Note the test cases in abstract_type_witnesses used to pass but are now
rejected. This is fine, because doing anything more complicated used to
crash, and the GSB would crash or misbehave with these examples.
I did this by requiring this in the typechecker. This will ensure that when we
emit a move, we are guaranteed to have a value decl ref that we can evaluate. It also ensures that we can't _move fields.
auto
The old syntax was
@opened("UUID") constraintType
Where constraintType was the right hand side of a conformance requirement.
This would always create an archetype where the interface type was `Self`,
so it couldn't cope with member types of opened existential types.
Member types of opened existential types is now a thing with SE-0309, so
this lack of support prevented writing SIL test cases using this feature.
The new syntax is
@opened("UUID", constraintType) interfaceType
The interfaceType is a type parameter rooted in an implicit `Self`
generic parameter, which is understood to be the underlying type of the
existential.
Fixes rdar://problem/93771238.
* [NFC] Add a getOptionalityDepth method to return the number of optionals a type has
* [NFC] Use getOptionalityDepth to remove existing code
* [AST] Add a new diagnostic note for suggesting optional chaining
* [CSDiagnostics] Tweak MissingOptionalUnwrapFailure to suggest using optional chaining on closure
* [Test] Add a couple of test cases for perform_optional_chain_on_closure
* [CSDiagnostics] Change DeclRefExpr casts to isa checks
* [AST] Update diagnostic phrasing
* [Sema] Use new diagnostic identifier and update comment
* [Test] Add a new test case for function type
* [Test] Update cfuncs_parse.swift tests to check for new diagnostic note
Calling computeExtendedNominal() won't catch the case where the
extended type did not exist at extension binding time but then
appeared later, like an inferred associated type witness for
example.
Remove a redundant bit of diagnostic logic now
that we better diagnose cases where an uncallable
variable is in an overload set for a mismatched
apply. Tweak another bit of diagnostic logic to
only apply to a single arg/param case, as that
seems to be what it's meant for.
I also created a SILType::isMoveOnly() helper that returns true if a type is
a move only wrapped type or a first class move only type. The verifier check
that move only types aren't copied in canonical SIL was rewired to use that as well.
This cannot be represented in the ABI because we expect that the
opaque return type can be instantiated from the generic arguments
of its declaring context. Reconstructing type metadata for the
DynamicSelfType would require opaque return type instantiation
to also take the self *value*.
Fixes rdar://72407668 / https://github.com/apple/swift/issues/56367
Diagnose uses of conformances in API when the conformance declaration
isn't imported by the local file. This uses the existing logic checking
for conformance availability combined with the implicit import check.
Report it as a warning until Swift 6 where it will be an error.
rdar://96437311
Upgrade the diagnostic from one that mentions the restriction to one
that
1) Mentions primary associated types at all
2) Removes the constraints
3) Points at the declaration of the protocol type so the user can fix it
The compiler enters "script" mode if there is a file called
`main.swift`, or if there is only one file. Parsing files as a script
means that anything in the file is interpreted as top-level code, which
is incompatible with a valid @main-annotated struct, so an error is
emitted.
Unfortunately, it is intentional in many cases, and the diagnostic
didn't provide anything actionable to indicate that the explicit main
function is intentional and that the file being passed in is not
actually a top-level context.
The new note indicates that passing `-parse-as-library` to the compiler
invocation will fix it if the explicit main function is intentional.
[Distributed] generic and inner test; without one edge case
[Distributed] fix distributed_thunk test; unsure about those extra hops, could remove later
[Distributed] Remove type pretending in getSILFunctionType; it is not needed
It seems our constant replacement in the earlier phases is enough, and
we don't need this trick at all.
[Distributed] Use thunk when calling cross-actor on DA protocols
When I began removing `convenience` from actor inits, I thought it would
be ok for an actor's extensions to be non-delegating. But that's wrong,
because the actor could be in a resilient module and still have its
properties change between stored and computed, meaning that there isn't
a practical way to allow this, unless if the extension and the actor
are in the same file.
For now, just re-ban this behavior before anybody notices :)
Implicitly imported decls may end up in inlinable code and break the
module API. This have been known to lead to deserialization crash and
could in theory break the generated swiftinterfaces files. Let's
explicitly check for such a case, keeping it to a warning until Swift 6
where we can make it an error.
rdar://95816286
For code such as the following:
```
let r = Regex {
/abc/
}
```
If RegexBuilder has not been imported, emit a
specialized diagnostic and fix-it to add
`import RegexBuilder` to the file.
Unfortunately we're currently prevented from
emitting the specialized diagnostic in cases where
the builder contains references to RegexBuilder
types, such as:
```
let r = Regex {
Capture {
/abc/
}
}
```
This is due to the fact that we bail from CSGen
due to the reference to `Capture` being turned
into an `ErrorExpr`. We ought to be able to
handle solving in the presence of such errors, but
for now I'm leaving it as future work.
rdar://93176036
This is possible because actors do not support inheritance. There
is one specific exception to that rule, which is that an actor
can inherit from `NSObject` just to support ObjC interop.
This means an actor is effectively a final class.
resolves rdar://87568153
