Package Details: deal-ii 9.6.2-1

Git Clone URL: https://aur.archlinux.org/deal-ii.git (read-only, click to copy)
Package Base: deal-ii
Description: An Open Source Finite Element Differential Equations Analysis Library
Upstream URL: https://www.dealii.org/
Keywords: fem pde science
Licenses: LGPL-2.1-or-later
Submitter: coincoin
Maintainer: coincoin (drwells)
Last Packager: drwells
Votes: 15
Popularity: 0.000205
First Submitted: 2013-09-11 11:11 (UTC)
Last Updated: 2024-12-24 03:10 (UTC)

Dependencies (28)

Required by (0)

Sources (1)

Latest Comments

« First ‹ Previous 1 2 3 4 5 6 7 Next › Last »

carlosal1015 commented on 2021-08-04 23:25 (UTC)

Thanks a lot :-), when I type pacman -Ql intel-mkl shows me /opt/intel/mkl/include/mkl.h.

drwells commented on 2021-08-04 20:32 (UTC)

You're welcome. That gist is very helpful.

  1. I think you figured it out but PETSc does not support compiling with one MPI version and linking against another so that should work.
  2. This is a known bug in deal.II's configuration script: see

https://github.com/dealii/dealii/issues/12435

In particular, we expect CMake to set LAPACK_INCLUDE_DIRS but it does not.

I can try to help but I am not that familiar with MKL. Do you know where the mkl.h header is installed? The test fails because it is not on the default search path.

carlosal1015 commented on 2021-08-04 19:24 (UTC) (edited on 2021-08-04 19:26 (UTC) by carlosal1015)

Thank you very much for the quick answer. I am sharing the log installation here. Since it is an unsupported (yet) package, I removed intel-openmpi. Then, I run yay -Syu deal-ii --noconfirm 2>&1 | tee error1.txt and I looked an error: #error "PETSc was configured with one OpenMPI mpi.h version but now appears to be compiling using a different OpenMPI mpi.h version" in the line 1301 of error1.txt. After I uninstall pets-complex from arch4edu repository and I run again yay -Syu deal-ii --noconfirm 2>&1 | tee error2.txt. I looked in the line 348 of error2.txt: Use other than Intel MKL implementation of BLAS/LAPACK (consult CMakeFiles/CMakeError.log for further information). I looked also that LAPACK_LIBRARIES are pointing to intel-mkl and finally finish with:

 74%] Linking CXX shared library ../lib/libdeal_II.g.so
collect2: fatal error: ld terminated with signal 9 [Killed]
compilation terminated.
make[2]: *** [source/CMakeFiles/deal_II.g.dir/build.make:1031: lib/libdeal_II.g.so.9.3.1] Error 1
make[1]: *** [CMakeFiles/Makefile2:1684: source/CMakeFiles/deal_II.g.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
[ 74%] Building CXX object source/matrix_free/CMakeFiles/obj_matrix_free_release.dir/shape_info.cc.o
[ 74%] Building CXX object source/non_matching/CMakeFiles/obj_non_matching_release.dir/immersed_surface_quadrature.cc.o
[ 74%] Building CXX object source/matrix_free/CMakeFiles/obj_matrix_free_release.dir/task_info.cc.o
[ 74%] Built target obj_non_matching_release
[ 75%] Building CXX object source/matrix_free/CMakeFiles/obj_matrix_free_release.dir/vector_data_exchange.cc.o
[ 75%] Built target obj_matrix_free_release

Finally I append the file /tmp/makepkg/deal-ii/src/build/detailed.log, /tmp/makepkg/deal-ii/src/build/CMakeFiles/CMakeError.log. If after I will try to remove the intel residue with lapack, etc, I'm sure it would build, surely someone else could confirm the successful compilation.

drwells commented on 2021-08-04 17:26 (UTC)

Thanks - I missed the intel-tbb -> tbb rename.

Issues with intel's MPI may be outside the scope of this PKGBUILD but if you have a bug report we can always send it upstream.

carlosal1015 commented on 2021-08-04 17:01 (UTC)

Thank you very much @drwells, @bartus for your valuable time on the update, I am having some error related to some MPI redefinitions using intel-mpi (orphan aur package). If this next attempt fails, I will remove intel-mpi.

intel-tbb was replaced by tbb a few months ago. I'm limiting my cpus cores and writing the log with tee command to post here in a few hours.

drwells commented on 2021-08-04 15:19 (UTC)

9.3.1 is now officially out so I have updated the AUR script.

Please post if you have problems - I will try to be more available for resolving issues. Special thanks to @bartus for the help (fortunately we no longer require patches so SRCDEST should work automatically).

archaeal commented on 2021-01-27 19:24 (UTC)

I was able to build deal.ii, but had two of the tests failed, one with p4est and the other with scalapack:

The following tests FAILED:
     6 - p4est.debug (Failed)
    16 - scalapack.debug (Failed)

The output from p4est:

./p4est.debug: symbol lookup error: ./p4est.debug: undefined symbol: _ZN6dealii8parallel11distributed13TriangulationILi2ELi2EEC1EP19ompi_communicator_tNS_13TriangulationILi2ELi2EE13MeshSmoothingENS3_8SettingsE

The output from scalapack:

./scalapack.debug: symbol lookup error: ./scalapack.debug: undefined symbol: _ZN6dealii15ScaLAPACKMatrixIdEC1EjRKSt10shared_ptrIKNS_9Utilities3MPI11ProcessGridEEjNS_13LAPACKSupport8PropertyE

bartus commented on 2021-01-22 09:30 (UTC) (edited on 2021-01-22 09:34 (UTC) by bartus)

@drwells: Nice, but you you still have an issue in build(). Fist you can't assume your sources will be in parent directory to $srcdir as user can sideload all their sources from different path by defining SRCDEST variable. Also since u've added the patches to the source array they automatically linked by makepkg to $srcdir rendering the cp call redundant not mention having wrong path (¬‿¬ ).

[Patch]

merge with: $git am -s < <(curl -s http://ix.io/2MSr)

drwells commented on 2021-01-22 00:44 (UTC)

@bartus I've updated the PR to correctly handle the patches. I opted to keep the patch files here rather than pull down two commits from GitHub - it's just a personal preference, but I like being able to see the patch files in AUR packages.

drwells commented on 2021-01-12 13:44 (UTC)

@bartus Sorry for the delay. I really appreciate you writing this up - this is a much better way to handle patches! I'll update the package later this week.

« First ‹ Previous 1 2 3 4 5 6 7 Next › Last »