[Trilinos-Users] building packages

Bartlett, Roscoe A rabartl at sandia.gov
Fri Jul 9 11:24:40 MDT 2010


Chris,

Here is the problem.  CMake is not getting all of your configure options and it thinks that only ML is explicitly being enabled as clearly shown by:

     Explicitly enabled packages on input (by user):  ML 1

The problem is that the package names are case sensitive (and you are not using the write case).  To see the correct package names with the correct case, either look at the file cmake/TrilinosPackages.cmake or just look at the output clearly telling you what packages are enable or not enabled in:


> Final set of enabled packages:  Teuchos ThreadPool Kokkos Epetra
> Zoltan Triutils Tpetra EpetraExt Isorropia AztecOO Galeri Amesos
> Ifpack ML 14
> 
> Final set of non-enabled packages:  TrilinosFramework RTOp Shards
> Thyra Pliris Pamgen Komplex Belos Tifpack Stratimikos Meros FEI
> Anasazi TriKota Stokhos STK Sacado Intrepid Phalanx Phdmesh NOX
> Moertel TrilinosCouplings Rythmos MOOCHO Piro Sundance PyTrilinos
> Didasko Optika Mesquite 31


I know this is a lot of output but everything you need to see about how packages and tests are enabled is right in the CMake configure output.

Fix the case of the package names to match those shown above and you will be fine.

Note that the problem of misspelling the name of a CMake cache variable on the cmake commandline and then having it ignored by CMake is a fundamental weakness of CMake.  It is something that we could get around but we would have to provide our own 'configure' wrapper script and we would have to provide our own enable options that we check on our own.  We really don't want to go down this road.

Kitware, at some point, might fix the problem of misspelled CMake cache variables being ignored but I have no idea if or when that would happen.  We have it on our backlog for our contract with Kitware but we have many other higher-priority issues that need to be addressed first (mostly with other tools, cmake is in good shape).  Unless this is done, we would have little choice but to 100% wrap CMake and provide our own (case insensitive) validation of user input parameters if we want to catch common user misspelling of input CMake cache variables.  Again, we really don't want to do this.  We would much rather pay to improve CMake and then every project that uses CMake would benefit as well (and would make sure that it was maintained).

Note that if you use a CMake GUI (the QT GUI or ccmake) that the pre-defined cache variables would be right there and you could not misspell them.  I think that is the way that Kitware intended most people to use CMake and there there is no issue with misspelling cache variables (because they are pre-loaded for you).

Sorry for the trouble.

