Gmsh problems

[Christophe Geuzaine]

"Landrø, Stig" wrote:
> 
> > Hello Christophe
> >
>         I have some additional comments concerning my Gmsh 2d meshing
> problems. When I did som tests with the Gmsh Windows NT version 0.93, I
> found that the occurence of memory access violations depended on both the
> direction of rotation of the surface loop, and the grid size that were set
> for the points. If the size at one point was set larger than the distance
> between the points, the grid would be generated OK for one rotational
> direction of the loop. For the other rotational direction of the loop, a
> memory access violation would occur. When I set a grid size smaller that the
> distance between the points, Gmsh would not crash, but the generated mesh
> was different for the two surface loop rotational directions. I have not
> tested this with the latest Linux version, but there could be a similar
> problem. I have not seen crashes when surface meshing with the Linux
> version, but some of the surface meshes have looked strange with some grid
> points far away from the actual surface.
> 

Hello,

I took a look at the files you sent (hold1*.geo). I managed to mesh
hold1.geo by setting a smaller characteristic length (0.08) for all
points, except for the 8 forming the exterior cube. Could you send me
your latest hold1.geo and hold2.geo (I didn't receive this one) with the
volumes defined? 

(You really shouldn't use the Windows version. It isn't distributed
anymore, and my collegue does not plan to release patched versions for
the moment.)

-- 
Christophe Geuzaine

Tel: +32-(0)4-366.37.10    mailto:Christophe.Geuzaine at ulg.ac.be
Fax: +32-(0)4-366.29.10    http://www.geuz.org