Should PetscObjectName be collective?

Satish Balay balay at mcs.anl.gov
Wed Nov 18 09:35:47 CST 2009


On Wed, 18 Nov 2009, Barry Smith wrote:

> 
> On Nov 18, 2009, at 8:52 AM, Jed Brown wrote:
> 
> > http://petsc.cs.iit.edu/petsc/petsc-dev/rev/ecbd2dcad71a
> > 
> > Note that there are a couple other places where keyvals are created (with no
> > way to free them).  Does it make sense to put those fields into this one?
> > 
> > Is there a reason that petscinitialize_() duplicates much of
> > PetscInitialize()?  Seems like we ought to have the common stuff in a
> > PetscInitialize_Private().
> 
>   There was a reason for this duplication many years ago. I do not know if it
> is still needed. There is the business of the fortran "initialization" being
> done twice which may be fragile. You could try merging but test it
> relentlessly :-)

My guess is that some operations had to be done in a different order
from fortran [due the missing argc/argv]..

Satish




More information about the petsc-dev mailing list