Search Criteria
Package Details: meep-git 1.4.3.r81.g856ccafe-1
| Git Clone URL: | https://aur.archlinux.org/meep-git.git (read-only) |
|---|---|
| Package Base: | meep-git |
| Description: | Free finite-difference time-domain (FDTD) software for electromagnetic simulations (includes python interface) |
| Upstream URL: | https://github.com/stevengj/meep |
| Licenses: | |
| Conflicts: | |
| Provides: | |
| Submitter: | heitzmann |
| Maintainer: | heitzmann |
| Last Packager: | heitzmann |
| Votes: | 0 |
| Popularity: | 0.000000 |
| First Submitted: | 2017-11-14 09:36 |
| Last Updated: | 2018-04-11 19:41 |
Dependencies (15)
- fftw (fftw-mpich, fftw-mpi)
- gsl
- harminv-git
- hdf5 (hdf5-mpich, hdf5-openmpi-java, hdf5-openmpi)
- libctl-git
- openmpi (openmpi-git, openmpi-cuda)
- python-numpy (python-numpy-openblas, python-numpy-mkl)
- suitesparse (suitesparse-mkl)
- swig (swig-git)
- gcc-fortran (gcc-fortran-multilib-x32, gcc-fortran-git) (make)
- git (git-git) (make)
- h5utils (h5utils-git) (optional) – visualization and conversion of scientific data
- mpb-git (optional) – calculation of electromagnetic modes
- python-h5py (python-h5py-git, python2-h5py-git, python-h5py-openmpi) (optional) – process HDF5 data from python
- python-mpi4py (python-mpi4py-intel) (optional) – use MPI from the python interface
Required by (2)
- python2-meep (requires meep)
- python2-meep-bzr (requires meep)
Latest Comments
ian11213 commented on 2018-06-05 20:59
Updating libctl to the latest labctl-git, solves the build problem mentioned below.
ian11213 commented on 2018-06-04 20:41
Package seems to need recompiling following the recent hdf5/python-h5py upgrades, as attempts to write results to hdf5 fail silently. But currently the build fails with: bend-flux-ll.cpp:97:23: error: ‘make_prism’ was not declared in this scope objects[0] = make_prism( dielectric, vertices, 4, height, axis); Suspect this is an upstream issue - see https://www.mail-archive.com/meep-discuss@ab-initio.mit.edu/msg06033.html and ff. Downgrading hdf5 to 1.10.1-3 and python-h5py to 2.7.1-1 gets me back on the road.
ian11213 commented on 2018-05-09 12:18
Thanks for the help and advice; I went through recompiling everything listed by pacman -Qqm which looked remotely relevant. Something solved the issue - I am not sure what. Now removed that evil link!
heitzmann commented on 2018-05-09 00:54
ian11213, recompiling the packages is the only thing I can think of. I had the exact same issue (with python-numpy in fact) and I solved it by recompiling openblas.
I strongly suggest you try to find the library that is looking for the old libgfortran and recompile it instead of simlinking to ligbfortran.so.4
ian11213 commented on 2018-05-08 19:19
Hi, Heitzmann - thanks. Yes, I understand I am doing very bad things here. But I had already recompiled all the AUR packages I can find; that did not fix it and I urgently needed meep running. Any other hints? Otherwise I'll just do another sweep through looking for any packages I may have missed previously. Thanks for your work maintaining the package.
heitzmann commented on 2018-05-08 10:41
ian11213, that is a bad solution.
After upgrading gcc-libs (which includes libgfortran) you have to recompile any packages you compiled locally, such as any *-git packages from AUR, openBLAS, etc.
ian11213 commented on 2018-05-07 23:00
After latest package upgrades (7 May) gives Python 3 error: Traceback (most recent call last): File "/home/me/sample.py", line 1, in <module> import meep as mp File "/usr/lib/python3.6/site-packages/meep/init.py", line 8, in <module> from . import _meep ImportError: libgfortran.so.4: cannot open shared object file: No such file or directory</module></module>
Working solution by making link titled libgfortran.so.4 to libgfortran.so.5 in /usr/lib
haawda commented on 2018-04-11 18:39
Please add git as makedepends.