dolfin team mailing list archive

["Åsmund Ødegård" <aasmund@xxxxxxxxx>]

On Jan 11, 2008 9:51 PM, Anders Logg <logg@xxxxxxxxx> wrote:

> On Thu, Jan 10, 2008 at 10:39:08AM -0500, Gideon Simpson wrote:
> > A summary of what I have discovered.
> >
> > With the following setup:
> >
> > os x 10.4.8 x86
> > gcc 4.0.1 (apple)
> > mpich2-1.06p1, dynamic libraries
> > pets-2.3.3-p8, dynamic libraries
> > hypre-2.0.0, static libraries,
> > umfpack 4.3, static libaries
> > numpy 1.0.4
> > swig 1.3.31
> > ffc-0.4.3
> > FIAT-0.3.3
> > UFC-1.0
> > dolfin -0.7.1
> >
> > I note the following behavior:
> >
> > If i configure dolfin, as in the tutorial, everything, including
> pydoflin,
> > compiles, but when I try to run any dolfin based application, i generate
> the
> > following error:
> >
> > valkyrie:cpp$ ./demo
> > dyld: Symbol not found: _HYPRE_ParCSRPilutDestroy
> >   Referenced from: /Users/gideon/software/petsc-2.3.3-p8/lib/
> > darwin8.11.1-cxx-debug/libpetscksp.dylib
> >   Expected in: dynamic lookup
> >
> > Trace/BPT trap
> >
> > If i give dolfin the LDFLAGS and LIB flags for HYPRE, then when the
> dolfin
> > libraries are being built, I get the error:
> >
> > ld: warning multiple definitions of symbol _mpipriv2_
> > /Users/gideon/software/petsc-2.3.3-p8
> /lib/darwin8.11.1-cxx-debug/libpetsc.dylib
> > (single module) definition of _mpipriv2_
> > /Users/gideon/software/lib/libmpich.dylib(single module) definition of
> > _mpipriv2_
> > ld: warning multiple definitions of symbol _mpipriv1_
> > /Users/gideon/software/petsc-2.3.3-p8
> /lib/darwin8.11.1-cxx-debug/libpetsc.dylib
> > (single module) definition of _mpipriv1_
> > /Users/gideon/software/lib/libmpich.dylib(single module) definition of
> > _mpipriv1_
> > ld: warning multiple definitions of symbol _mpiprivc_
> > /Users/gideon/software/petsc-2.3.3-p8
> /lib/darwin8.11.1-cxx-debug/libpetsc.dylib
> > (single module) definition of _mpiprivc_
> > /Users/gideon/software/lib/libmpich.dylib(single module) definition of
> > _mpiprivc_
> > ld: common symbols not allowed with MH_DYLIB output format with the
> > -multi_module option
> > /Users/gideon/software/petsc-2.3.3-p8/externalpackages/hypre-2.0.0/
> > darwin8.11.1-cxx-debug/lib/libHYPRE.a(globalObjects.o) definition of
> common
> > _msgBuf_dh (size 1024)
> > /usr/bin/libtool: internal link edit command failed
> >
> > Finally, if I include the flags for HYPRE, but only built static dolfin
> > libraries and disable pydolfin, it compiles and applications work.
>  However, I
> > have to make modifications to dolfin.pc, as it fails to include the
> HYPRE flags
> > and the -framework vecLib for apple's BLAS/LAPACK.
> >
> >
> >
> >
> > -gideon
>
> A new scons-based build system for DOLFIN is on its way. Hopefully we
> can get all this sorted out then.
>
> If Åsmund reads this, maybe he can give a status update on the build
> system?


I read it. There are really just a few minor things that remains to be
sorted out. Most important I think is picking the right compiler when
linking with PETSc.

Although I have a bit too much to do these days, I'm making progress. Stay
tuned.


>
> --
> Anders
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev
>



-- 
[simula.research laboratory]
                Åsmund Ødegård
                IT-Manager
                phone: +4790069915
                http://www.simula.no/