dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Fri, Apr 25, 2008 at 11:13:25PM +0200, Martin Sandve Alnæs wrote:
> 2008/4/25, Anders Logg <logg@xxxxxxxxx>:
> > On Fri, Apr 25, 2008 at 08:11:44PM +0200, Martin Sandve Alnæs wrote:
> >  > Is it necessary to have VectorNormType as a parameter? Since the
> >  > number of different norms is quite small I think it's much easier for
> >  > a user to relate to one function for each. The parameter is just
> >  > another thing to look up in the documentation somehow.
> >
> >
> > I think it should be ok as long as it's easy to look up.
> >
> >  If one may write help(x.norm) and get the appropriate information, I
> >  think it's worth it not having to add normL2, normL1, normInf.
> 
> Ok.
> 
> While on the subject of documentation, be aware that using *args in
> the python interface completely destroys the documentation. Seriously,
> *args is _never_ a good idea. I can _tolerate_ it when the docstring
> is superior (pylab does this pretty well, and is a rare case of when
> it makes sense), but most often it should rather be replaced by a set
> of named keyword arguments with default values "None".

I agree.

-- 
Anders