If a class was introduced in an OS newer than the deployment target,
suppress the inference of @_staticInitializeObjCMetadata, because
the resulting eager initialization will crash when run on older OS's.
This is a stop-gap solution; we want the eager initialization code to
check availability before registering the class, but that requires more
effort.
Fixes the main part of SR-6203 / rdar://problem/35161939.
Now that we pass in the correct type metadata for 'Self', it is
sound for a class to conform to a protocol with a default implementation
for a method returning 'Self'.
Fixes <rdar://problem/23671426>.
Now that we pass in the correct type metadata for 'Self', it is
sound for a class to conform to a protocol with a default implementation
for a method returning 'Self'.
Fixes <rdar://problem/23671426>.
Replace a where Type-pointer-equality check with what it intended,
i.e., match up ParenTypes at the top level and perform a deeper
equality comparison of the underlying types.
Fixes SR-5166 / rdar://problem/32666189.
This time, the warnings only fire when the class in question directly
conforms to NSCoding. This avoids warning on cases where the user has
subclassed something like, oh, UIViewController, and has no intention
of writing it to a persistent file.
This also removes the warning for generic classes that conform to
NSCoding, for simplicity's sake. That means
'@NSKeyedArchiverEncodeNonGenericSubclassesOnly' is also being
removed.
Actually archiving a class with an unstable mangled name is still
considered problematic, but the compiler shouldn't emit diagnostics
unless it can be sure they are relevant.
rdar://problem/32314195
This is accomplished by recognizing this specific situation and
replacing the 'objc' attribute with a hidden '_objcRuntimeName'
attribute. This /only/ applies to classes that are themselves
non-generic (including any enclosing generic context) but that have
generic ancestry, and thus cannot be exposed directly to Objective-C.
This commit also eliminates '@NSKeyedArchiverClassName'. It was
decided that the distinction between '@NSKeyedArchiverClassName' and
'@objc' was too subtle to be worth explaining to developers, and that
any case where you'd use '@NSKeyedArchiverClassName' was already a
place where the ObjC name wasn't visible at compile time.
This commit does not update diagnostics to reflect this change; we're
going to change them anyway.
rdar://problem/32414557
None of the clients of this care about distinguishing between immediate
failures in getWitnessType() vs. earlier errors that result in
getWitnessType() returning ErrorType, so simplify the interface by
having it always return Type() if there is a problem.
Update the clients that were not already checking for null result to do
so.
Fixes rdar://problem/32215763.
Protocol requirements involving same-type-to-Self constraints cannot
be witnessed by declarations in non-final classes that have the same
form of same-type requirement to the corresponding class type, because
it creates a soundness hole with subclasses:
protocol Q {
func foo<T: P>(_: T, _: T.T) where T.T == Self
}
class C: Q {
func foo<T: P>(_: T, _: C) where T.T == C {}
}
class D: C {
// in D, T.T == D does not hold
}
Warn about this in Swift 3 compatibility mode, error on it in Swift 4
mode. When possible, provide a note + Fix-It suggesting that the
same-type constraint might be weakened to a superclass constraint
(which works with subclassing).
Fixes rdar://problem/30398503.
Adoption so far shows that the criteria we set up here are too broad.
This is particularly problematic for subclasses of NS/UIView and the
like that might never be encoded at all.
rdar://problem/32306355
Swift 3 allowed a class to explicitly conform to AnyObject, although
it was meaningless. Recent AnyObject-related changes started rejecting
such conformances as ill-formed; allow them with a warning + Fix-It in
Swift 3 compatibility mode.
When we have a potential assignment of associated types to type
witnesses during associated type inference, check that set of type
witnesses against the requirements in the requirement signature, so
that we can reject any solutions that fail some of the protocol's
requirements.
This is most of rdar://problem/31830524 --- but gets thwarted by the
inability of associated type inference to work across multiple
protocols.
Infer @_staticInitializeObjCMetadata in those cases where we need a
static initializer to make an NSCoding-conforming class visible to the
Objective-C runtime. This does *not* include classes with one of the
@NSKeyedArchive attributes:
* @NSKeyedArchiveLegacy implies that we'll register the class
directly, with the necessary side effect of initialize Objective-C
metadata.