- Ross


> -----Original Message-----
> From: chris at paratools.com [mailto:chris at paratools.com]
> Sent: Friday, July 09, 2010 10:56 AM
> To: Bartlett, Roscoe A
> Subject: RE: [Trilinos-Users] building packages
> 
> Hi Ross,
>    here you go..
> Chris
> 
> Configuring Trilinos build directory
> 
> -- Found PythonInterp: /usr/bin/python2.6
> 
> Setting up major user options ...
> 
> -- Setting Trilinos_ENABLE_EXAMPLES=ON because Trilinos_ENABLE_TESTS=ON
> 
> Reading the list of packages from TrilinosPackages.cmake ...
> 
> -- Trilinos_NUM_PACKAGES='45'
> 
> Reading the list of TPLs from TrilinosTPLs.cmake ...
> 
> -- Trilinos_NUM_TPLS='43'
> 
> Explicitly enabled packages on input (by user):  ML 1
> 
>  ...
>
> Final set of enabled packages:  Teuchos ThreadPool Kokkos Epetra
> Zoltan Triutils Tpetra EpetraExt Isorropia AztecOO Galeri Amesos
> Ifpack ML 14
> 
> Final set of non-enabled packages:  TrilinosFramework RTOp Shards
> Thyra Pliris Pamgen Komplex Belos Tifpack Stratimikos Meros FEI
> Anasazi TriKota Stokhos STK Sacado Intrepid Phalanx Phdmesh NOX
> Moertel TrilinosCouplings Rythmos MOOCHO Piro Sundance PyTrilinos
> Didasko Optika Mesquite 31
>
> ...
> 
> 
> Quoting "Bartlett, Roscoe A" <rabartl at sandia.gov>:
> 
> > Chris,
> >
> > Can you please attach the entire configure output?  The problem
> > should be obvious if I can see that.
> >
> > - Ross
> >
> >> -----Original Message-----
> >> From: trilinos-users-bounces at software.sandia.gov [mailto:trilinos-
> >> users-bounces at software.sandia.gov] On Behalf Of chris at paratools.com
> >> Sent: Friday, July 09, 2010 10:26 AM
> >> To: trilinos-users at software.sandia.gov
> >> Subject: [Trilinos-Users] building packages
> >>
> >> Hi,
> >>      here is copy of my do-configure script.
> >> EXTRA_ARGS=$@
> >> TRILINOS_PATH=/home/chris/Work/Apps/trilinos/10.2.2
> >> TRILINOS_HOME=/home/chris/Testing/trilinos-10.2.2-Source
> >>
> >> cmake \
> >> -D CMAKE_INSTALL_PREFIX:PATH=$TRILINOS_PATH \
> >> -D CMAKE_BUILD_TYPE:STRING=RELEASE \
> >> -D CMAKE_CXX_FLAGS:STRING="-m32" \
> >> -D CMAKE_C_FLAGS:STRING="-m32" \
> >>            -D Trilinos_ENABLE_MESQUITE:BOOL=ON \
> >> -D Trilinos_ENABLE_ML:BOOL=ON \
> >> -D Trilinos_ENABLE_EPETRA:BOOL=ON \
> >> -D Trilinos_ENABLE_EPETRAEXT:BOOL=ON \
> >> -D Trilinos_ENABLE_TEPETRA:BOOL=ON \
> >> -D Trilinos_ENABLE_JEPETRA:BOOL=ON \
> >> -D Trilinos_ENABLE_AZTECOO:BOOL=ON \
> >> -D Trilinos_ENABLE_ZOLTAN:BOOL=ON \
> >> -D Trilinos_ENABLE_TESTS:BOOL=ON \
> >> -D BLAS_LIBRARY_DIRS:PATH=/home/chris/Work/Apps/ATLAS/3.9.25/lib \
> >> -D BLAS_LIBRARY_NAMES:STRING="atlas;cblas;f77blas" \
> >> -D LAPACK_LIBRARY_DIRS:PATH=/home/chris/Work/Apps/ATLAS/3.9.25/lib \
> >> -D LAPACK_LIBRARY_NAMES:STRING="lapack" \
> >> -D TPL_ENABLE_MPI:BOOL=ON \
> >> -D MPI_BASE_DIR:PATH="/usr/local/packages/openmpi-1.3.2" \
> >> -D MPI_EXEC:FILEPATH="mpirun" \
> >> -D MPI_EXEC_MAX_NUMPROCS:STRING=4 \
> >> -D MPI_EXEC_NUMPROCS_FLAG:STRING=-np \
> >> -D DART_TESTING_TIMEOUT:STRING=600 \
> >> $EXTRA_ARGS \
> >> ${TRILINOS_HOME}
> >>
> >>
> >> and when I run it I get....
> >>
> >> Setting up testing support ...
> >>
> >>
> >> Configuring individual enabled Trilinos packages ...
> >>
> >> -- Processing enabled package: Teuchos
> >> -- Performing Test HAVE_GCC_ABI_DEMANGLE
> >> -- Performing Test HAVE_GCC_ABI_DEMANGLE - Failed
> >> -- Performing Test HAVE_TEUCHOS_BLASFLOAT
> >> -- Performing Test HAVE_TEUCHOS_BLASFLOAT - Failed
> >> -- Performing Test CXX_COMPLEX_BLAS_WORKS
> >> -- Performing Test CXX_COMPLEX_BLAS_WORKS - Failed
> >> -- Performing Test HAVE_FIXABLE_COMPLEX_BLAS_PROBLEM
> >> -- Performing Test HAVE_FIXABLE_COMPLEX_BLAS_PROBLEM - Failed
> >> -- Processing enabled package: ThreadPool
> >> -- Processing enabled package: Kokkos
> >> -- Processing enabled package: Epetra
> >> -- Processing enabled package: Zoltan
> >> -- Processing enabled package: Triutils
> >> -- Processing enabled package: Tpetra
> >> -- Processing enabled package: EpetraExt
> >> -- Processing enabled package: Isorropia
> >> -- Processing enabled package: AztecOO
> >> -- Processing enabled package: Galeri
> >> -- Processing enabled package: Amesos
> >> -- Processing enabled package: Ifpack
> >> -- Processing enabled package: ML (tests, examples)
> >>
> >> Exporting library dependencies ...
> >>
> >>
> >> So CMAKE is not enabling MESQUITE and only the tests/examples are
> for
> >> ML.
> >>
> >> -- Performing Test HAVE_GCC_ABI_DEMANGLE
> >> -- Performing Test HAVE_GCC_ABI_DEMANGLE - Failed
> >> -- Performing Test HAVE_TEUCHOS_BLASFLOAT
> >> -- Performing Test HAVE_TEUCHOS_BLASFLOAT - Failed
> >> -- Performing Test CXX_COMPLEX_BLAS_WORKS
> >> -- Performing Test CXX_COMPLEX_BLAS_WORKS - Failed
> >> -- Performing Test HAVE_FIXABLE_COMPLEX_BLAS_PROBLEM
> >> -- Performing Test HAVE_FIXABLE_COMPLEX_BLAS_PROBLEM - Failed
> >>
> >> What are these tests? Why are they Failed? Is it important ?????
> >>
> >> The configure script does complete and have been able to compile the
> >> code but can only run the ML tests (via make test).
> >>
> >> gcc version 4.4.1 20090725 (Red Hat 4.4.1-2) (GCC)
> >> fedora  2.6.30.10-105.2.23.nic.fc11.i686.PAE
> >>
> >> thanks
> >> Chris
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> Trilinos-Users mailing list
> >> Trilinos-Users at software.sandia.gov
> >> http://software.sandia.gov/mailman/listinfo/trilinos-users
> >
> >
> 
> 
> 




More information about the Trilinos-Users mailing list