dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Mon, Mar 17, 2008 at 07:28:37PM +0100, Anders Logg wrote:
> On Mon, Mar 17, 2008 at 07:16:20PM +0100, Anders Logg wrote:
> > On Mon, Mar 17, 2008 at 06:07:05PM +0000, Garth N. Wells wrote:
> > > 
> > > 
> > > Garth N. Wells wrote:
> > > > 
> > > > Anders Logg wrote:
> > > >> On Mon, Mar 17, 2008 at 05:18:03PM +0000, Garth N. Wells wrote:
> > > >>> I'm having a tough time getting the demos to work again :(.
> > > >>>
> > > >>> Disabling PETSc, I can plot using pyDOLFIN, but running a cpp demo (like 
> > > >>> poisson) I get
> > > >>>
> > > >>> *** glibc detected *** python: free(): invalid next size (fast): 
> > > >>> 0x084667f0 ***
> > > >>>
> > > >>> when the demo tries to plot. Any ideas?
> > > >> Same here:
> > > >>
> > > >> *** glibc detected *** python: free(): invalid next size (fast):
> > > >> 0x084fdbf8 ***
> > > >>
> > > >> If you write
> > > >>
> > > >>   ls -lrt /tmp/*.py
> > > >>
> > > >> you'll see the Python script that is run when calling plot() from C++.
> > > >>
> > > >> Here's what it looks like:
> > > >>
> > > >>   from dolfin import *
> > > >>
> > > >>   object = Function("/tmp/filepjlanT.xml")
> > > >>   plot(object)
> > > >>
> > > >> Running just this script reproduces the error one gets when running
> > > >> the Poisson demo. From the trace it looks like something goes wrong in
> > > >> DiscreteFunction::interpolate() when DOLFIN is compiled with PETSc.
> > > >>
> > > > 
> > > > I'm getting the error with PETSc disabled.
> > > > 
> > > > When I enable PETSc, I get a message that plotting is not configured.
> > > > 
> > > >> I looked at it briefly but couldn't figure it out.
> > > >>
> > > >
> > > 
> > > Looks like the problem was introduced by
> > > 
> > >    http://www.fenics.org/hg/dolfin?cs=2576b428fd0e
> > > 
> > > I haven't had a look at the details.
> > > 
> > > Garth
> > 
> > My fault then. The problem is definitely in
> > DiscreteFunction::interpolate() to get the vertex values.
> > 
> > I'll see if I can find it.
> 
> Found it, fix on its way...

ok, try now.

-- 
Anders