* @NSKeyedArchiveSubclassesOnly promises not to archive the class
directly anyway.
Introduce the @NSKeyedArchiveSubclassesOnly attribute, which can be
placed on a class that conforms to NSCoding to suppress the
unstable-name diagnostics by promising to only archive
subclasses---not this class directly.
The diagnostic regarding NSCoding classes with unstable names can be
suppressed by adding @objc (the preferred solution for new code) or
@NSKeyedArchiveLegacy (for existing archives). Provide those as
Fix-Its, in that order.
(Thanks, Jordan!)
Currently inactive, this attribute indicates that a static initializer should be emitted to register the Objective-C metadata when the image is loaded, rather than on first use of the Objective-C metadata. Infer this attribute for NSCoding classes that won’t have static Objective-C metadata or have an @NSKeyedArchiveLegacy attributed.
This attribute allows one to provide the "legacy" name of a class for
the purposes of archival (via NSCoding). At the moment, it is only
useful for suppressing the warnings/errors about classes with unstable
archiving names.
The name mangling changed from Swift 3 to Swift 4, and may get slight
tweaks as we lock down ABI stability. Identify and warn about (in
Swift 3) or error about (in Swift 4) the cases where we don't have
obviously-stable name mangling, e.g.,
* private/fileprivate classes (whose mangled names involve the file name)
* nested classes (whose mangled names depend on their enclosing type)
* generic classes (whose mangled names involve the type arguments)
When an extension introduces a conformance that already exists, the
type checker will reject the conformance and then ignore it. In cases
where the original conformance is in the same module as the type or
protocol (but the new, redundant conformance is in some other module),
downgrade this error to a warning. This helps with library-evolution
cases where a library omitted a particular conformance for one of its
own types/protocols in a previous version, then introduces it in a new
version.
The specific driver for this is the conformance of String to
Collection, which affects source compatibility. Fixes
rdar://problem/31104415.
SourceKit always sets it positively. This may lead to more aggressive fixits however
less informative messages. We currently use the flag only for filling protocol stubs.
This reverts part of #4038 which made the compiler consider it to be an `Explicit` conformance, breaking source code that was accepted in Swift 3.0 which declared a raw type as well as explicit conformance to `RawRepresentable` (reported as rdar://problem/30386658). While I'm here, a couple of spot fixes:
- Ensure an enum's raw value exprs are type-checked before checking conformances of any of its extensions, since the RawRepresentable conformance derivation will blow up if the raw value exprs haven't been checked. Fixes an order dependency issue if `extension Foo: RawRepresentable {}` gets checked before `enum Foo: Int { ... }`.
- Don't display the custom `enum_declares_rawrep_with_raw_type` diagnostic if the source location for the enum's inheritance clause is invalid, so that we don't emit a dislocated diagnostic.
Until recently we didn't allow nested generic types at all.
In Swift 3, generic typealiases were added, and we forgot to
guard against them in witness matching, leading to a crash if
a generic typealias witnesses an associated type requirement.
Now that nested generic nominals are allowed too, add a check.
The diagnostic is not very good, but I'll revisit this later.
This fixes a regression from the previous patch which got
rid of PrintOptions::StripDynamicSelf.
When printing protocol declarations with a BaseType set in
PrintOptions, we can end up with a DynamicSelfType wrapping
a non-class type, if the protocol requirement returned
Self.
Note that this changes the diagnostic for missing protocol
requirements slightly; we used to sometimes refer to 'Self'
even if the conforming type is not a class, which is not
accepted by the type checker anyway. I believe the new
diagnostics are more correct.
The @objc optional requirement near-miss heuristics were too
permissive, and could occasionally produce ridiculous results that
were nowhere close to a "near" miss. Make the diagnostics more
conservative, and fix an issue with an errant sentinel value.
Fixes rdar://problem/26380688.
An overriding declaration doesn't have a choice about its signature,
because the signature is dictated by the overridden
declaration. Therefore, don't produce a warning for it.
Fixes rdar://problem/28524237.
