Re: [Libmesh-devel] ex4 Parallel run error.

[Roy S. <roy...@ic...>]

On Thu, 19 Nov 2009, Roy Stogner wrote:

>
> On Thu, 19 Nov 2009, Vijay S. Mahadevan wrote:
>
>> I just updated, clobbered and ran make on my head svn version of
>> libmesh. Just to test if the library is working correctly, ran ex4 in
>> both 1 and 2 processors. In serial, the code ran as intended but on
>> two processors, I get segfaults.
>
> I can reproduce this; I'm running it through the debugger now...
>
>> I wonder if disabling libHilbert is causing this ? Any help on this
>> would be great because this is the simplest test case I have.

Yeah, it looks like disabling libHilbert is the problem.  I didn't
realize that we were dependent on it now; I thought it was only being
used for getting a partitioning-independent ordering in output.

But it looks like we need a non-NULL fallback implementation for
MeshCommunication::find_global_indices().  A simple "index them in the
order they're already in" should work for SerialMesh, at least... and
I think for the partitioner.C usage it should work for ParallelMesh as
well, since that's operating on unpartitioned elements.

I'll give that a try and see.
---
Roy