We noticed that a project failed to build with prepared overloads enabled,
but built successfully with prepared overloads disabled.
It turns out that the code clearly should never have been accepted in the
first place, because the type checker was making an arbitrary choice between
two nominal type declarations with the same name.
Further inspection revealed this was because of a FIXME added in 2013 which
was far too broad. Tighten up the logic here to only disambiguate if at least
one of the two declarations is a type alias, and it has the same underlying
type as the other one. A choice between unrelated nominal type declarations
should always be ambiguous.
This still isn't perfect, because we might have two generic type aliases
that are referenced in both solutions with different substitution maps,
in which case we will still erroneously pick the first one. But this
is better than the old logic, at least.
Fixes rdar://165863775.
If arguments are represented by a single tuple it's possible
that the issue is not about missing parameters but instead
about tuple destructuring. Fix `fixMissingArguments` to check for
overruns after destructuring and stop if that produces more
arguments then parameters because such situations are better
diagnosed as a general conversion failure rather than a missing
argument(s) problem.
Resolves: rdar://159408715
This fixes an ambiguity introduced by the stdlib change in
0f99458900.
Since (borrowing T) -> () and (T) -> () both convert to
each other, we could end up with ambiguous solutions where
neither one was better than the other. Generalize the
existing trick we use for labeled vs unlabeled tuples to
also strip off ownership specifiers and @convention(...)
from function types. This fixes the regression, as well
an existing FIXME in a test I added a while ago where
the same problem arises with @convention(block).
All linux and Android CI show this being repeated many times, eg when
[building the Docc binary on the linux
CI](https://ci.swift.org/job/swift-PR-Linux-smoke-test/24247/consoleText):
```
"/home/build-user/build/buildbot_linux/unified-swiftpm-build-linux-x86_64/x86_64-unknown-linux-gnu/release/docc.build/main.swift.o" "-lswiftObservation" ... repeated many times ... "-lswiftObservation" "-lswiftCore"
```
This reverts commit 6852bc9834.
In addition to the original change, this makes sure that C++ `std::string` and Swift `String` are given distinct score, in order to prevent ambiguity which was causing build failures in some projects.
rdar://158439395
We now report properties referencing types imported from
implementation-only dependencies in non-library-evolution mode as the
memory layouts of structs and enums may be exposed to clients. Here we
exempt non-open classes from this check as clients hold pointers to
classes which hides the class internal memory layouts. Keep forbidding
it for open classes in order to error early, a client trying to subclass
an open class with such a reference from their property would already
see an error about members not deserializable.
Exclude properties with initial values from the memberwise initializer
if they are less accessible than the most accessible property, up to
`internal`. Introduce a compatibility overload that continues to
include the same properties as before until the next language mode.
This is gated behind the `ExcludePrivateFromMemberwiseInit` feature.
rdar://122416579
If the property isn't memberwise initializable we need to emit its
initial value and call the init accessor. This currently isn't necessary
since all init accessor properties are included in the memberwise
initializer, but the following commit will change this.
Currently `init` accessors are always memberwise initialized, but I'm
planning on changing that. As such we need to make sure we record
the `init` accessor regardless of whether it is included in the memberwise
init.
This serves as the backing request for both InitAccessorPropertiesRequest
and MemberwiseInitPropertiesRequest, ensuring the former does not
care about the init accessor is included in the memberwise initializer.
Store the original VarDecl in the same map we use for tracking the
original wrapper var for property wrappers. This will allow us to
use the same logic to determine the original var for both.
rdar://163631865
Under certain circumstances the same symbol can be used for different implementations of the forwarders.
The forwarders themselves are already emitted with "internal" LLVM linkage, but some particularities in
the mapping from SILLinkage to LLVM linkage caused async function pointers to partial apply forwarders
to be emitted with linkonce_odr instead, which caused the wrong forwarder to be called at runtime, causing
unexpected behavior and crashes.
Ensure that we nativize the path and consider only the filename starting
prefix to account for file suffixes (e.g. `.exe` on Windows). This
ensures that we properly support other platforms.
If we didn't find an extension result, try again disregarding
invertible requirements since `demangleGenericSignature` won't
include them. This is just meant to be a quick low risk fix that we
can cherry-pick, the proper fix here is to delete all this logic and
just return the nominal along with the ABI module name to filter
lookup results.
rdar://165639044
