================
@@ -14527,20 +14525,24 @@ Sema::CreateOverloadedUnaryOp(SourceLocation OpLoc,
UnaryOperatorKind Opc,
UnaryOperator::getOpcodeStr(Opc), OpLoc);
return ExprError();
- case OR_Deleted:
+ case OR_Deleted: {
// CreateOverloadedUnaryOp fills the first element of ArgsArray with the
// object whose method was called. Later in NoteCandidates size of
ArgsArray
// is passed further and it eventually ends up compared to number of
// function candidate parameters which never includes the object parameter,
// so slice ArgsArray to make sure apples are compared to apples.
+ StringLiteral *Msg = Best->Function->getDeletedMessage();
CandidateSet.NoteCandidates(
PartialDiagnosticAt(OpLoc, PDiag(diag::err_ovl_deleted_oper)
<< UnaryOperator::getOpcodeStr(Opc)
+ << !!Msg
----------------
erichkeane wrote:
> > it is something that we don't ever do
>
> Weird, I recall seeing that in other places in the codebase which is the only
> reason why I used it; I’d normally just write e.g. `bool(Msg)` or `Msg !=
> nullptr` etc. instead.
Yikes! If you see it, feel free to NFC fix those :) I personally find that
pattern absurdly distracting, and it isn't really necessary most of the time
(other than with our operator <<'s), since pointer-to-bool conversions are well
defined. The `!= nullptr` is, IMO, the 'most readable' and clear.
https://github.com/llvm/llvm-project/pull/86526
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
