dolfin team mailing list archive

[Johan Hake <johan.hake@xxxxxxxxx>]

On Thursday June 3 2010 09:48:20 Anders Logg wrote:
> On Thu, Jun 03, 2010 at 09:33:08AM -0700, Johan Hake wrote:
> > On Thursday June 3 2010 09:25:46 Anders Logg wrote:
> > > On Thu, Jun 03, 2010 at 08:37:55AM -0700, Johan Hake wrote:
> > > > On Thursday June 3 2010 08:09:03 Anders Logg wrote:
> > > > > On Thu, Jun 03, 2010 at 07:59:30AM -0700, Johan Hake wrote:
> > > > > > On Thursday June 3 2010 04:28:19 Anders Logg wrote:
> > > > > > > On Thu, Jun 03, 2010 at 01:24:30PM +0200, Marie Rognes wrote:
> > > > > > > > On 03. juni 2010 13:22, Anders Logg wrote:
> > > > > > > > > On Thu, Jun 03, 2010 at 01:16:27PM +0200, Marie Rognes 
wrote:
> > > > > > > > >> What happened to the member variable 'domain' in
> > > > > > > > >> DirichletBC in bcs.py ?
> > > > > > > > > 
> > > > > > > > > It was replaced by 'domain_args' to handle MeshFunctions in
> > > > > > > > > homogenize().
> > > > > > > > 
> > > > > > > > So, domain == domain_args[0] ?
> > > > > > > 
> > > > > > > Yes, if len(domain_args) == 1. Otherwise, it will be of length
> > > > > > > 2 and contain the MeshFunction and the domain number.
> > > > > > 
> > > > > > I think I have said this before...
> > > > > > 
> > > > > > Where is the docstring?
> > > > > > 
> > > > > > :)
> > > > > > 
> > > > > > Johan
> > > > > 
> > > > > I'm not sure this should be documented in the docstring since this
> > > > > is really an internal trick of the Python DirichletBC class. So
> > > > > the problem might really be the variable name (missing
> > > > > underscore).
> > > > 
> > > > Well, something is better than nothing. Now it looks like:
> > > > 
> > > > class DirichletBC(cpp.DirichletBC):
> > > >     def __init__(self, *args):
> > > >         "Create Dirichlet boundary condition."
> > > > 
> > > > Which is really nothing. A user need to look in a demo to understand
> > > > how to use it. Especially with the magical AutoSubDomain feature...
> > > 
> > > Aha, now I see your point. Yes, there should definitely be a better
> > > docstring, but I think we decided that docstring should end up in the
> > > online manual and then be "included" in the Python code so it might
> > > not be worth the effort to write a very detailed docstring right now.
> > 
> > He, he, always an excuse!
> 
> I'm saving my strength for a possible night of pain if/when we decide
> to change from a(v, u) to a(u, v)... :-)

Urk...

Yes I noticed that it might go in that direction. As my weight is not in FEM, 
I have kept quite about this. But if you ask me, I vote to keep it as it is. 
:) Together with the notion that Hans Petter had in the thread where this was 
discussed.

Johan