https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81702
--- Comment #8 from Nathan Sidwell <nathan at gcc dot gnu.org> --- Ah, I remember now. The ABI specifies the data layout of the various RTTI objects like _ZTVN10__cxxabiv117__class_type_infoE. It does not specify the vtable layout -- implementations are free to add additional member functions: (ABI doc) 2.9.4 The class definitions below are to be interpreted as implying a memory layout following the class layout rules for the host ABI. They specify data members only, except for the Standard-specified member functions of the std::type_info class given below, and do not imply anything about the member functions of these classes. Virtual member functions of these classes may only be used within the target systems' respective runtime libraries. The data members must be laid out exactly as specified. So the vtable is defined when the definition(s) of our implementations of those structures is seen. >From random user code, we generate the type info object and reference the (undefined) vtable. Ergo, I believe eliding the assert is the right thing.
