Currently the following code doesn't work when `callConstructor()` is called from Swift:
```cc
inline int increment(int value) {
return value + 1;
}
struct Incrementor {
int incrementee;
Incrementor(int value) : incrementee(increment(value)) {}
}
int callConstructor(int value) {
return Incrementor(value).incrementee;
}
```
The issue is that we don't generate `IR` for the `increment()` function when it's only called from a constructor or a method.
Swift is aware of the existence of `increment()` and we see it in `IR` as `declare incrementEi`, however, as we don't to emit a definition, we get the following error:
```
Incrementor::Incrementor(int): error: undefined reference to 'increment(int)'
```
This PR fixes this by visiting constructors and methods in `IRGen` and calling `HandleTopLevelDecl()` with all used declarations, which results in emitting definitions for the used declarations.
Co-authored-by: Marcel Hlopko <hlopko@google.com>
It is not completely uncommon for a record to get imported while
importing it's members (for example, if the member points back to the
parent record). In this case, simply use the already-imported record.
This should improve performance but also prevent an error where a member
accidentally had two parents.
This patch also moves around some of the member import/add logic to
allow for the above "optimization."
Add an convert to the new `target-swiftxx-frontend` substitution in lit
to control the C++ interop enabling in Swift. This allows for a single
site which will enable control of both an overridden standard (for
testing multiple C++ standards) and simplify writing tests.
The C++ interop modules require C++ support. Explicitly require C++ as
a feature when building these modules. This has no impact on the
changes as all the tests enable C++ already.
Use the FullyQualified<Type> abstraction from the prior commit plus DescriptiveDeclKind to give a bit more information when issuing a missing member type diagnostic during type resolution.
It's possible to have a "UnresolvedUsingValueDecl" with a "CXXConstructorName". If that's the case (or if there's any other unkown decl) bail instead of crashing.
* Replace `??1` with `??0` prefix in SILGen tests.
* Replace `call void <constructor>` with `call {{.*}} <constructor>`
because Windows ABI sometimes returns a pointer from the constructor for
non-trivial types.
* Make `noalias` an optional check.
* Check isAggregate instead of hasUserDeclaredConstructor.
* Rename addEmptyArgNamesForCxxFunc -> addEmptyArgNamesForClangFunction.
* Other minor fixes and cleanups.
* Update tests that relied on old behavior.
* Use mangleCXXName instead of mangleCXXCtor.
* Call VisitCXXRecordDecl not VisitRecordDecl from
VisitClassTemplateSpecializationDecl. This allows template constructors
to be imported and called correctly.
- Adapt tests to changes that have happened in the meantime (e.g.
`HasVirtualBase` is rightly no longer considered loadable)
- Don't import copy or move constructors (noticed this because
references are now imported correctly, so copy and move constructors
suddenly started showing up in the SIL test)
- Don't try to define an implicitly-deleted default constructor (this
previously broke loadable-types-silgen.swift)
The current "ClangRecordTypeInfo" derives from "LoadableTypeInfo" and is
only meant for loadable types. While we have not yet run into problems,
this may cause issues in the future and as more logic is needed around
copying, moving, and destroying C++ objects, this needs to be fixed.
