[cfe-dev] Is it permitted to explicitly specialize a deleted function template?

[Jeff Walden]

Is it an error to explicitly specialize a deleted function template?  gcc and msvc seem to think it's fine, and both compile this example:

   template<int i> int f() = delete;
   template<> int f<4>() { return 4; }
   int main() { return f<4>(); }

In contrast clang (r152210) sees an explicit specialization of a deleted function template as a redefinition:

> [jwalden at wheres-wally tmp]$ ~/Programs/clang-build/build/Release/bin/clang++ -std=c++11 clang-delete-bug.cpp
> clang-delete-bug.cpp:2:16: error: redefinition of 'f'
> template<> int f<4>() { return 4; }
>                ^
> clang-delete-bug.cpp:2:16: note: previous definition is here
> clang-delete-bug.cpp:4:21: error: call to deleted function 'f'
> int main() { return f<4>(); }
>                     ^~~~
> clang-delete-bug.cpp:2:16: note: candidate function [with i = 4] has been explicitly deleted
> template<> int f<4>() { return 4; }
>                ^
> 2 errors generated.

The N2346 default/delete proposal says that "One can define a template with a deleted definition. Specialization and argument deduction occur as they would with a regular template function, but explicit specialization is not permitted." and proposes editing 14.7.3 to allow an explicit specialization to declare only non-deleted function templates.  But N3242 at least didn't make this change (I don't have the final spec), so I'm guessing this is a clang bug but don't know.  Who's right?

Jeff