dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

Anders Logg wrote:
> On Wed, Jul 09, 2008 at 11:53:31AM +0200, Martin Sandve Alnæs wrote:
> > 2008/7/9 Anders Logg <logg@xxxxxxxxx>:
> > > On Tue, Jul 08, 2008 at 03:41:43PM +0200, Martin Sandve Alnæs wrote:
> > > > 2008/7/8 Johan Hake <hake@xxxxxxxxx>:
> > > > > On Tuesday 08 July 2008 15:29:03 Johan Hake wrote:
> > > > > > On Tuesday 08 July 2008 15:06:39 DOLFIN wrote:
> > > > > > > One or more new changesets pushed to the primary dolfin repository.
> > > > > > > A short summary of the last three changesets is included below.
> > > > > > >
> > > > > > > changeset:   4387:b9cdb218b6b993acc959f3aabac11658c92613db
> > > > > > > tag:         tip
> > > > > > > user:        "Martin Sandve Alnæs <martinal@xxxxxxxxx>"
> > > > > > > date:        Tue Jul 08 15:07:17 2008 +0200
> > > > > > > files:       site-packages/dolfin/function.py
> > > > > > > description:
> > > > > > > Bugfix: some demos failed after things were moved around.
> > > > > > You should probably fix bc.py too.
> > > > > I looked at your fix and there are some more fixes that should be done to
> > > > > function.py.
> > > > >
> > > > > Look at the patch from
> > > > >
> > > > > <www.fenics.org/pipermail/dolfin-dev/2008-July/008586.html>
> > > > >
> > > > > Johan
> > > > >
> > > > > > I also probably fixed this in my patches above...
> > > > > >
> > > > > > I think I have to quit sending patches, as it breaks my system when they
> > > > > > are not accepted before someone else fixes the same. :P
> > > > > >
> > > > > > Cheers,
> > > > > >
> > > > > > Johan
> > > > I just fixed those as well... Sorry I didn't look at your patches
> > > > before I did this!
> > > >
> > > > These bugs shouldn't have made it into the repository in the first
> > > > place since they broke the demos.
> > > I just ran a few of the demos manually before pushing.
> > >
> > > > Is it possible to sort the demo runs such that the slowest ones are
> > > > at the end?
> > > That would be very difficult. The test scripts find all demos
> > > automatically and specifying the order would mean that we would need
> > > to list all the demos manually.
> > >
> > > > Is it possible to only run the python demos?
> > > > This would greatly speed up the test process...
> > > It is now. Just add the option --only-python when running the main
> > > test script.
> > Forgot to check this in? Doesn't work here.
>
> I thought it was enough to commit the changes on my laptop... :-)
> Pushed now.
>
> > Anyway, I added some timing code to the demo runs, and
> > nls/cahn-hillard/cpp and ode/lorenz/python eat 75% (50/25) of the
> > time. We could just push those to the end manually in the test script.
>
> Sounds good. We should also investigate if we can do something about
> those demos (coarser mesh etc) to reduce the time.

The Cahn-Hilliard demo can be skipped. The nature of the problem means 
that a coarse mesh won't deliver sensible results.

Garth

> ------------------------------------------------------------------------
>
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev