https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87012
Alexandre Oliva <aoliva at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
CC| |aoliva at gcc dot gnu.org
Assignee|unassigned at gcc dot gnu.org |aoliva at gcc dot
gnu.org
--- Comment #3 from Alexandre Oliva <aoliva at gcc dot gnu.org> ---
Created attachment 45007
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45007&action=edit
combination of two potential fixes
The problem is that tsubst creates a new type variant of reference type for the
template instantiation, and passes it for convert_nontype_argument to create
the indirect_ref(nop_expr(var_decl)) using the reference type for the nop_expr,
but later on, strip_typedefs replaces the canonical reference type in the
nop_expr within the indirect_ref and verify_unstripped_args_1 doesn't like
that.
The attached patchlet fixes both spots, because I can't decide which change is
better. Canonicalizing the type would save strip_typedefs some effort, but
preserving symbolic typedef names might give us better messages at a later
time. Thoughts?
