https://gcc.gnu.org/bugzilla/show_bug.cgi?id=39970
--- Comment #10 from Brandon Nickell <bbnickell at gmail dot com> --- Well I did open this bug 11 years ago (time flies!) and C++0x was still in mid flight before it turned into C++11, so quite a bit has changed since this was originally reported (for instance there was no constexpr or decltype which is used in the previous comment). At the time there was nothing in the name mangling scheme that clarified how you would even handle mangling the global.member template parameter as it was unusual that it would even be accepted. FWIW, I suspect this test case is correctly rejected in Edison Design Group's front end but cannot immediately confirm. Thanks.
