https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98712
Patrick Palka <ppalka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2021-01-19
Status|UNCONFIRMED |NEW
Ever confirmed|0 |1
CC| |ppalka at gcc dot gnu.org
--- Comment #2 from Patrick Palka <ppalka at gcc dot gnu.org> ---
Confirmed.
It seems that when operator== is declared constexpr, we synthesize it eagerly
from check_bases_and_members before doing layout_class_type, which means that
at the point of synthesis the type's TYPE_FIELDS lacks the FIELD_DECLs that
correspond to its bases, so build_comparison_op ends up completely ignoring the
type's bases.
When operator== is not declared constexpr, we synthesize the method lazily at
the point of first use, and thus well after layout_class_type has been done
from finish_struct_1.
