On 11 September 2015 at 09:33, Julian Andrej <j...@tf.uni-kiel.de> wrote:

> Would it be possible to include a precompiled version into the ppa
> install? Just for the convenience installation or a quick install.
>
>
That would require someone building and maintaining the a PETSc package for
the ppa. Someone might be willing (but it won't be me!).

Personally, I don't think having a PETSc ppa package is a good idea because
PETSc should be allowed to download it's own dependencies because it picks
compatible versions and applies patches for bugs that upstream developers
have neglected to fix. This isn't compatible with the 'proper' Debian
packaging approach.

Garth


> On Fri, Sep 11, 2015 at 10:31 AM, Anders Logg <anders.l...@gmail.com>
> wrote:
> > Agree!
> >
> > --
> > Anders
> >
> >
> > fre 11 sep. 2015 kl 10:11 skrev Garth N. Wells <gn...@cam.ac.uk>:
> >>
> >> I would like to propose that we switch to DOLFIN supporting PETSc 3.6
> >> only. Martin has just reported a number of bugs that in all likelihood
> are
> >> due to bugs in the Ubuntu/Debian PETSc 3.4 package, or in old versions
> of
> >> PETSc dependencies. Finding work-arounds for these is not a good use of
> >> anyone's time.
> >>
> >> A consequence of bumping the PETSc version is that the Debian/Ubuntu
> >> packages would no longer be configured with PETSc. This isn't such a big
> >> deal since:
> >>
> >> 1. I understand that the Debian/Ubuntu PETSc package is without a
> >> maintainer, so is likely to disappear at some point;
> >> 2. Anyone who needs PETSc should be using a supported version of PETSc,
> >> which at present means building it themselves;
> >> 3. PETSc is for performance and parallel simulations, in which case it
> >> should be configured for the machine it's used on; and
> >> 4. With the fallback of EIgen for linear algebra, a DOLFIN package
> without
> >> PETSc would still be fast in serial and a wide range of linear solvers
> are
> >> available.
> >>
> >> Garth
> >
> >
> > _______________________________________________
> > fenics mailing list
> > fenics@fenicsproject.org
> > http://fenicsproject.org/mailman/listinfo/fenics
> >
>
_______________________________________________
fenics mailing list
fenics@fenicsproject.org
http://fenicsproject.org/mailman/listinfo/fenics

Reply via email to