dolfin team mailing list archive

["Matthew Knepley" <knepley@xxxxxxxxx>]

On 11/16/06, Garth N. Wells <g.n.wells@xxxxxxxxxx> wrote:
> Perhaps branch is the wrong word. I presume that it will take a bit of
> time to gets things in place and tested, so I don't want this to put
> non-build system developments within DOLFIN on ice.
>
> Could you sketch out how the transition will take place?

There will be no code separation. The build system needs certain information
in order to proceed. Right now, this information is coming from pkg-config
files which has two manin deficiencies: they may not exist and this information
was not being tested. We can rectify both, but you can still use the pkg-config
files with this system to replicate what you get with the current
build. The only
real transition is in the input needed and the level of safety.

  Matt

> Garth
>
> Matthew Knepley wrote:
> > On 11/16/06, Garth N. Wells <g.n.wells@xxxxxxxxxx> wrote:
> >> OK, so effectively there will be a temporary build system branch. Could
> >> this be placed on fenics.org so that we can follow it?
> >
> > Actually, I do not see any need for a branch. The build part is exactly
> > the same
> > for now as it has been. The only thing that will change is now you do
> > not have
> > to specify all the configure options since we will have search, and
> > more exhaustive
> > tests will be run on what you specify. If you want to replicate the
> > previous functionality,
> > just specify everything up front like you have been. Asmund, do you agree?
> >
> >   Matt
> >
> >> Also, what time frame are we looking at?
> >>
> >> Garth
> >>
> >> Anders Logg wrote:
> >> > I don't think there will be a problem. The new build of DOLFIN will be
> >> > developed separately and we will all get a chance to try it out before
> >> > making the move.
> >> >
> >> > /Anders
> >> >
> >> >
> >> > On Thu, Nov 16, 2006 at 12:47:13PM +0100, Garth N. Wells wrote:
> >> >> I'm all for a new build system. I like radical changes :).
> >> >>
> >> >> I would prefer to make a temporary branch "dolfin-build" since we
> >> don't
> >> >> know for sure how long it will take. I don't want a broken build
> >> system
> >> >> to stop developments in the DOLFIN kernel.
> >> >>
> >> >> Garth
> >> >>
> >> >>
> >> >> Anders Logg wrote:
> >> >>> On Thu, Nov 16, 2006 at 11:29:59AM +0100, jjan@xxxxxxxxxxx wrote:
> >> >>>>>> On Thu, Nov 16, 2006 at 11:10:31AM +0100, Johan Hoffman wrote:
> >> >>>>>>> Sounds good to me.
> >> >>>>>>>
> >> >>>>>>> The stable branch will be available under dolfin-stable i guess
> >> >>>>>>> (Johan?).
> >> >>>>>> What do you mean?
> >> >>>>>>
> >> >>>>>> The new build system will be used for all branches, stable or not.
> >> >>>>>>
> >> >>>>>> /Anders
> >> >>>>> Yes of course, but if the switch means that the build system
> >> will be
> >> >>>>> broken for 2 weeks, then the stable branch should provide an
> >> alternative
> >> >>>>> in the mean time. Right? That must be the idea of a dolfin-stable?
> >> >>>>>
> >> >>>>> /Johan
> >> >>>>>
> >> >>>> I'm also very positive to the move to a new build system, Åsmund
> >> and Matt
> >> >>>> seem to be very capable of taking care of the strategy. But Johan
> >> H. has a
> >> >>>> point, I think we should release before the move. Then the
> >> release can be
> >> >>>> used as the dolfin-stable branch and applications developed
> >> against that.
> >> >>>> When the build system starts working again, we merge the branches
> >> and
> >> >>>> release again.
> >> >>> Sounds good.
> >> >>>
> >> >>> The switch to the new build system is not immediate. It will take
> >> some
> >> >>> time to put things together and convert the current build system
> >> for DOLFIN.
> >> >>>
> >> >>>> Is there anything left to do before the release?
> >> >>> Testing. Please try out all the demos, benchmarks, unit test etc.
> >> >>> Then we can make a release of DOLFIN + FFC (+FIAT?).
> >> >>>
> >> >>> /Anders
> >> >>>
> >> >>>
> >> >>>>   Johan
> >> >>>
> >> >>>
> >> >>>> _______________________________________________
> >> >>>> DOLFIN-dev mailing list
> >> >>>> DOLFIN-dev@xxxxxxxxxx
> >> >>>> http://www.fenics.org/mailman/listinfo/dolfin-dev
> >> >>> _______________________________________________
> >> >>> DOLFIN-dev mailing list
> >> >>> DOLFIN-dev@xxxxxxxxxx
> >> >>> http://www.fenics.org/mailman/listinfo/dolfin-dev
> >> >>>
> >> >> _______________________________________________
> >> >> DOLFIN-dev mailing list
> >> >> DOLFIN-dev@xxxxxxxxxx
> >> >> http://www.fenics.org/mailman/listinfo/dolfin-dev
> >> > _______________________________________________
> >> > DOLFIN-dev mailing list
> >> > DOLFIN-dev@xxxxxxxxxx
> >> > http://www.fenics.org/mailman/listinfo/dolfin-dev
> >> >
> >>
> >> --
> >>
> >> Dr. Garth N. Wells
> >> Faculty of Civil Engineering and Geosciences
> >> Delft University of Technology
> >> Stevinweg 1
> >> 2628 CN Delft
> >> Netherlands
> >>
> >> tel.   +31 15 278 7922
> >> fax.   +31 15 278 6383
> >> e-mail g.n.wells@xxxxxxxxxx
> >> url    http://www.mechanics.citg.tudelft.nl/~garth
> >> _______________________________________________
> >> DOLFIN-dev mailing list
> >> DOLFIN-dev@xxxxxxxxxx
> >> http://www.fenics.org/mailman/listinfo/dolfin-dev
> >>
> >
> >


--
"Failure has a thousand explanations. Success doesn't need one" -- Sir
Alec Guiness