[Gmsh] progress with CGNS boundary data

Nigel Nunn nnnunnn at gmail.com
Thu Aug 5 19:17:55 CEST 2010


Dear Professor Geuzaine,

Thank you for your reply, and many thanks also to Stepen Guzik.

Stepen's code in MZone and MZoneBoundary is the most spectacular use of
standard containers I have ever seen.  For interest, I built a cgns lib
(VC++ 2008), and after a simple change in MZoneBoundary (replacing embedded
FaceT with a pointer to FaceT), writeCGNS(...) runs and writes without
fuss.  That embedded FaceT object caused undefined behaviour in the custom
memory management of CCon::FaceVector<T>.  Only problem is I have no
software to *read* cgns, so cannot confirm or deny correctness.

However, by reusing MZone and MZoneBoundary to assemble the mesh
topology, I've added GModel::writeNEU(...) to export in gambit neutral
format (*.neu).  This gives the face-boundary data required by Tim
Warburton's family of multi-GPU DG solvers.  As a first test, I have these
solvers communicating with the gmsh gui via sockets (as suggested by the
docs).  Run-time rendering of solution fields is working fine.  Also, added
a button to Pause/Resume the client process.

Nigel


On Wed, Aug 4, 2010 at 7:11 AM, Christophe Geuzaine <cgeuzaine at ulg.ac.be>wrote:


> Hi Nigel - I've always been intrigued by CGNS, but never had a chance to
> look it up in detail. The CGNS code in Gmsh was contributed by Stepen Guzik
> (in CC). His code is quite complex: he's probably the best person to
> contact.
>
>
> On 24/07/10 21:17, Nigel Nunn wrote:
>


>  Hi to gmsh developers,
>>
>> The svn code for writeCGNS looks great.  I’m stepping through it trying to
>> get the region-surface boundary data our solver needs.
>>
>> I've found a few places where unexpected things happen.  Is there someone
>> among the active developers currently interested in sorting this out, or
>> simply interested in getting face boundary data?
>>
>> To export such data directly from Netgen, we had to call
>> mesh.UpdateTopology() which builds all the necessary data structures. It
>> looks like something similar is being done in GModelIO_CGNS.cpp by zoneTask?
>>
>> If anyone is interested in discussing this, please let me know.  I think I
>> failed to subscribe to the mailing list, so might be best to also email me
>> at nnnunnn at gmail.com
>>
>> thanks for all the great stuff in gmsh!
>> Nigel
>>
>>
>
> --
> Prof. Christophe Geuzaine
> University of Liege, Electrical Engineering and Computer Science
> http://www.montefiore.ulg.ac.be/~geuzaine
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20100806/f9be7f7d/attachment.html>