segmentation fault when saving png of Constant

[Chaffra Affouda]

I am getting a seg fault when doing

mesh = UnitCubeMesh(10,10,10)
plotter = plot(Constant(1.0),mesh=mesh)
plotter.write_png('test.png') # breaks here

but if I do

mesh = UnitCubeMesh(10,10,10)
Q = FunctionSpace(mesh,'CG',1)
f = Function(Q)
f.vector()[:] = 1.0
plotter = plot(f)
plotter.write_png('test.png') # works here

I could track down the problem to "_function.lock()->compute_vertex_values(vertex_values, *mesh);" in VTKPlottableGenericFunction.cpp. Apperently _function is a std::weak_ptr and that seems to be the problem. Can anyone reproduce?

[Mattia Tamellini]

Hi,

I also found the snippet posted above to give a segmentation fault. Actually, I get a segmentation fault with both the versions posted (or better, with the C++ equivalent of both the versions). Has this been fixed in the latest commits?

[Martin Sandve Alnæs]

I get this:

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-5-b76fd35ce204> in <module>()
----> 1 plotter.write_png('test.png') # breaks here

RuntimeError: 

*** -------------------------------------------------------------------------
*** DOLFIN encountered an error. If you are not able to resolve this issue
*** using the information listed below, you can ask for help at
***
***     [e-mail]
***
*** Remember to include the error message listed below and, if possible,
*** include a *minimal* running example to reproduce the error.
***
*** -------------------------------------------------------------------------
*** Error:   Unable to complete call to function update().
*** Reason:  Assertion !_function.expired() failed.
*** Where:   This error was encountered inside ../../dolfin/plot/VTKPlottableGenericFunction.cpp (line 232).
*** Process: 0
*** 
*** DOLFIN version: 1.7.0dev
*** Git changeset:  da023d201c0e75fafdb72617bfb0b0effa9b3537
*** -------------------------------------------------------------------------

[Mattia Tamellini]

So I take it the answer to my question is "no" :)

[Prof Garth Wells]

Looks like the version of plot that takes a references (to meshes and expressions) needs to be ignored in the SWIG layer.

Simpler fix would be to just remove it and force use of the shared_ptr version.

[Prof Garth Wells]

Fix Issue #554.

Should really remove the plot functions that take references.

→ <<cset 43fc46a0d5df>>

[Anders Logg (Chalmers)]

Removing milestone: 1.7 (automated comment)