libMesh: A C++ Finite Element Library

On Mon, 11 Apr 2011, Derek Gaston wrote:

> Just tried to make a copy of a ParallelMesh... and got a segfault in
> UnstructuredMesh::copy_nodes_and_elements().  I'm going to go ahead
> and dive in...  but I thought I would see if you had any thoughts on
> whether or not I should expect the copy constructor for ParallelMesh
> to work... or if there is some philosophical reason why it won't
> ever work.

No philosophical reason, but probably some oversight: it's easy to
write mesh code that assumes no sparsity and then get bitten by that
assumption later.

Glancing at the code I don't see an obvious problem, though...
---
Roy

Ok - I got it sorted.  The call to add_point() was just allowing the mesh to assign the node an id... but that's not exactly what we want.... we actually want it to have the same node id that node has in the copied mesh.

So I just used this call to add_point instead:

        Node *newn = this->add_point(*oldn, oldn->id(), oldn->processor_id());

That fixed it and my code runs fine now.

Derek

On Apr 11, 2011, at 10:13 AM, Roy Stogner wrote:

> 
> On Mon, 11 Apr 2011, Derek Gaston wrote:
> 
>> Just tried to make a copy of a ParallelMesh... and got a segfault in
>> UnstructuredMesh::copy_nodes_and_elements().  I'm going to go ahead
>> and dive in...  but I thought I would see if you had any thoughts on
>> whether or not I should expect the copy constructor for ParallelMesh
>> to work... or if there is some philosophical reason why it won't
>> ever work.
> 
> No philosophical reason, but probably some oversight: it's easy to
> write mesh code that assumes no sparsity and then get bitten by that
> assumption later.
> 
> Glancing at the code I don't see an obvious problem, though...
> ---
> Roy
> 
> ------------------------------------------------------------------------------
> Xperia(TM) PLAY
> It's a major breakthrough. An authentic gaming
> smartphone on the nation's most reliable network.
> And it wants your games.
> http://p.sf.net/sfu/verizon-sfdev
> _______________________________________________
> Libmesh-devel mailing list
> Libmesh-devel@...
> https://lists.sourceforge.net/lists/listinfo/libmesh-devel

On Mon, 11 Apr 2011, Derek Gaston wrote:

> Ok - I got it sorted.  The call to add_point() was just allowing the
> mesh to assign the node an id... but that's not exactly what we
> want.... we actually want it to have the same node id that node has
> in the copied mesh.

That would do it, yeah.

> So I just used this call to add_point instead:
>
>        Node *newn = this->add_point(*oldn, oldn->id(), oldn->processor_id());
>
> That fixed it and my code runs fine now.

Looks good, thanks - put it in svn?
---
Roy

On Apr 11, 2011, at 10:53 AM, Roy Stogner wrote:

>> So I just used this call to add_point instead:
>> 
>>       Node *newn = this->add_point(*oldn, oldn->id(), oldn->processor_id());
>> 
>> That fixed it and my code runs fine now.
> 
> Looks good, thanks - put it in svn?

Will do... but I'm going to do a bit more testing first.... I'm still seeing a segfault in parallel.... more to do apparently.

Derek