[Patch, Fortran] PING: Revised DEFERRED bindings

[Tobias Burnus]

Daniel Kraft wrote:
> here's a rediffed version of my DEFERRED implementation, see below. 
> No regressions on a fresh test.
> Ok for trunk?
Yes, it is OK. Thanks for the patch. Can you update the wiki (GFortran
-> news, Fortran2003, Fortran2003status, Patchtracker) after check in?

Tobias
> -------- Original Message --------
> Subject: [Patch, Fortran] Revised DEFERRED bindings
> Date: Wed, 11 Feb 2009 15:18:20 +0100
> From: Daniel Kraft <d@domob.eu>
> To: Fortran List <fortran@gcc.gnu.org>, Tobias Burnus
> <burnus@net-b.de>,   gcc-patches <gcc-patches@gcc.gnu.org>
>
> Hi,
>
> this is a new version of my DEFERRED patch.  The previous one was
> already approved by Tobias, except that we discussed what gfortran
> should do about calls like this:
>
> TYPE(concrete_t) :: obj ! concrete_t extends abstract abstract_t
> CALL obj%abstract_t%proc ()
>
> IIRC, we concluded that calls of type-bound procedures (no matter if
> DEFERRED or not) are illegal in this case (i.e., with an ABSTRACT typed
> base-object).  The only change in the new patch is to enforce this, see
> abstract_type_5.f03 in the patch.  While this is in fact an ABSTRACT
> type fix, I think it fits in well with this patch.
>
> The changed code is in resolve.c, namely the new procedure
> check_typebound_baseobject, and its calls in resolve_compcall and
> resolve_typebound_call.
>
> Regression-testing at the moment on GNU/Linux-x86-32.  Ok for 4.5 (with
> this change) if no failures?
>
> Thanks,
> Daniel
>