[petsc-users] How can PETSc configure with MSMPI

[Satish Balay]

On Sun, 2 Feb 2014, Jed Brown wrote:

> Satish Balay <balay at mcs.anl.gov> writes:
> >> http://wiki.mpich.org/mpich/index.php/Frequently_Asked_Questions#Q:_Why_can.27t_I_build_MPICH_on_Windows_anymore.3F
> >
> > Yes - but that has no bearing on cygwin/gnu compiler usage [which was
> > the scope of this discussion]
> 
> They don't test it, support it, or track bug reports on any Windows
> platforms (with or without Cygwin).  It is the same canned response on
> an almost daily basis on the mailing list.  The last commit to mention
> Cygwin was in 2011.

I'm not sure what issue we are discussing here..

If a user wants to use PETSc with cygwin, MPICH or OpenMPI should work.

However if the user wants to use PETSc with cygwin - and needs
'upstream' support for all these packages - most of the packages that
work with --download-package are not really tested or supported by
upstream. [supleru? metis? hypre? others?]

If the user wants an MPI properly supported by cygwin - then they can
install OpenMPI thats packages as part of cygwin. [If not upstream -
the cygwin maintainger supports it?]

Most of the statements from MPICH folks regarding Windows is wrt MPICH
binary distribution for MS/Intel compilers. I don't know if they had a
position on cygwin before - or if that position changed [along with
their position on this binary distribution]

Satish