dolfin team mailing list archive

[Ilmar Wilbers <ilmarw@xxxxxxxxx>]

Anders Logg wrote:
> On Mon, Mar 31, 2008 at 02:56:47PM +0200, Ilmar Wilbers wrote:
>   
> > Ilmar Wilbers wrote:
> >     
> > > demo/mesh/meshfunction/python
> > > demo/mesh/partitioning/python
> > > demo/mesh/refinement/python
> > > demo/pde/dg/python
> > > demo/pde/mixed-poisson/python
> > > The plot is only displayed a few milliseconds (Viper problem).
> > >
> > >   
> > >       
> > It would be better to replace
> >
> > plot(sigma)
> > plot(u)
> > interactive()
> >
> > with
> >
> > plot(sigma, interactive=True)
> > plot(u, interactive=True)
> >     
>
> I thought it would be nice to be able to see both plots at the same
> time? (Which is possible if you just move one of the windows.)
>
>   
Sure, I agree. It is just not clear that there is a second window below 
the first, so when done with the top one, and pressing 'q', both 
dissapear. But nevermind, this is not very important.
> > in demo/pde/mixed-poisson/python, else the two plots will be on top of 
> > each other and both will be closed upon pressing 'q'. The rest it fine.
> >     
> > > demo/mesh/subdomains/cpp
> > > demo/mesh/subdomains/python
> > > Maybe give some output from running the program? Like: 'generated file 
> > > subdomains.xml'?
> > >
> > >   
> > >       
> > No changes.
> >     
>
> I added a line here to increase the debug level. Don't you get any
> output?
>
>   
Nope, nothing...
> > > demo/ode
> > > Demos are not compiled, do not follow the directory structure of the 
> > > other demos, and don't work.
> > >
> > >   
> > >       
> > No changes.
> >     
>
> ok, try again.
>
> The Python interface of the ODE solver is broken. It shouldn't be much
> work to fix it, but it hasn't been done.
>
>   
What am I looking for here, except for the aliev-panfilov directory 
(cpp/demo works), all directories only contain a main.cpp and plot.py. 
Nothing is built.
> > > demo/pde/convection-diffusion/python
> > > ...
> > > Applying boundary conditions to linear system.
> > > Traceback (most recent call last):
> > >    File "demo.py", line 65, in <module>
> > >      solve(A, u1.vector(), b)
> > >    File "/usr/local/lib/python2.5/site-packages/dolfin/dolfin.py", line 
> > > 3683, in solve
> > >      return _dolfin.solve(*args)
> > > TypeError: in method 'solve', argument 2 of type 'dolfin::Vector &'
> > >
> > >   
> > >       
> > Seems fine. But should add interactive=True to plot.
> >     
>
> There seems to be some more trouble with this demo. The solution
> remains constant (it should change with time). I couldn't figure out
> what goes wrong.
>
>   
> > > demo/pde/elasticity/cpp
> > > The plot looks completely weird! Python version gives the correct plot.
> > >
> > >   
> > >       
> > Still broken.
> >     
>
> Here too.
>
>   
> > > demo/pde/mixed-poisson/python
> > > demo/pde/stokes/stabilized/python
> > > ...
> > > Ordering mesh entities...
> > > swig/python detected a memory leak of type 'SubFunction *', no 
> > > destructor found.
> > >
> > >   
> > >       
> > Still there, but that is OK, perhaps?
> >     
>
> It's been there for a while and I don't know how to fix it.
>
>   
> > > demo/pde/periodic/python
> > > ...
> > >    Applying boundary conditions to linear system.
> > > Traceback (most recent call last):
> > >    File "demo.py", line 68, in <module>
> > >      u = pde.solve()
> > >    File "/usr/local/lib/python2.5/site-packages/dolfin/assemble.py", 
> > > line 193, in solve
> > >      cpp_DirichletBC.apply(bc, A, b, self.dof_maps.sub(1), compiled_form)
> > > TypeError: unbound method apply() must be called with cpp_DirichletBC 
> > > instance as first argument (got PeriodicBC instance instead)
> > >
> > >   
> > >       
> > No changes.
> >     
>
> Haven't looked at this yet.
>
>   
> > > demo/pde/stokes/stabilized/python
> > > demo/pde/stokes/taylor-hood/python
> > > Only the second of two plot is displayed.
> > >
> > >   
> > >       
> > The same goes for these two,
> >
> > plot(U)
> > plot(P)
> > interactive()
> >
> > should be replaced by
> >
> > # Plot solution
> > plot(U, interactive=True)
> > plot(P, interactive=True)
> >     
>
> Yes, if that's what we want (see above).
>
>   
> > > demo/quadrature/python
> > > ilmarw@*****:python$ python demo.py lobatto 10
> > > Traceback (most recent call last):
> > >    File "demo.py", line 43, in <module>
> > >      sys.exit(main(sys.argv[1:]))
> > >    File "demo.py", line 33, in main
> > >      q = LobattoQuadrature(n);
> > > NameError: global name 'LobattoQuadrature' is not defined
> > >
> > >   
> > >       
> > Fixed...
> >     
>
> Yes.
>
>   
> > > In addition, there are a few 'fix me' demos.
> > >       
>
> Yes. Note that not all Python demos have worked before. The state the
> demos are in now is close to the best they have ever been in. So it's
> not the new build system, it's just that the Python interface and
> demos are work in progress.
> dd
>   
ilmar