dolfin team mailing list archive

[Johan Hake <hake@xxxxxxxxx>]

On Tuesday 22 January 2008 18:43:30 Anders Logg wrote:
> On Tue, Jan 22, 2008 at 03:15:02PM +0100, Ola Skavhaug wrote:
> > Anders Logg skrev den 22/01-2008 følgende:
> > > On Tue, Jan 22, 2008 at 03:04:44PM +0100, Martin Sandve Alnæs wrote:
> > > > I think this is definitely a "bug". (It's not a function, but a
> > > > built-in type.)
> > >
> > > Agree. Suggestions?
> >
> > %rename(dolfin_set) dolfin::set;
> >
> > ??
> >
> > Ola
>
> If possible, it would be good to have the same function names in
> Python and C++ (or I will be confused).
>
> Here are two options:
>
> A. Rename all free functions in DOLFIN to dolfin_foo. This will
> include
>
>   dolfin_message
>   dolfin_error
>   dolfin_debug (already called this today for technical reasons)
>
>   dolfin_set
>   dolfin_get
>
>   dolfin_init

I think that these will be too cumbersome to use. 

> We would not touch the free functions assemble() and solve() which are
> prominent enough not to be prefixed by "dolfin_"
>
> B. Don't prefix the functions and find a better name for
> dolfin::set(). For example, we could have
>
>   setparm()
>   getparm()

I vote for these. 

I notice that pylab use setp and getp, so these are not any options.


Is it an option to not include message, error, debug when you 

from dolfin import *

and for example include them as 

from dolfin.debug import [debug, message, error] or something?

I do not know if this is possible the way pydolfin is built now, i.e. all 
functions in one module, generated from one swig interface file. 

Johan