[petsc-dev] DMSetDefaultSection destroys old sections

[Jed Brown]

On Thu, Aug 30, 2012 at 5:06 PM, Chris Eldred <chris.eldred at gmail.com>wrote:

> The code for DMSetDefaultSection is:
>
> 3016:   PetscSectionDestroy(&dm->defaultSection);
> 3017:   PetscSectionDestroy(&dm->defaultGlobalSection);
> 3018:   dm->defaultSection = section;
> 3019:   return(0);
>

This code is wrong. It should likely have used
PetscObjectReference((PetscObject)section) before these lines.


>
> which destroys the "current" default section before setting the new
> one. Since my sections are passed by reference in Fortran, the only
> recourse I can see is to create a copy of the section before it is
> passed to DMSetDefaultSection. Any other ideas?
>
> On Thu, Aug 30, 2012 at 4:03 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> > I think the bug is not that the old section is destroyed, but that the
> new
> > one is not referenced.
> >
> >
> > On Thu, Aug 30, 2012 at 5:00 PM, Chris Eldred <chris.eldred at gmail.com>
> > wrote:
> >>
> >> I am using DMSetDefaultSection and DMCreateLocalVector to create
> >> vectors in my code, but errors are occurring because
> >> DMSetDefaultSection destroys old sections. My usage pattern is
> >>
> >> for each variable
> >> call DMSetDefaultSection
> >> call DMCreateLocalVector
> >>
> >> which fails if there is more than one 1 variable that uses the same
> >> section since the sections are being destroyed when a new one is set.
> >> Is there a reason for this behavior in DMSetDefaultSection (ie can a
> >> version of DMSetDefaultSection that does not destroy the old sections
> >> be provided)?
> >>
> >> --
> >> Chris Eldred
> >> DOE Computational Science Graduate Fellow
> >> Graduate Student, Atmospheric Science, Colorado State University
> >> B.S. Applied Computational Physics, Carnegie Mellon University, 2009
> >> chris.eldred at gmail.com
> >
> >
>
>
>
> --
> Chris Eldred
> DOE Computational Science Graduate Fellow
> Graduate Student, Atmospheric Science, Colorado State University
> B.S. Applied Computational Physics, Carnegie Mellon University, 2009
> chris.eldred at gmail.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120830/a6af02a7/attachment.html>