https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124312
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jason at gcc dot gnu.org,
| |ppalka at gcc dot gnu.org
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Seems the cast to that pointer type rather than ADDR_EXPR of a COMPONENT_REF of
a DECL_ARTIFICIAL member inside of that class is created by build_base_path:
386 /* Don't bother with the calculations inside sizeof; they'll ICE if
the
387 source type is incomplete and the pointer value doesn't matter.
In a
388 template (even in instantiate_non_dependent_expr), we don't have
vtables
389 set up properly yet, and the value doesn't matter there either;
we're
390 just interested in the result of overload resolution. */
391 if (uneval)
392 {
393 expr = build_nop (ptr_target_type, expr);
394 goto indout;
395 }
So, because it is in unevaluated context we don't bother (or can't?) do the
normal handling, and then are upset during the constant evaluation about that.
