http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56749



             Bug #: 56749

           Summary: weird interaction between scoped enum used as non-type

                    template parameter and template lookup

    Classification: Unclassified

           Product: gcc

           Version: 4.8.1

            Status: UNCONFIRMED

          Severity: normal

          Priority: P3

         Component: c++

        AssignedTo: unassig...@gcc.gnu.org

        ReportedBy: abiga...@gmail.com





Hi, g++-4.81 built from svn r197010 on x86_64-apple-darwin12.3.0 with clang3.2.

I find the following unexpected:





enum normal_enum

{

    not_scoped1,

    not_scoped2

};



enum class scoped_enum

{

    scoped1,

    scoped2

};



template <normal_enum N=not_scoped1>

class A

{

public:

    template <typename T>

        void fun ()

        {

        }

};



template <scoped_enum N=scoped_enum::scoped1>

class B

{

public:

    template <typename T>

        void fun ()

        {

        }

};





template <typename T>

void tfun ()

{

    A<> a;

    a.fun<char>(); <------------ THIS IS FINE



    B<> b_defaulted;

    B<scoped_enum::scoped1> b_explicited;



    //b_defaulted.fun<char>();          <------------ UNEXPECTED: THIS FAILS 

    b_defaulted.template fun<char>();   <------------ THIS IS FINE



    //b_explicited.fun<char>();         <------------ UNEXPECTED: THIS FAILS 

    b_explicited.template fun<char>();  <------------ THIS IS FINE

}



int main(int argc, char const *argv[])

{

    tfun<int>();

    return 0;

}





IOW, it seems like using a scoped enum as a non-type template for a class,

induces  a requirement for template disambiguation when invoking a member

template which is not happpening using an normal enum.

Note that neither gcc 4.7.2 nor clang have this behavior.

I assume this is not expected, as I would say "fun" is never a dependent name

in the snippet above as neither a, b_defaulted and b_eplicited are, regardless

of using a scoped enum or a normal enum as the non-type template.



Regards,

Andrea Bigagli.

Reply via email to