Two more observations/questions from a novice user: * IPython is not part of the hashdist. That is if you load dolfin from IPython you will load the non-hashdist version (if there is one).
* System Python modules libraries are not in the PYTONPATH when following the instructions (for example scipy). On Ubunutu machines you get them by running: "export PYTHONPATH=/usr/lib/python2.7/dist-packages/:$PYTHONPATH" * Is there an easy way to add more packages to the on click installation (e.g. I like having ipython, scipy and dolfin-adjoint)? 2014-12-10 10:57 GMT+01:00 Anders Logg <[email protected]>: > This is something I haven't figured out. It would be good to have a > discussion on this. > > One option might be to just keep all of FEniCS with dependencies in > HashDist (using the fenics-install.sh script). > > Then for all FEniCS Python components, one sets up a separate directory > for installation during development, in combination with a config file that > can be sourced to put that directory first in the path. > > For DOLFIN development, one would source the hashdist config and work with > cmake.local. > > But there are likely better ways to handle this using suitable hashdist > commands. Even if the install script does not install hashdist, I assume it > will be a good thing for developers to actually install hashdist so that > they may talk to hashdist outside of running the install script. > > -- > Anders > > > Wed Dec 10 2014 at 8:47:50 AM skrev Martin Sandve Alnæs < > [email protected]>: > >> Do you have a suggestion for a work flow while developing? I.e. >> building intermediate uncommitted work repeatedly on top of a hashdist >> stack without storing every version in hashdist? Maybe best to have a >> hashdist stack with all dependencies only and just work with fenics >> components outside hashdist? >> >> Martin >> 10. des. 2014 08:06 skrev "Anders Logg" <[email protected]>: >> >> Yes, it is quite fantastic isn't it... :-) >>> >>> The problem you noted with the name of the config file has been fixed. >>> >>> The idea is that one can build many different configurations/versions >>> of FEniCS and keep the config files around for switching between them. >>> >>> Note that each config file contains info about which versions of >>> FEniCS were used to build so it's easy to go back later and check which >>> build one has. >>> >>> -- >>> Anders >>> >>> >>> Wed Dec 10 2014 at 6:41:59 AM skrev Simon Funke < >>> [email protected]>: >>> >>>> This is fantastic! The installation of the dev version with hasdist >>>> works fine on my Ubuntu 14.10 laptop. >>>> >>>> My only comment is that the script tells me at the end: >>>> >>>> """ >>>> For your convenience, a configuration file named fenics.dev has been >>>> created in the current directory. For setting up your environment, you >>>> should issue the following command: >>>> >>>> source fenics.master >>>> """ >>>> >>>> The last line should be "source fenics.dev" >>>> >>>> >>>> >>>> 2014-12-09 22:46 GMT+01:00 Lizao Li <[email protected]>: >>>> >>>>> @Johannes, the SLEPc patched worked. Thanks~ On Archlinux, these two >>>>> lines builds FEniCS successfully: >>>>> >>>>> TMP=/tmp/test; mkdir $TMP; cd $TMP; ln -s `which python2` python >>>>> >>>> >>>>> wget -O - http://fenicsproject.org/fenics-install.sh | bash >>>>> >>>>> Since python is included in the package, it is probably a good >>>>> idea to include ipython as well. The default repl is a pain to use. >>>>> >>>>> @Jan, I tried GCC4.8.2 on the same CentOS and it still fails with >>>>> the same error. Upon digging further into the build log, it seems dolfin >>>>> did something funny: >>>>> >>>>> 2014/12/09 14:50:53 - INFO: [package:run_job] -- The C compiler >>>>> identification is GNU 4.4.7 >>>>> 2014/12/09 14:50:53 - INFO: [package:run_job] -- The CXX compiler >>>>> identification is GNU 4.4.7 >>>>> 2014/12/09 14:50:53 - INFO: [package:run_job] -- Check for working C >>>>> compiler: * /usr/bin/cc* >>>>> 2014/12/09 14:50:54 - INFO: [package:run_job] -- Check for working C >>>>> compiler: * /usr/bin/cc* -- works >>>>> 2014/12/09 14:50:54 - INFO: [package:run_job] -- Detecting C compiler >>>>> ABI info >>>>> 2014/12/09 14:50:54 - INFO: [package:run_job] -- Detecting C compiler >>>>> ABI info - done >>>>> 2014/12/09 14:50:54 - INFO: [package:run_job] -- Check for working CXX >>>>> compiler: /usr/bin/c++ >>>>> 2014/12/09 14:50:54 - INFO: [package:run_job] -- Check for working CXX >>>>> compiler: /usr/bin/c++ -- works >>>>> >>>>> Note that in the above cc and c++ (wrong version), not gcc and g++ >>>>> (correct version), were used. I think symlink gcc probably won't work here >>>>> because libc and includes are also going to be wrong, unlike the python >>>>> case where the symlinked python is only used to run the hashdist script. >>>>> Shouldn't dolfin cmake choose the compiler following some priority like >>>>> mpic++>g++>c++ ? >>>>> >>>> >>>>> On Tue, Dec 9, 2014 at 11:16 AM, Jan Blechta < >>>>> [email protected]> wrote: >>>>> >>>>>> Building mpi4py fails on Ubuntu 12.04 because of some problem with >>>>>> system Cython (see below for full log) >>>>>> >>>>>> ImportError: /usr/lib/pymodules/python2.7/Cython/Compiler/Code.so: >>>>>> undefined symbol: PyUnicodeUCS4_DecodeUTF8 >>>>>> >>>>>> Shouldn't be Cython built