.
Originally, this may have been necessary in order to change the pointer type,
but with opaque pointers, it is no longer necessary.
Rollup of 8 pull requests
Successful merges:
- #121202 (Limit the number of names and values in check-cfg diagnostics)
- #121301 (errors: share `SilentEmitter` between rustc and rustfmt)
- #121658 (Hint user to update nightly on ICEs produced from outdated nightly)
- #121846 (only compare ambiguity item that have hard error)
- #121961 (add test for #78894#71450)
- #121975 (hir_analysis: enums return `None` in `find_field`)
- #121978 (Fix duplicated path in the "not found dylib" error)
- #121991 (Merge impl_trait_in_assoc_types_defined_by query back into `opaque_types_defined_by`)
r? `@ghost`
`@rustbot` modify labels: rollup
only set noalias on Box with the global allocator
As discovered in https://github.com/rust-lang/miri/issues/3341, `noalias` and custom allocators don't go well together.
rustc can now check whether a Box uses the global allocator. This replaces the previous ad-hoc and rather unprincipled check for a zero-sized allocator.
This is the rustc part of fixing that; Miri will also need a patch.
const_eval_select: make it safe but be careful with what we expose on stable for now
As this is all still nightly-only I think `````@rust-lang/wg-const-eval````` can do that without involving t-lang.
r? `````@oli-obk`````
Cc `````@Nilstrieb````` -- the updated version of your RFC would basically say that we can remove these comments about not making behavior differences visible in stable `const fn`
Implement async closure signature deduction
Self-explanatory from title.
Regarding the interaction between signature deduction, fulfillment, and the new trait solver: I'm not worried about implementing closure signature deduction here because:
1. async closures are unstable, and
2. I'm reasonably confident we'll need to support signature deduction in the new solver somehow (i.e. via proof trees, which seem very promising).
This is in contrast to #109338, which was closed because it generalizes signature deduction for a *stable* kind of expression (`async {}` blocks and `Future` traits), and which proliferated usage may pose a stabilization hazard for the new solver.
I'll be certain to make sure sure we revisit the closure signature deduction problem by the time that async closures are being stabilized (which isn't particularly soon) (edit: Put it into the async closure tracking issue). cc `````@lcnr`````
r? `````@oli-obk`````
Stop using Bubble in coherence and instead emulate it with an intercrate check
r? `````@compiler-errors`````
This change is kinda funny, because all I've done is reimplement `Bubble` behaviour for coherence without using `Bubble` explicitly.
Add basic i18n guidance for `Display`
I've tried to be relatively noncommittal here. The part I think is most important is to mention the concept of "display adapters" *somewhere* in the `std::fmt` documentation that has some chance of being discovered when people go looking for ways to provide context when `Display`ing their type.
Rendered:
> ### Internationalization
>
> Because a type can only have one `Display` implementation, it is often preferable to only implement `Display` when there is a single most "obvious" way that values can be formatted as text. This could mean formatting according to the "invariant" culture and "undefined" locale, or it could mean that the type display is designed for a specific culture/locale, such as developer logs.
>
> If not all values have a justifiably canonical textual format or if you want to support alternative formats not covered by the standard set of possible [formatting traits], the most flexible approach is display adapters: methods like [`str::escape_default`] or [`Path::display`] which create a wrapper implementing `Display` to output the specific display format.
>
> [formatting traits]: https://doc.rust-lang.org/nightly/std/fmt/index.html#formatting-traits
> [`str::escape_default`]: https://doc.rust-lang.org/nightly/std/primitive.str.html#method.escape_default
> [`Path::display`]: https://doc.rust-lang.org/nightly/std/path/struct.Path.html#method.display
The module docs do already have a [localization header](https://doc.rust-lang.org/nightly/std/fmt/index.html#localization), so maybe this header should be l10n instead of i18n, or maybe this information should live under that header? I'm not sure, but here on the `Display` trait at least isn't a *bad* spot to put it.
The other side of this that comes up a lot is `FromStr` compatibility, but that's for a different PR.
Merge impl_trait_in_assoc_types_defined_by query back into `opaque_types_defined_by`
Instead, when we're collecting opaques for associated items, we choose the right collection mode depending on whether we're collecting for an associated item of a trait impl or not.
r? ```@compiler-errors```
follow up to https://github.com/rust-lang/rust/pull/121838
Fix duplicated path in the "not found dylib" error
While working on the gcc backend, I couldn't figure out why I had this error:
```
error: couldn't load codegen backend /checkout/compiler/rustc_codegen_gcc/target/release/librustc_codegen_gcc.so/checkout/compiler/rustc_codegen_gcc/target/release/librustc_codegen_gcc.so: cannot open shared object file: No such file or directory
```
As you can see, the path is duplicated for some reason. After investigating a bit more, I realized that `libloading::Error::LoadLibraryExW` starts with the path of the not found dylib, making it appear twice in our error afterward (because we do render it like this: `{path}{err}`, and since the `err` starts with the path...).
Thanks to `````@bjorn3````` for linking me to https://github.com/rust-lang/rust/pull/121392. :)
hir_analysis: enums return `None` in `find_field`
Fixes#121757.
Unnamed union fields with enums are checked for, but if `find_field` causes an ICE then the compiler won't get to that point.
