Hi Wolfgang, Thanks for the pointer about removing the enum. I also managed to get past that point but by removing the template <int dim_1> dependence (and any occurrences of dim_1). This obviously doesn't produce usable code, but it does seem to confirm that the problem is related to the template. This seems to be again confirmed when make reaches the grid_refinement class because it fails with template related errors (I presume exactly what you saw):
source/grid/grid_refinement.cc:508: error: non-template used as template source/grid/grid_refinement.cc:515: error: non-template used as template source/grid/grid_refinement.cc:522: error: non-template used as template source/grid/grid_refinement.cc:529: error: non-template used as template source/grid/grid_refinement.cc:539: error: non-template used as template etc It looks very much like the 3.3.3 releases are not handling the templates which deal.II needs and which make it so useful. I have a bug report filed with PathScale for an error it throws processing the trilinos library so maybe I can get some answers from them. Pretty much all compilers on the cluster are 3.3.3 but I see that they were built in 2006 so hopefully they have moved forward and maybe these issues will convince the cluster admin to get a newer version of the compilers. On the other cluster and the Trilinos front I can confirm that when I compiled --with-mpi-incdir=/... ,--with-mpi-libdir=/... and --enable-mpi, the mpi headers were installed correctly and deal.II compiled first time. The --enable-mpi switch seems to be needed even if you give mpi lib and inc directories. Thanks, Michael On Fri, Jun 12, 2009 at 1:31 PM, Wolfgang Bangerth<[email protected]> wrote: > > Michael, > >> Thank you for that response. I have followed up on the CellSimilarity >> compile error and just to confirm the syntax error you refer to was >> due to my earlier debugging attempts (I had tried to declare a >> CellSimilarity enum for it to find in the fe_system.h file). The true >> error that I am seeing is: >> >> In file included from source/fe/fe_system.cc:21: >> /CHPC/home/mrapson/lib/gcc/deal.II-6.2.1/deal.II/include/fe/fe_system.h: >>688: error: use of enum `enum dealii::CellSimilarity::Similarity' without >> previous declaration > > I've tried this with gcc 3.3.1 and indeed I see the same problem but > there's a simple solution: in all the lines the compiler complains about, > simply remove the 'enum' and leave 'CellSimilarity::Similarity' in place. > It shouldn't matter (and more recent compilers also don't complain), and > taking out the one word makes things compile. > > I've run into more problems in the file grid/grid_refinement.cc, though, > which I'm working on now. It may be that those problems are fixed in later > dot releases of 3.3.x. > > >> Concerning the trilinos. Both deal.II and trilinos were installed with >> mpi, but trilinos I used the --with-mpi-incdir=/... >> --with-mpi-libdir=/... rather than --with-mpi. I am now testing >> building trilinos with openmpi and with the --with-mpi switch as well, >> in case it is important for installing the mpi headers. > > Let me know what you find. > > >> I will send you my code for MonitorNode and a test case in a separate >> email. > > I'm looking forward to it! > > Best > W. > > ------------------------------------------------------------------------- > Wolfgang Bangerth email: [email protected] > www: http://www.math.tamu.edu/~bangerth/ > > _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
