#241 Sematics of entity::mightBeCoarsened()
Metadata
Property | Value |
---|---|
Reported by | unknown (unknown) |
Reported at | Jan 16, 2007 14:32 |
Type | Bug Report |
Version | Git (pre2.4) [autotools] |
Operating System | Unspecified / All |
Last edited by | Oliver Sander (oliver.sander@tu-dresden.de) |
Last edited at | Feb 16, 2009 18:04 |
Closed by | Oliver Sander (oliver.sander@tu-dresden.de) |
Closed at | Feb 16, 2009 18:04 |
Closed in version | Unknown |
Resolution | Fixed |
Comment | The semantics question is settled. I will open a new issue for the UG bug. |
Description
When marking an entity with -1, then this entity will be coarsened, when adapt() is called, if coarsening is at all possible. The routine mightBeCoarsened() is responsible for telling a client, if this might actually happen. In order to be a useful method, mightBeCoarsened()==false should imply that the corresponding entity will be present after a grid change (otherwise the information we get is next to meaningless). This is, however, not the case. The problem is, that if an entity is coarsened, then all entities with the same father (that might not be marked) will be coarsened, too, but mightBeCoarsened==false for these (unmarked) entities. Strictly speaking, the documentation does not promise the behaviour I expected, and I don't think that it will be hard for a client to write a workaround by marking the entities for coarsening in a consistent manner, but of course it is difficult to spot a bug that results from an inconsistent marking strategy. So I suggest to either implement the semantics "mightBeCoarsened()==false implies that the entity remains after grid change", or to document more explicitely the weak semantics that is present.