On Fri, Sep 17, 2010 at 9:03 AM, John Peterson <peterson@cfdlab.ae.utexas.edu> wrote:
I don't think we should change all of them!  There are cases in the
numerics classes where the intent is to downcast to a more derived
type, e.g. a PetscVector.  You'd have to look at each case by hand,
which sounds like a pain. 

"Fixing" libmesh shouldn't be our concern here.  We HAVE to be able to reliably dynamic cast.  If the compiler isn't going to give that to us that's a non-starter.

This isn't just a "make it work on OSX" moment.  We shouldn't have to change any valid code just because Apple has a bug in their compilers.