packaging mpich2 -- conflicts with lam and file layoutb

[Ed Hill]

On Tue, 2005-11-08 at 14:17 +0000, Paul Howarth wrote:
> Tom 'spot' Callaway wrote:
> > I'd rather do this:
> > 
> > %{_libdir}/${MPI_IMPL_NAME}/
> > %{_includedir}/${MPI_IMPL_NAME}/
> > 
> > And use something like alternatives to manage the binaries, without
> > putting them in a "magic" dir that violates the FHS. 
> > 
> > lam will probably need some work to meet these guidelines, but I'll push
> > internally to get it fixed for FC5.
> 
> Is alternatives the best way of dealing with this? It provides only for 
> changing the system-wide default, and doesn't help if regular users want 
>   to try different implementations themselves. At least if different 
> implementations are in different directories, the implementations can be 
> selected by changing environment variables such as PATH. The FHS doesn't 
> really cater for that, does it?

Hi Tom & Paul,

Thanks for the feedback!  And can someone please explain how the
alternatives thing works or, even better, point me towards a doc that
explains it?

I see how the libs/headers layout that Tom suggests above will work for
multiple different MPI implementations.  But I *don't* see how it works
for the different MPI binaries and man pages.  Or am I missing
something?

Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464