On Thu, 3 Apr 2008, Lorenzo Botti wrote:

> The problem is that there are no subactive elem and elem->node(n) is
> never == Node::invalid_id. There could be different nodes with the
> same id?
> The right answer is no, is it?

Correct, but if you want to throw an assert() in there to double-check
it couldn't hurt.

It is possible, if you're using ParallelMesh, to on occasion have node
ids which aren't in use by any node, so that the maximum id exceeds
the number of nodes, but the reverse should never happen, and I
believe neither should happen with SerialMesh.

> Maybe we have some active elem that should be ancestor?
> Do you have any other idea?

Not at the moment.  As John said, can you send us a test case?

> Because of this in System::write_serialized_blocked_dof_objects(...)
> written_lenght match with not_subactive_node_ids, not with vec_lenght
> (vec_lenght == n_nodes()) and so the written solution is wrong.

Hmm... Can you take a mesh on which this fails, rotate it fractionally
to avoid any two nodes sharing the same x value, project "u = x" on
it, and write it out?  That would make it easy to see any nodal values
that were being skipped or duplicated in the I/O.
---
Roy

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to