Relate existential associated types with variance Invariant
Fixes#71550#72315
r? @nikomatsakis
The test case reported in that issue now errors with the following message ...
```
error[E0495]: cannot infer an appropriate lifetime for lifetime parameter 'a in function call due to conflicting requirements
--> /tmp/test.rs:25:5
|
25 | bad(&Bar(PhantomData), x)
| ^^^^^^^^^^^^^^^^^^^^^^^^^
|
note: first, the lifetime cannot outlive the lifetime `'a` as defined on the function body at 24:11...
--> /tmp/test.rs:24:11
|
24 | fn extend<'a, T>(x: &'a T) -> &'static T {
| ^^
note: ...so that reference does not outlive borrowed content
--> /tmp/test.rs:25:28
|
25 | bad(&Bar(PhantomData), x)
| ^
= note: but, the lifetime must be valid for the static lifetime...
note: ...so that the types are compatible
--> /tmp/test.rs:25:9
|
25 | bad(&Bar(PhantomData), x)
| ^^^^^^^^^^^^^^^^^
= note: expected `&'static T`
found `&T`
error: aborting due to previous error
For more information about this error, try `rustc --explain E0495`.
```
I could also add that test case if we want to have a weaponized one too.
Fixes#69977
When we parse a chain of method calls like `foo.a().b().c()`, each
`MethodCallExpr` gets assigned a span that starts at the beginning of
the call chain (`foo`). While this is useful for diagnostics, it means
that `Location::caller` will return the same location for every call
in a call chain.
This PR makes us separately record the span of the function name and
arguments for a method call (e.g. `b()` in `foo.a().b().c()`). This
`Span` is passed through HIR lowering and MIR building to
`TerminatorKind::Call`, where it is used in preference to
`Terminator.source_info.span` when determining `Location::caller`.
This new span is also useful for diagnostics where we want to emphasize
a particular method call - for an example, see
https://github.com/rust-lang/rust/pull/72389#discussion_r436035990
Don't lose empty `where` clause when pretty-printing
Previously, we would parse `struct Foo where;` and `struct Foo;`
identically, leading to an 'empty' `where` clause being omitted during
pretty printing. This will cause us to lose spans when proc-macros
involved, since we will have a collected `where` token that does not
appear in the pretty-printed item.
We now explicitly track the presence of a `where` token during parsing,
so that we can distinguish between `struct Foo where;` and `struct Foo;`
during pretty-printing
Don't create impl candidates when obligation contains errors
Fixes#72839
In PR #72621, trait selection was modified to no longer bail out early
when an error type was encountered. This allowed us treat `ty::Error` as
`Sized`, causing us to avoid emitting a spurious "not sized" error after
a type error had already occured.
However, this means that we may now try to match an impl candidate
against the error type. Since the error type will unify with almost
anything, this can cause us to infinitely recurse (eventually triggering
an overflow) when trying to verify certain `where` clauses.
This commit causes us to skip generating any impl candidates when an
error type is involved.
normalize adt fields during structural match checking
fixes#72896
currently only fixes the issue itself and compiles stage 1 libs.
I believe we have to use something else to normalize the adt fields here,
as I expect some partially resolved adts to cause problems 🤔
stage 1 libs and the test itself pass, not sure about the rest...
Will spend some more time looking into it tomorrow.
r? @pnkfelix cc @eddyb
improper ctypes: normalize return types and transparent structs
Fixes#66202.
See each commit individually (except the first which adds a test) for more detailed explanations on the changes made.
In summary, this PR ensures that return types are normalized before being checked for FFI-safety, and that transparent newtype wrappers are FFI-safe if the type being wrapped is FFI-safe (often true previously, but not if, after substitution, all types in a transparent newtype were zero sized).
Resolve E0584 conflict
Adds a new error code (`E0761`) to indicate ambiguity in module file names and an accompanying expanded description to resolve a conflict over `E0584`.
Resolves#73116
This commit ensures that if a `repr(transparent)` newtype's only
non-zero-sized field is FFI-safe then the newtype is also FFI-safe.
