I noticed few things on OS X (10.6.4) when compiling this (fresh svn
checkout), the problem I can't fix is at the bottom:
0) The p4est script appears to work without modification, and the library
builds ok.
1) Everything configures and compiles fine when using the config line:
./configure --with-umfpack --with-blas --with-lapack
--with-p4est=/path_to/prog/lib/p4est --enable-mpi --disable-threads CC=mpicc
CXX=mpicxx F77=mpif77
2) Linking seems to be the issue:
A) The linker looks for 'libp4est.so' rather than 'libp4est.dylib', which is
what gets built on the mac. Same goes for 'libsc.so'. I fixed this by
modifying the source/Makefile, below is the modified snippet of code:
# and p4est
ifeq ($(USE_CONTRIB_P4EST),yes)
deplibs.g += $(DEAL_II_P4EST_DIR)/DEBUG/lib/libp4est$(shared-lib-suffix)
\
$(DEAL_II_P4EST_DIR)/DEBUG/lib/libsc$(shared-lib-suffix) \
-Wl,-rpath,$(DEAL_II_P4EST_DIR)/DEBUG/lib
deplibs.o += $(DEAL_II_P4EST_DIR)/FAST/lib/libp4est$(shared-lib-suffix)
\
$(DEAL_II_P4EST_DIR)/FAST/lib/libsc$(shared-lib-suffix) \
-Wl,-rpath,$(DEAL_II_P4EST_DIR)/FAST/lib
endif
B) I got a duplicate symbol error on linking, which I have not been able to
track down:
=====deal.II==========debug========== Linking library: libdeal_II.g.dylib
ld: duplicate symbol
dealii::FiniteElementData<3>::face_to_cell_index(unsigned int, unsigned int,
bool, bool, bool) const in
/Users/jsp203/prog/lib/deal_petsc/lib/debug/2d_fe_fe_data.o and
/Users/jsp203/prog/lib/deal_petsc/lib/debug/1d_fe_fe_data.o
collect2: ld returned 1 exit status
make[1]: *** [/Users/jsp203/prog/lib/deal_petsc/lib/libdeal_II.g.dylib]
Error 1
make: *** [debug] Error 2
Thank you to everyone for getting the distributing triangulation and
supporting functions implemented. I will certainly benefit from your hard
work.
Jon
On Mon, Oct 25, 2010 at 6:59 PM, Wolfgang Bangerth
<[email protected]>wrote:
>
> > Won't we kill Windows users with the dot? I would have preferred dealii.
>
> Ah, sorry. But I think we should be ok -- Windows today can handle dots
> just
> fine. Spaces would have been a bigger problem. That said, by using
> "deal.II"
> we would make the install paths the same as the package name -- there's
> something to be said for that.
>
> W.
>
> -------------------------------------------------------------------------
> Wolfgang Bangerth email: [email protected]
> www:
> http://www.math.tamu.edu/~bangerth/<http://www.math.tamu.edu/%7Ebangerth/>
>
> _______________________________________________
> dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
>
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
