← Back to team overview

dolfin team mailing list archive

Re: demo report

 

On Sat, Feb 16, 2008 at 05:28:14PM -0500, Gideon Simpson wrote:
> i fixed it by blowing away my existing installation and using newly  
> pulled copies of al the software.  i was able to fix my plotting; it  
> was  vtk problem.  BTW, perhaps it is because I'm using the VTK CVS  
> source, but I do get the following command line output when running  
> something with viper plotting:
> 
> python[57321:613] *** _NSAutoreleaseNoPool(): Object 0x5ba0060 of  
> class vtkCocoaServer autoreleased with no pool in place - just leaking
> Stack: (0x96d6612f 0x96c72ec2 0x5112b14 0x5113510 0x4ddbf45 0x18d806  
> 0x18d9e8 0x18f45b 0x139c27 0x11fd3d 0x18dfb8 0x18f45b 0x139c27  
> 0x11fd3d 0x18dfb8 0x18f45b 0x139c27 0x11fd3d 0x1285f8 0x11fd3d  
> 0x1629f1 0x15cbe6 0x11fd3d 0x18dfb8 0x18f45b 0x18da85 0x18f45b  
> 0x18f548 0x1a69ec 0x1a7016 0x1a8982 0x1b3c03)
> 
> Also, a remark on installation.  I use the Apple python distribution,  
> but like to keep all my personal software libraries in /opt.  However,  
> for some reason dolfin wants to install pydolfin in
> /Library/Python/2.5/site-packages, despite setting --prefix=/opt when  
> configuring.  there's a similar issue with UFC.
> 
> -gideon

Strange, what does UFC tell you when you install it?

    "Installing UFC under..."

The setup.py script is fairly simple so it should be easy to see what
goes wrong with a few print statements.

-- 
Anders


> On Feb 16, 2008, at 3:11 AM, Anders Logg wrote:
> 
> > On Fri, Feb 15, 2008 at 09:28:29PM -0500, Gideon Simpson wrote:
> >> So, up to the extent that the viper plotting seems to be broken, the
> >> poisson demo works in both c++ and python.
> >
> > Good, was it ffc-clean that helped?
> >
> > Why is plotting broken? Has it worked for you before?
> >
> > To debug it, run a demo and then look in /tmp for the latest Python
> > script:
> >
> >  # ls -lrt /tmp/*.py | tail -1
> >  -rw-r--r-- 1 logg logg 128 2008-02-15 18:45 /tmp/fileaD6dZd.py
> >
> > Edit this script and remove the try/except part so you just have
> > the small piece of code that does the plotting. That should tell us
> > more about what goes wrong.
> >
> >> However, neither c++ nor
> >> python elasticity or stokes demos run properly, both generating
> >> segmentation faults after:
> >>
> >> Initializing PETSc (ignoring command-line arguments).
> >
> > Try removing as much stuff as you can (beginning at the end of the
> > script) to have something which is as simple as possible but still
> > breaks. Then post the script to give us a hint of what may go wrong.
> >
> 
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev


References