Re: [C++ Patch] PR 65579 ("gcc requires definition of a static constexpr member...")

[Jason Merrill <jason at redhat dot com>]

On Thu, Oct 26, 2017 at 6:17 AM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi again,
>
> On 24/10/2017 20:58, Jason Merrill wrote:
>>
>> This seems like an odd place to add the complete_type call.  What
>> happens if we change the COMPLETE_TYPE_P (type) in
>> cp_apply_type_quals_to_decl to COMPLETE_TYPE_P (complete_type (type))?
>
> Finally I'm back with some information.
>
> Simply doing the above doesn't fully work. The first symptom is the failure
> of g++.dg/init/mutable1.C which is precisely the testcase that you added
> together with the "|| !COMPLETE_TYPE_P (type)" itself: clearly, the
> additional condition isn't able anymore to do its work, because, first, when
> the type isn't complete, TYPE_HAS_MUTABLE_P (type) is false and then, when
> in fact it would be found true, we check !COMPLETE_TYPE_P (complete_type
> (type)) which is false, because completing succeeded.

> Thus it seems we need at least something like:
>
>    TREE_TYPE (decl) = type = complete_type (type);
>
>    if (TYPE_HAS_MUTABLE_P (type) || !COMPLETE_TYPE_P (type))
>      type_quals &= ~TYPE_QUAL_CONST;
>
> But then, toward the end of the testsuite, we notice a more serious issue,
> which is unrelated to the above: g++.old-deja/g++.pt/poi1.C
>
> // { dg-do assemble  }
> // Origin: Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
>
> template <class T>
> class TLITERAL : public T
>     {
>     int x;
>     };
>
> class GATOM;
>
> typedef TLITERAL<GATOM> x;
> extern TLITERAL<GATOM> y;
>
> also fails:
>
> poi1.C: In instantiation of ‘class TLITERAL<GATOM>’:
> poi1.C:13:24:   required from here
> poi1.C:5:7: error: invalid use of incomplete type ‘class GATOM’
>  class TLITERAL : public T
> poi1.C:10:7: note: forward declaration of ‘class GATOM’
>  class GATOM;
>
> that is, trying to complete GATOM at the 'extern TLITERAL<GATOM> y;" line
> obviously fails. Note, in case isn't obvious, that this happens exactly for
> the cp_apply_type_quals_to_decl call at the end of grokdeclarator which I
> tried to change in my first try: the failure of poi1.C seems rather useful
> to figure out what we want to do for this bug.
>
> Well, as expected, explicitly checking VAR_P && DECL_DECLARED_CONSTEXPR_P
> works again - it seems to me that after all it could make sense given the
> comment precisely talking about the additional complexities related to
> constexpr. Anyway, I'm attaching the corresponding complete patch.

Looking at the code again, it seems that the problem is the difference
between start_decl_1 and grokfield, in that the former has

  /* If an explicit initializer is present, or if this is a definition
     of an aggregate, then we need a complete type at this point.
     (Scalars are always complete types, so there is nothing to
     check.)  This code just sets COMPLETE_P; errors (if necessary)
     are issued below.  */
  if ((initialized || aggregate_definition_p)
      && !complete_p
      && COMPLETE_TYPE_P (complete_type (type)))
    {
      complete_p = true;
      /* We will not yet have set TREE_READONLY on DECL if the type
         was "const", but incomplete, before this point.  But, now, we
         have a complete type, so we can try again.  */
      cp_apply_type_quals_to_decl (cp_type_quals (type), decl);
    }

and grokfield/finish_static_data_member_decl don't.  How about
completing the type and re-applying the quals in
finish_static_data_member_decl if there's an initializer?  Your most
recent patch ought to work, but is less parallel.  Sorry for the
churn.

Jason