Nathan Ridge <zeratul976@xxxxxxxxxxx> writes: >> > Is the code ill-formed and gcc thus non-conforming? >> >> Those are two separate questions, > > Doesn't accepting invalid code make a compiler non-conforming? There are a number of cases where the standard doesn't require a conforming implementation to do anything in particular for an ill-formed program. >> but as far as I can tell, this code is >> indeed ill-formed, and gcc should give an error for it. Please consider >> filing an accepts-valid bug as described at http://gcc.gnu.org/bugs/ . > > I have filed http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48409. Thanks. >> > I ask because libstdc++ uses specializations like this for std::is_function in , >> > and as a result, clang refuses to compile anything that includes . >> >> I don't see that in the current . I see something >> different in : the function is a template parameter. The >> const qualifier is permitted if the function type is a >> pointer-to-member. I think that may be possible in . In >> that case, the const qualifier is not an error, and any use of it with >> something that is not a pointer-to-member should be dropped due to >> SFINAE. But I am not a C++ expert, and I could be wrong here. > > I was referring to the following: > > template<typename> > struct is_function > : public false_type { }; > template<typename _Res, typename... _ArgTypes> > struct is_function<_Res(_ArgTypes...)> > : public true_type { }; > template<typename _Res, typename... _ArgTypes> > struct is_function<_Res(_ArgTypes...) const> > : public true_type { }; Yes, me too, but that is very different from the example in your original message. Your example was about template specialization. The code quoted above is template definitions. Ian