also? >>>>>> >>>>>> Jan >>>>>> >>>>>> >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] running install >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] running build >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] running build_src >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] cythonizing >>>>>> 'mpi4py.MPI.pyx' -> 'mpi4py.MPI.c' 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] Traceback (most recent call last): 2014/12/09 >>>>>> 18:02:56 - INFO: [package:run_job] File "setup.py", line 624, in >>>>>> <module> 2014/12/09 18:02:56 - INFO: [package:run_job] main() >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] File "setup.py", line >>>>>> 621, in main 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> run_setup() 2014/12/09 18:02:56 - INFO: [package:run_job] File >>>>>> "setup.py", line 493, in run_setup 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] **metadata) 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/tmp/mpi4py-dps5vsljzl7g-1/conf/mpidistutils.py", >>>>>> line 441, in setup 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> return fcn_setup(**attrs) 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/core.py", >>>>>> line 151, in setup 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> dist.run_commands() 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/dist.py", >>>>>> line 953, in run_commands 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] self.run_command(cmd) 2014/12/09 18:02:56 - >>>>>> INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/dist.py", >>>>>> line 972, in run_command 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] cmd_obj.run() 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/command/install.py", >>>>>> line 563, in run 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> self.run_command('build') 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/cmd.py", >>>>>> line 326, in run_command 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] self.distribution.run_command(command) >>>>>> 2014/12/09 >>>>>> 18:02:56 - INFO: [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/dist.py", >>>>>> line 972, in run_command 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] cmd_obj.run() 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/command/build.py", >>>>>> line 127, in run 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> self.run_command(cmd_name) 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/cmd.py", >>>>>> line 326, in run_command 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] self.distribution.run_command(command) >>>>>> 2014/12/09 >>>>>> 18:02:56 - INFO: [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/bld/python/ctzx2dm36c6d/lib/python2.7/distutils/dist.py", >>>>>> line 972, in run_command 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] cmd_obj.run() 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File "setup.py", line 594, in build_sources >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] wdir='src', >>>>>> force=cmd.force, VERSION=CYTHON) 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File "setup.py", line 573, in run_cython >>>>>> 2014/12/09 >>>>>> 18:02:56 - INFO: [package:run_job] wdir=wdir) 2014/12/09 18:02:56 >>>>>> - >>>>>> INFO: [package:run_job] File >>>>>> >>>>>> "/srv/groot/blechta/.hashdist/tmp/mpi4py-dps5vsljzl7g-1/conf/cythonize.py", >>>>>> line 9, in cythonize 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> from Cython.Compiler.Main import \ 2014/12/09 18:02:56 - INFO: >>>>>> [package:run_job] File >>>>>> "/usr/lib/pymodules/python2.7/Cython/Compiler/Main.py", line 18, in >>>>>> <module> 2014/12/09 18:02:56 - INFO: [package:run_job] import Code >>>>>> 2014/12/09 18:02:56 - INFO: [package:run_job] >>>>>> ImportError: /usr/lib/pymodules/python2.7/Cython/Compiler/Code.so: >>>>>> undefined symbol: PyUnicodeUCS4_DecodeUTF8 2014/12/09 18:02:56 - >>>>>> ERROR: >>>>>> [package:run_job] Command '[u'/bin/bash', '_hashdist/build.sh']' >>>>>> returned non-zero exit status 1 2014/12/09 18:02:56 - ERROR: >>>>>> [package:run_job] command failed (code=1); raising >>>>>> _______________________________________________ >>>>>> fenics mailing list >>>>>> [email protected] >>>>>> http://fenicsproject.org/mailman/listinfo/fenics >>>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> Lizao (Larry) Li >>>>> Univeristy of Minnesota >>>>> >>>>> _______________________________________________ >>>>> fenics mailing list >>>>> [email protected] >>>>> http://fenicsproject.org/mailman/listinfo/fenics >>>>> >>>>> >>>> >>>> >>>> -- >>>> Simon Funke >>>> >>>> Imperial College London >>>> Applied Modelling and Computation Group >>>> Office: 3.56R Royal School of Mines Building >>>> Tel: +44 (0)20 759 41363 >>>> Email: [email protected] >>>> >>> >>> _______________________________________________ >>> fenics mailing list >>> [email protected] >>> http://fenicsproject.org/mailman/listinfo/fenics >>> >>> > _______________________________________________ > fenics mailing list > [email protected] > http://fenicsproject.org/mailman/listinfo/fenics > > -- Simon Funke Center for Biomedical Computing Simula Research Laboratory Oslo, Norway Email: [email protected]
_______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
