The following:
template< typename >
struct S;
template< typename T >
S< T >& f (S< T >& s) {
const typename S< T >::nested ignore;
return s;
}
template S< char >& f (S< char >&);
template< typename >
struct S {
struct nested { };
};
compiles with gcc, but not with clang:
$ clang -c /tmp/t.cpp
/tmp/t.cpp:6:20: error: implicit instantiation of undefined template 'S<char>'
const typename S< T >::nested ignore;
^
/tmp/t.cpp:10:21: note: in instantiation of function template specialization 'f<char>' requested here
template S< char >& f (S< char >&);
^
/tmp/t.cpp:2:8: note: template is declared here
struct S;
^
1 error generated.
I believe clang to be right in that, at the point of instantiation, the function f refers an incomplete definition of S. OTOH, a later specialization of S might provide the correct definition that makes the dependent 'nested' well-formed. Any opinions?
Both compilers are correct.
[temp.point]/p6, 8:
6 An explicit instantiation definition is an instantiation point for the specialization or specializations specified by the explicit instantiation.
8 A specialization for a function template [...] may have multiple points of instantiations within a translation unit, and in addition to the points of instantiation described above, for any such specialization that has a point of instantiation within the translation unit, the end of the translation unit is also considered a point of instantiation. [...] If two different points of instantiation give a template specialization different meanings according to the one definition rule (3.2), the program is ill-formed, no diagnostic required.
There are two points of instantiation for f<char>
: at the explicit instantiation definition, and at the end of the TU. Because those two points of instantiation would result in different meanings (as lookup for S<T>::nested
would yield different results), the program is ill-formed NDR.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With