Dear all,
sorry for flooding this list, but I discovered the following:
This results in a segfault:
boundary_parts=MeshFunction("size_t", mesh, "beam_facet_region.xml")
But this works flawlessly:
boundary_parts=MeshFunction("size_t", mesh, "./beam_facet_region.xml")
Is there any explanation for this (to my mind) very strange behaviour?
Apparently,
mesh = Mesh("beam.xml")
or
mesh = Mesh("./beam.xml")
makes no difference and neither segfaults?!
Regards,
Stephan Schmidt
On 20 Sep 2013, at 00:59, Schmidt, Stephan wrote:
Dear all,
I have a peculiar problem with FEniCS on OSX installed via dorsal
(stable=false, snapshots=false). My "print dolfin.__version__" shows 1.2.0+
The following code works perfect:
from dolfin import *
mesh = Mesh("beam.xml")
plot(mesh, interactive=True)
However, when opening the corresponding MeshFunction, the following crashes
with a segfault
from dolfin import *
mesh = Mesh("beam.xml")
boundary_parts=MeshFunction('size_t', mesh, "beam_facet_region.xml")
plot(mesh, interactive=True)
the segfault is
Python(8825,0x7fff74232960) malloc: *** error for object 0x7fff7318d860:
pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
[noname:08825] *** Process received signal ***
[noname:08825] Signal: Abort trap: 6 (6)
[noname:08825] Signal code: (0)
[noname:08825] *** End of error message ***
Abort trap: 6
Any ideas why this could be happening? I suppose the i/o is working fine,
because it can open the xml mesh file. However, it apparently cannot open the
facet region file?! Both files were automatically generated using
dolfin-convert from a gmsh beam.msh file. I will send the files in a separate
mail due to size restrictions, they are 300kb.
Any hints would be greatly appreciated.
Regards,
Stephan Schmidt
_______________________________________________
fenics-support mailing list
[email protected]<mailto:[email protected]>
http://fenicsproject.org/mailman/listinfo/fenics-support
_______________________________________________
fenics-support mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics-support
