[Gmsh] Gmsh doesn't run

Martin Vymazal martin.vymazal at vki.ac.be
Tue Dec 22 23:50:49 CET 2009


Hello Matti, hello Christophe,

  I can run gmsh now. Thank you for the fix.

  However, gmsh is still unusable for me as I don't see any mesh after  
meshing (doesn't matter if it's 2D or 3D). The behaviour is following:

Tools->Options->Geometry->Visibility->[Points | Lines | Surfaces |  
Volumes] - everything works fine.

Then I mesh and I still see my geometry, but no mesh.

Tools->Options->Mesh->Visibility-> Nodes ... works fine
                                 -> Lines ... I see some more nodes,  
but no lines
                                 -> Surface edges ... even more nodes, no lines
                                 -> Volume edges ... has no effect
                                 -> Volume faces ... has no effect

I also tried to mesh in command line (gmsh -3 myfile.geo) and open  
myfile.msh with gmsh. All I get is an empty window.

Playing with mesh visibility options, I see a lot of error output in  
the terminal from which I run gmsh. The messages usually say something  
like:

XRequest.135: BadLength (poly request too large or internal Xlib  
length error) 0x2
XRequest.135: GLXBadLargeRequest 0x2
XRequest.135: BadLength (poly request too large or internal Xlib  
length error) 0x3800013
XRequest.135: GLXBadLargeRequest 0x2
XRequest.135: BadLength (poly request too large or internal Xlib  
length error) 0x2
XRequest.135: GLXBadLargeRequest 0x2
XRequest.135: BadLength (poly request too large or internal Xlib  
length error) 0x2


In the end, the only thing I have left from the mesh is a bunch of  
nodes. I have this problem in my laptop with Nvidia Quadro FX 570M and  
binary nvidia drivers version 190.42-1. I have no problem with gmsh  
and intel graphics card (different computer, the same linux  
distribution, identical system settings). Therefore I suspect this to  
be a graphics-related issue.

Does anyone have a similar problem?


Best regards,

    Martin Vymazal



Quoting Matti Pellikka <matti.pellikka at tut.fi>:

> 2009/12/17 Christophe Geuzaine <cgeuzaine at ulg.ac.be>
>
>>
>> Indeed - we recently changed the default config options to include Matti's
>> homogy work, which depends on libgmp. It seems that, since we link with a 32
>> bit shared libgmp lib, the resulting binary will not work when trying to
>> load a 64 bit version instead...
>>
>> Matti - do you know if it would be easy to link with a static version of
>> libgmp?
>>
>>
> libgmp should be now statically linked in the newest nightly builds. Please
> observe whether the problem persists.
>
>
>> Christophe
>>
>>
>>
>> -------- Original Message --------
>> Subject: Re: [Gmsh] Gmsh doesn't run
>> Date: Thu, 17 Dec 2009 12:47:36 +0100
>> From: Umut Tabak <u.tabak at tudelft.nl>
>> To: Martin Vymazal <martin.vymazal at vki.ac.be>
>> CC: Gmsh mai. lis. <gmsh at geuz.org>
>>
>> Martin Vymazal wrote:
>>
>>>
>>> So I gave a try to the nightly build of gmsh.
>>> This is what happens in both my laptop and work machine (which are
>>> both 64 bit):
>>>
>>> ./gmsh: error while loading shared libraries: libgmp.so.3: wrong ELF
>>> class: ELFCLASS64
>>>
>> Yesterday, from the nightly build I got the same error from a similarly
>> configured machine(64 bit laptop), but did not have the time to continue
>> digging. And turned to the stable release. Indeed I am not the only one
>> so this deserves further inspection.
>> Best regards,
>> Umut
>>
>> _______________________________________________
>> gmsh mailing list
>> gmsh at geuz.org
>> http://www.geuz.org/mailman/listinfo/gmsh
>>
>>
>>
> --
> Matti Pellikka
>



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.