Skip to content

Latest commit

 

History

History
132 lines (93 loc) · 4.68 KB

README.rst

File metadata and controls

132 lines (93 loc) · 4.68 KB
Raw
https://travis-ci.org/diffpy/pyobjcryst.svg?branch=master http://codecov.io/github/diffpy/pyobjcryst/coverage.svg?branch=master

pyobjcryst

Python bindings to ObjCryst++, the Object-Oriented Crystallographic Library.

The documentation for this release of pyobjcryst can be found on-line at http://diffpy.github.io/pyobjcryst.

REQUIREMENTS

pyobjcryst requires Python 2.7, C++ compiler and the following software:

  • libobjcryst - Object-Oriented Crystallographic Library for C++, https://github.com/diffpy/libobjcryst
  • setuptools - tools for installing Python packages
  • NumPy - library for scientific computing with Python
  • scons - software constructions tool (1.0 or later)
  • python-dev - header files for interfacing Python with C
  • libboost-all-dev - Boost C++ libraries and development files

We recommend to use Anaconda Python as it allows to install all software dependencies together with pyobjcryst. For other Python distributions it is necessary to install the required software separately. As an example, on Ubuntu Linux the required software can be installed using

sudo apt-get install \
   python-setuptools python-numpy scons \
   build-essential python-dev libboost-all-dev

INSTALLATION

The preferred method is to use Anaconda Python and install from the "diffpy" channel of Anaconda packages

conda config --add channels diffpy
conda install pyobjcryst

pyobjcryst is also included in the "diffpy-cmi" collection of packages for structure analysis

conda install diffpy-cmi

If you prefer to use other Python distribution or install from sources, you must first install the libobjcryst library as per the instructions at https://github.com/diffpy/libobjcryst. Make sure other required software is also in place and then run:

python setup.py install

You may need to use sudo with system Python so the process is allowed to copy files to system directories. If administrator (root) access is not available, see the usage information from python setup.py install --help for options to install to a user-writable location. The installation integrity can be verified by changing to the HOME directory and running

python -m pyobjcryst.tests.run

An alternative way of installing pyobjcryst is to use the SCons tool, which can speed up the installation by compiling C++ files in several parallel jobs (-j4):

scons -j4 install

See scons -h for description of build targets and options.

DEVELOPMENT

pyobjcryst is an open-source software developed as a part of the DiffPy-CMI complex modeling initiative at the Brookhaven National Laboratory. The pyobjcryst sources are hosted at https://github.com/diffpy/pyobjcryst.

Feel free to fork the project and contribute. To install pyobjcryst in a development mode, where its sources are directly used by Python rather than copied to a system directory, use

python setup.py develop --user

When developing it is preferable to compile the C++ files with SCons using the build=develop option, which compiles the extension module with debug information and C-assertions checks

scons -j4 build=debug develop

The build script checks for a presence of sconsvars.py file, which can be used to permanently set the build variable. The SCons construction environment can be further customized in a sconscript.local script. The package integrity can be verified by executing unit tests with scons -j4 test.

When developing with Anaconda Python it is essential to specify header path, library path and runtime library path for the active Anaconda environment. This can be achieved by setting the CPATH, LIBRARY_PATH and LDFLAGS environment variables as follows:

# resolve the prefix directory P of the active Anaconda environment
P="$(conda info --json | grep default_prefix | cut -d\" -f4)"
export CPATH=$P/include
export LIBRARY_PATH=$P/lib
export LDFLAGS=-Wl,-rpath,$P/lib
# compile and re-install pyobjcryst
scons -j4 build=debug develop

On Mac OS X the distributed Anaconda packages are built for operating system version 10.6, which may be incompatible with codes compiled on a newer OS. To avoid this problem set the environment variable MACOSX_DEPLOYMENT_TARGET=10.6. This allows to build pyobjcryst against the Anaconda package for the libobjcryst library.

CONTACTS

For more information on pyobjcryst please visit the project web-page

http://www.diffpy.org

or email Prof. Simon Billinge at sb2896@columbia.edu.