dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Tue, Oct 11, 2005 at 01:16:44PM +0200, dolfin@xxxxxxxxxx wrote:

> Commit from johanjan (2005-10-11 13:16 CEST)
> --------------------
> 
> PyDOLFIN:
> 
> ODE::f() now works in Python (both mono and multi variants), simple
> test produces reasonable result.

Very good!

Now that the Python interface is becoming more useful, maybe we should
promote it with a better position in the source tree?

I suggest we put it in one of

  src/pydolfin
  src/swig

instead of
 
  dolfin  src/demo/scripting/pydolfin/

It should also be installed in the usual place where Python packages
get installed during 'make install'. The test examples could still be
under src/demo. This could wait until after the next release.

Another question we've discussed previously: should we keep the
current system where 'make' does a fake install in $toplevel or should
it be necessary to to a 'make install' before doing 'make demo'?

The advantage would be that we could simplify the build system: Faheem
Mitha reported some problems with the fake install not being cleaned
during a 'make clean', and we could remove all the

    '../../../dolfin-config --cflags'

stuff in the demo Makefiles. The disadvantage would be that it would
introduce one extra step to run the demos.

/Anders