[petsc-dev] [petsc-maint] Updating PETSc to support Elemental 0.80

[Satish Balay]

On Mon, 24 Jun 2013, Jack Poulson wrote:

> On Mon, Jun 24, 2013 at 11:31 AM, Satish Balay <balay at mcs.anl.gov> wrote:
> 
> >
> > Looks like parmetis.py is using the cmake option
> > '-DGKLIB_PATH=../headers' to set the path to GKlib.h
> >
> > [I'm not sure if this is prefereable to changing
> > pkg-parmetis/CMakeLists.txt as you've indicated]
> >
> >
> Thanks, Satish. I would strongly lean towards having pkg-metis/pkg-parmetis
> be a bit more intuitive to build as standalone packages (which is to say, I
> think it makes sense to have the include_directories(headers) statement).
> It is not obvious that one is expected to:
> 1) Install metis
> 2) Set METIS_PATH to the installation directory
> 3) Set GKLIB_PATH to ./headers
> 
> 
> > > Is PETSc using these CMakeLists.txt files? I can now build pkg-parmetis
> > on
> > > top of a pkg-metis install if I make the above change (and uncomment out
> > > lines 9-13 of pkg-parmetis/CMakeLists.txt and actually search for MPI).
> >
> > Hm - PETSc configure sets up MPI correctly before building
> > parmetis. So it should't automatically be searching for MPI.
> >
> >
> A compromise might be to only perform the search if the requisite MPI
> variables aren't already defined.
> 
> If I am to have Clique build on top of pkg-metis and pkg-parmetis, I will
> need to modify them to install enough header files so that it can make use
> of metislib.h and parmetislib.h.

So clique requires additional include files? I supporse these are
private include files from metis/parmetis..

>  I'll try to propose a patch over the next
> few days and include the 'include_directories(headers)' and replace the
> strange 'include(FindMpi); if(NOT MPI_FOUND)' statements with guarded
> 'find_package(MPI); if(NOT MPI_C_FOUND)' statements.


Looks like that piece of code is from upstream. It has the comment:
>>
# GK commented this out as it seems to be creating problems
<<

Satish