[Bug fortran/55282] [OOP] openmp directive and classes

2013-03-17 Thread janus at gcc dot gnu.org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



janus at gcc dot gnu.org changed:



   What|Removed |Added



 Status|UNCONFIRMED |RESOLVED

 Resolution||DUPLICATE



--- Comment #7 from janus at gcc dot gnu.org 2013-03-17 17:06:41 UTC ---

resolving as duplicate of PR 52531



*** This bug has been marked as a duplicate of bug 52531 ***


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-13 Thread burnus at gcc dot gnu.org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



Tobias Burnus burnus at gcc dot gnu.org changed:



   What|Removed |Added



 CC||burnus at gcc dot gnu.org



--- Comment #5 from Tobias Burnus burnus at gcc dot gnu.org 2012-11-13 
10:44:14 UTC ---

(In reply to comment #1)

 Note that Fortran 2003 is not supported in OpenMP 3.1. This may change with

 OpenMP 4, but I'm not sure of that.



The OpenMP ARB wasn't as active as I had hoped for, cf. Fortran 2003 in 1.6

Normative References in the OpenMP 4.0 RC, available at

http://openmp.org/wp/openmp-specifications/  At a glance, it looks as if almost

none of the Fortran 2003 features have been taken care of.


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-13 Thread valeryweber at hotmail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



--- Comment #6 from Valery Weber valeryweber at hotmail dot com 2012-11-13 
16:57:28 UTC ---

Dear All 

I posted a comment on the openmp forum about the f2003 features. Complaining

there may help, who knows?

Valery


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-12 Thread janus at gcc dot gnu.org

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282

janus at gcc dot gnu.org changed:

   What|Removed |Added

 CC||janus at gcc dot gnu.org
Summary|openmp directive and|[OOP] openmp directive and
   |classes |classes

--- Comment #1 from janus at gcc dot gnu.org 2012-11-12 14:35:42 UTC ---
(In reply to comment #0)
 The following code doesnt compile at all with the lattest gfortran. 
 The problem seems to be in the class definition of the variable this 
 (works fine while declared as type).

 [...]
 
 bug.f90: In function ‘prog’:
 bug.f90:21:0: error: ‘__vtab_mod_My_t’ not specified in enclosing parallel

This is a known problem, see PR 52531.

Note that Fortran 2003 is not supported in OpenMP 3.1. This may change with
OpenMP 4, but I'm not sure of that.


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-12 Thread janus at gcc dot gnu.org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



--- Comment #2 from janus at gcc dot gnu.org 2012-11-12 14:47:11 UTC ---

(In reply to comment #1)

 Note that Fortran 2003 is not supported in OpenMP 3.1. This may change with

 OpenMP 4, but I'm not sure of that.



Just checked: The public RC at http://openmp.org/wp/openmp-specifications/ says

the following:



This OpenMP API specification refers to ISO/IEC 1539-1:2004 as Fortran 2003.

The

following features are not supported:

 * ...

 * Polymorphic entities

 * ...



So, it looks like CLASS declarations are still not allowed in OpenMP 4. Too

bad!


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-12 Thread valeryweber at hotmail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



--- Comment #3 from Valery Weber valeryweber at hotmail dot com 2012-11-12 
19:18:34 UTC ---

Thanks pointing that. Is there any reason for not allowing the classes in

openmp?

I noticed that other compilers (eg ifort, xlf) can accommodate with this

deviation from the standard, is gfortran going in the same direction?


[Bug fortran/55282] [OOP] openmp directive and classes

2012-11-12 Thread kargl at gcc dot gnu.org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55282



kargl at gcc dot gnu.org changed:



   What|Removed |Added



 CC||kargl at gcc dot gnu.org



--- Comment #4 from kargl at gcc dot gnu.org 2012-11-12 20:02:02 UTC ---

(In reply to comment #3)

 Thanks pointing that. Is there any reason for not allowing the classes in

 openmp?

 I noticed that other compilers (eg ifort, xlf) can accommodate with this

 deviation from the standard, is gfortran going in the same direction?



Of course there is a good reason.  What if a future OpenMP

standard introduces classes in manner that conflicts with

the way gfortran implements the extension?  gfortran would

then need an option to toggle between the standard conforming

code and the GNU Fortran extension.  Which, then, is the

default?