[cfe-commits] r83429 - in /cfe/trunk: lib/Sema/SemaOverload.cpp test/SemaCXX/builtin-ptrtomember-overload.cpp
Douglas Gregor
dgregor at apple.com
Wed Oct 7 09:01:13 PDT 2009
On Oct 6, 2009, at 4:08 PM, Fariborz Jahanian wrote:
> Author: fjahanian
> Date: Tue Oct 6 18:08:05 2009
> New Revision: 83429
>
> URL: http://llvm.org/viewvc/llvm-project?rev=83429&view=rev
> Log:
> Patch to implement C++ [over.built]p11 of overload resolution.
> Doug, please review. There is a FIXME in the test case with a question
> which is unrelated to this patch (that is, error is issued
> before set of builtins are added to the candidate list).
Very nice. Some comments below.
> case OO_ArrowStar:
> - // FIXME: No support for pointer-to-members yet.
> + // C++ [over.built]p11:
> + // For every quintuple (C1, C2, T, CV1, CV2), where C2 is a
> class type,
> + // C1 is the same type as C2 or is a derived class of C2, T
> is an object
> + // type or a function type, and CV1 and CV2 are cv-qualifier-
> seqs,
> + // there exist candidate operator functions of the form
> + // CV12 T& operator->*(CV1 C1*, CV2 T C2::*);
> + // where CV12 is the union of CV1 and CV2.
> + {
> + for (BuiltinCandidateTypeSet::iterator Ptr =
> + CandidateTypes.pointer_begin();
> + Ptr != CandidateTypes.pointer_end(); ++Ptr) {
> + QualType C1Ty = (*Ptr);
> + QualType C1;
> + if (const PointerType *PointerTy = C1Ty->getAs<PointerType>
> ()) {
> + C1 = PointerTy->getPointeeType();
> + C1 = Context.getCanonicalType(C1).getUnqualifiedType();
> + if (!isa<RecordType>(C1))
> + continue;
> + }
> + for (BuiltinCandidateTypeSet::iterator
> + MemPtr = CandidateTypes.member_pointer_begin(),
> + MemPtrEnd = CandidateTypes.member_pointer_end();
> + MemPtr != MemPtrEnd; ++MemPtr) {
> + const MemberPointerType *mptr = cast<MemberPointerType>
> (*MemPtr);
> + QualType C2 = QualType(mptr->getClass(), 0);
> + C2 = Context.getCanonicalType(C2).getUnqualifiedType();
> + if (C1 != C2 && !IsDerivedFrom(C1, C2))
> + break;
> + QualType ParamTypes[2] = { *Ptr, *MemPtr };
> + // build CV12 T&
> + QualType T = mptr->getPointeeType();
> + unsigned CV1 = (*Ptr).getCVRQualifiers();
> + unsigned CV2 = T.getCVRQualifiers();
For both CV1 and CV2, we should get the canonical type (of *Ptr and T)
before getting the CV qualifiers. Otherwise, we won't see qualifiers
applied within a typedef.
> + T = Context.getCVRQualifiedType(T, (CV1 | CV2));
> + QualType ResultTy = Context.getLValueReferenceType(T);
> + AddBuiltinCandidate(ResultTy, ParamTypes, Args, 2,
> CandidateSet);
> + }
> + }
> + }
> break;
> case OO_Conditional:
>
> Added: cfe/trunk/test/SemaCXX/builtin-ptrtomember-overload.cpp
> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/SemaCXX/builtin-ptrtomember-overload.cpp?rev=83429&view=auto
>
> =
> =
> =
> =
> =
> =
> =
> =
> ======================================================================
> --- cfe/trunk/test/SemaCXX/builtin-ptrtomember-overload.cpp (added)
> +++ cfe/trunk/test/SemaCXX/builtin-ptrtomember-overload.cpp Tue Oct
> 6 18:08:05 2009
> @@ -0,0 +1,19 @@
> +// RUN: clang-cc -fsyntax-only -verify %s -std=c++0x
> +
> +struct A {};
> +
> +struct B {
> + operator A*();
> +};
> +
> +struct C : B {
> +
> +};
> +
> +
> +void foo(C c, B b, int A::* pmf) {
> + // FIXME. Bug or correct? gcc accepts it. It requires
> derived-to-base followed by user defined conversion to work.
> + int j = c->*pmf; // expected-error {{left hand operand to ->* must
> be a pointer to class compatible with the right hand operand, but is
> 'struct C'}}
GCC is correct. I'm guessing it's because of this FIXME in
BuiltinCandidateTypeSet::AddTypesConvertedFrom:
// FIXME: Visit conversion functions in the base classes, too.
so, we're not seeing the conversion function B::operator A*() when we
look for user-defined conversions from C.
Thanks, Fariborz!
- Doug
More information about the cfe-commits
mailing list