[petsc-dev] checking if a vector is ghosted

[mspieg]

Thanks everyone for getting back so fast (as usual).

VecIsGhosted would be nice…

I assume that would go into dev…

In the meantime, I think the error handler trick will work in 3.2…

thanks
marc

On Nov 8, 2011, at 3:41 PM, Barry Smith wrote:

> 
>  We should add a VecIsGhosted()
> 
>   Barry
> 
> On Nov 8, 2011, at 12:07 PM, Jed Brown wrote:
> 
>> On Tue, Nov 8, 2011 at 14:05, Matthew Knepley <knepley at gmail.com> wrote:
>> Just want to check that the following is a kosher use of the return error handler
>> 
>>   bool _is_ghosted = true;
>>   PetscErrorCode ierr;
>>   PetscPushErrorHandler(PetscReturnErrorHandler,PETSC_NULL);
>>   ierr= VecGhostGetLocalForm(*x, x_ghosted.get());
>>   if (ierr != 0 )
>>     _is_ghosted = false;
>>   PetscPopErrorHandler();
>> 
>> Do you need to mess with the error handler?
>> 
>> Yes, otherwise an error message would be printed.
>> 
>> Mark, what you describe is okay. We, or you, could make a VecIsGhosted(). The code shows how it determines whether a Vec is ghosted.
>> 
>> PetscErrorCode  VecGhostGetLocalForm(Vec g,Vec *l)
>> {
>>  PetscErrorCode ierr;
>>  PetscBool      isseq,ismpi;
>> 
>>  PetscFunctionBegin;
>>  PetscValidHeaderSpecific(g,VEC_CLASSID,1);
>>  PetscValidPointer(l,2);
>> 
>>  ierr = PetscTypeCompare((PetscObject)g,VECSEQ,&isseq);CHKERRQ(ierr);
>>  ierr = PetscTypeCompare((PetscObject)g,VECMPI,&ismpi);CHKERRQ(ierr);
>>  if (ismpi) {
>>    Vec_MPI *v  = (Vec_MPI*)g->data;
>>    if (!v->localrep) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Vector is not ghosted");
>>    *l = v->localrep;
>>  } else if (isseq) {
>>    *l = g;
>>  } else {
>>    SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Vector type %s does not have local representation",((PetscObject)g)->type_name);
>>  }
>>  ierr = VecGhostStateSync_Private(g,*l);CHKERRQ(ierr);
>>  ierr = PetscObjectReference((PetscObject)*l);CHKERRQ(ierr);
>>  PetscFunctionReturn(0);
>> }
> 

----------------------------------------------------
Marc Spiegelman
Lamont-Doherty Earth Observatory
Dept. of Applied Physics/Applied Math
Columbia University
http://www.ldeo.columbia.edu/~mspieg
tel: 845 704 2323 (SkypeIn)
----------------------------------------------------