https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66338
--- Comment #3 from Jonathan Wakely <redi at gcc dot gnu.org> --- I think this is expected, because your type S has an unconstrained constructor template that accepts any argument, including objects of type tuple<S>, but in the signature of that constructor (specifically in the exception-specification) you use the argument in an expression that might be invalid. The tuple constructor is constexpr, so that invalid expression causes an error. I'm not sure, but this might be fixed by implementing this change to the standard: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4387
