[Bug c++/79424] New: dtor synthesized before abstractness correctly determined

["nathan at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79424

            Bug ID: 79424
           Summary: dtor synthesized before abstractness correctly
                    determined
           Product: gcc
           Version: 7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: nathan at gcc dot gnu.org
  Target Milestone: ---

Created attachment 40694
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=40694&action=edit
Exemplar

I stumbled upon this looking at 79393. In the example, we think A2::~A2 is
deleted because we haven't yet determined it's abstractness.  Consequently we
think we need VDT::~VDT, which is private.  Then we give an error about
overriding non-deleted A1::~A1 with a deleted function.  Further fallout when
the cleanup for global a3 attempts to use deleted A3::~A3 (which is deleted
because A2::~A2 was erroneously deleted).

dfs_declare_virt_assop_and_dtor calls lazily_declare_fn (sfk_destructor, t)
before CLASSTYPE_PURE_VIRTUALS is determined (we've not looked at inherited
pure virts, nor local overriders for such inherited pure virts.  It only
contains the pure virts declared in the class).

That's from check_bases_and_members called early in from finish_struct_1.  Pure
virtuals are finally determined from finish_struct_bits called much later in
finish_struct_1.