Previously, ZSTs were ignored for the purposes of linting FFI-safety
in transparent structs - thus, only the single non-ZST would be checked
for FFI-safety. However, if the non-zero-sized field is a generic
parameter, and is substituted for a ZST, then the type would be
considered FFI-unsafe (as when every field is thought to be zero-sized,
the type is considered to be "composed only of `PhantomData`" which is
FFI-unsafe).
In this commit, for transparent structs, the non-zero-sized field is
identified (before any substitutions are applied, necessarily) and then
that field's type (now with substitutions) is checked for FFI-safety
(where previously it would have been skipped for being zero-sized in
this case).
To handle the case where the non-zero-sized field is a generic
parameter, which is substituted for `()` (a ZST), and is being used
as a return type - the `FfiUnsafe` result (previously `FfiPhantom`) is
caught and silenced.
Signed-off-by: David Wood <david@davidtw.co>
This commit moves the check that skips unit return types to after
where the return type has been normalized - therefore ensuring that
FFI-safety lints are not emitted for types which normalize to unit.
Signed-off-by: David Wood <david@davidtw.co>
This commit adds a test of the improper ctypes lint, checking that
return type are normalized bethat return types are normalized before
being checked for FFI-safety, and that transparent newtype wrappers
are FFI-safe if the type being wrapped is FFI-safe.
Signed-off-by: David Wood <david@davidtw.co>
Revert #71956
...since it caused unsoundness in #73137. Also adds a reduced version of #73137 to the test suite. The addition of the `MaybeInitializedLocals` dataflow analysis has not been reverted, but it is no longer used.
Presumably there is a more targeted fix, but I'm worried that other bugs may be lurking. I'm not yet sure what the root cause of #73137 is.
This will need to get backported to beta.
r? @tmandry
This is only really useful in debug messages, so I've switched to
calling `span_to_string` in any place that causes a `Span` to end up in
user-visible output.
Previously, we would parse `struct Foo where;` and `struct Foo;`
identically, leading to an 'empty' `where` clause being omitted during
pretty printing. This will cause us to lose spans when proc-macros
involved, since we will have a collected `where` token that does not
appear in the pretty-printed item.
We now explicitly track the presence of a `where` token during parsing,
so that we can distinguish between `struct Foo where;` and `struct Foo;`
during pretty-printing
Free `default()` forwarding to `Default::default()`
It feels a bit redundant to have to say `Default::default()` every time I need a new value of a type that has a `Default` instance.
Especially so, compared to Haskell, where the same functionality is called `def`.
Providing a free `default()` function that forwards to `Default::default()` seems to improve the situation.
The trait is still there, so if someone wants to be explicit and to say `Default::default()` - it still works, but if imported as `std::default::default;`, then the free function reduces typing and visual noise.
Add `-Z span-debug` to allow for easier debugging of proc macros
Currently, the `Debug` impl for `proc_macro::Span` just prints out
the byte range. This can make debugging proc macros (either as a crate
author or as a compiler developer) very frustrating, since neither the
actual filename nor the `SyntaxContext` is displayed.
This commit adds a perma-unstable flag `-Z span-debug`. When enabled,
the `Debug` impl for `proc_macro::Span` simply forwards directly to
`rustc_span::Span`. Once #72618 is merged, this will start displaying
actual line numbers.
While `Debug` impls are not subject to Rust's normal stability
guarnatees, we probably shouldn't expose any additional information on
stable until `#![feature(proc_macro_span)]` is stabilized. Otherwise,
we would be providing a 'backdoor' way to access information that's
supposed be behind unstable APIs.
Update annotate-snippets-rs to 0.8.0
#59346
I made major changes to this library. In the previous version we worked with owned while in the current one with borrowed.
I have adapted it without changing the behavior.
I have modified the coverage since the previous one did not return correctly the index of the character in the line.
When creating default values a trait method needs to be called with an
explicit trait name. `Default::default()` seems redundant. A free
function on the other hand, when imported directly, seems to be a better
API, as it is just `default()`. When implementing the trait, a method
is still required.
