[petsc-users] How can PETSc configure with MSMPI

Matthew Knepley knepley at gmail.com
Sun Feb 2 13:09:37 CST 2014

On Sun, Feb 2, 2014 at 12:52 PM, Jed Brown <jed at jedbrown.org> wrote:

> Matthew Knepley <knepley at gmail.com> writes:
> > I don't think this is apples-to-apples since there are later versions
> > of PETSc that do everything the earlier versions do. This is a case of
> > the developer getting run over by a bus (or at least their courage)
> > and us finding a way to keep the functionality.
> We removed DMMG, which is "loss of functionality" in the sense that old
> code doesn't work without significant refactoring.  We got rid of it

But it is not a loss of functionality since we replaced it with something
can do everything it did. Whereas MPICH did not replace this with something
that works on Windows. It is wholly different.


> because it was high-maintenance and non-composable.  It's not perfect
> parity, but Windows/Cygwin was dropped because it was high-maintenance,
> so it's not that different.  Developer time is in limited supply and
> Windows users tend to prefer binary installs and non-Cygwin solutions,
> so they're more prone to grumble about your Windows support than to
> thank you for maintaining it on a volunteer basis.
> We can disagree with the decision of the MPICH project, but I see their
> rationale.

What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140202/ae3a5738/attachment.html>

More information about the petsc-users mailing list