sieve-dev IndicesVisitor

[Matthew Knepley]

This is a bug. I pushed the fix.

  Thanks,

     Matt

On Mon, Sep 15, 2008 at 11:19 PM, Shi Jin <jinzishuai at gmail.com> wrote:
> Hi there,
>
> I have the following loop:
> typedef
> ALE::ISieveVisitor::IndicesVisitor<PETSC_MESH_TYPE::real_section_type ,
>     PETSC_MESH_TYPE::order_type,PetscInt> visitor_type;
>
>     int numIndices;
>     int e=0;
>     for (PETSC_MESH_TYPE::label_sequence::iterator c_iter = cells->begin();
> c_iter
>             !=cells->end(); ++c_iter, e++)//Loop thru local elements
>     {
>         visitor_type velVisitor(*vel[0], *velGlobalOrder, (int) (pow(
>                 sieve->getMaxConeSize(), m->depth())+1)*m->getMaxDof(),
>                 m->depth() > 1);
>         visitor_type preVisitor(*p, *preGlobalOrder, (int) (pow(
>                 sieve->getMaxConeSize(), m->depth())+1)*m->getMaxDof(),
>                 m->depth() > 1);
>
>         computeIndicesForCell(e, vel[0], &numIndices, velVisitor);
>         for (int i=0; i<numIndices; i++)
>             velCellIndices[e*numIndices+i]=elementIndices[i];
>
>         computeIndicesForCell(e, p, &numIndices, preVisitor);
>         for (int i=0; i<numIndices; i++)
>             preCellIndices[e*numIndices+i]=elementIndices[i];
>     }
>
> Where the function computeIndicesForCell() calls the visitor by reference.
> I want to take the definitions of the two visitors out of loop since I think
> they can be reused. However, if I do that, an exception will be thrown at
> void visitPoint(const point_type& point, const int orientation) {
>         if (p >= size) {
>           ostringstream msg;
>           msg << "Too many points (>" << size << ")for IndicesVisitor
> visitor";
>           throw ALE::Exception(msg.str().c_str());
>         }
>         if (points) {
>           int pp;
>           for(pp = 0; pp < p; ++pp) {if (points[pp] == point) break;}
>           if (pp != p) return;
>           points[p++] = point;
>         }
> ...
> }
> due to too large p value.
> I wonder if this is normal.  The value of p increments in each call to
> visitPoint() and it does not get reset in the clear() call as i does.
>
> Should I just keep the definitions inside the loop? Thanks a lot.
> --
> Sincerely,
> Shi Jin, Ph.D.
> http://www.ualberta.ca/~sjin1/
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which
their experiments lead.
-- Norbert Wiener