https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64372
Harald van Dijk <harald at gigawatt dot nl> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |harald at gigawatt dot nl
--- Comment #2 from Harald van Dijk <harald at gigawatt dot nl> ---
(In reply to Manuel López-Ibáñez from comment #1)
> Probably even:
>
> const char&
> foz(const char* p)
> {
> return *p;
> }
>
> is undefined.
No, that is fine. *p is an lvalue already, so no temporary gets created.
The original code was supposed to get a warning because ptr ? *ptr : throw ptr
was not an lvalue: for a ? b : c to be an lvalue, both b and c needed to be
lvalues.
However, the C++ standard has since changed, and if one operand is a
throw-expression, and the other is an lvalue, the result is now an lvalue. CWG
issue 1560:
http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1560
So in -std=c++11 mode, there should be no temporary, and no warning. (I haven't
checked if current versions of GCC already implement that, but I can find no
bug reports about it.)
