Ah... That is different then... I have no problem etags'ing the source tree and replacing HAVE_MPI. Let me know if you want me to do that.

Still, -DHAVE_CONFIG_H is such a generic thing it suprises me (a) they use it and (b) users are required to use it.

Seems like it might be worth a complaint on the NOX d
evelopment list...


From: Derek Gaston
To: Kirk, Benjamin (JSC-EG)
Cc: libmesh-devel@lists.sourceforge.net
Sent: Tue Sep 16 12:06:06 2008
Subject: Re: [Libmesh-devel] HAVE_MPI collisions...

On Sep 16, 2008, at 11:01 AM, Kirk, Benjamin (JSC-EG) wrote:

So long as they mean the same thing I suppose we could

#ifndef HAVE_MPI
#define HAVE_MPI

Wouldn't it be _them_ that would need the #ifndef?  Their #defines are clobbering _ours_.

A question, though...  I have not seen this, and I see you are including from the trilinos source directory.  Do you see this if you 'make install' and include from the install dir?

That's a possibility, I'll try it soon.  But more likely the reason you haven't seen this is that for just the Epetra stuff you don't have to use -DHAVE_CONFIG_H... but for NOX you do.  And when you do that it pulls in these Package_config.h files (which, as I say is necessary for NOX).  What's currently checked into the repo doesn't have this flag, but what I'm working on now does.

I'll report back when I get a chance to try the install dir stuff.