[MOAB-dev] Changing adjacency behavior

[Jason Kraftcheck]

On 10/21/2010 05:37 PM, James Porter wrote:
> Last month, the ITAPS list discussed[1] how adjacency should work,
> specifically whether an entity is adjacent to itself. Everyone seemed to
> agree that an entity is not adjacent to itself and that a call to
> iMesh_getEntAdj where the source dimension is the same as the target
> dimension should return an empty list.
>
> Currently, MOAB returns an error for vertex-vertex adjacency, and the
> entity itself for higher dimensions. I have a patch to change the
> behavior to the expected ITAPS way, but I wanted to make sure no one
> needed the old way before I got rid of it. If someone does have a use
> for this, let me know and I'll make it optional.
>

The inconsistent behavior clearly needs to be fixed.  I don't think we 
need to worry about users who expected the vertex case to fail. 
Hopefully there aren't too many relying on the other case(s) succeeding. 
  But the group seems adamant on choosing theoretically correct over 
practically useful, so if we're going to conform to the interface we 
might as well make the change.  If the change does result in lots of 
problems perhaps we can use those use cases as examples of why the 
current behavior (less the vertex inconsistency) is more practical.

- jason