Simple library checking with SCONS

This collection of library checks is provided as a convenience helper. It offers the possibility to perform a platform independent check for the library (including compilation and execution) and automatically adds command line configuration options for it. It works with Python 2 and 3 and is tested on Windows and Linux.

Usage

Every library has an assigned library id string. Currently, checks are implemented for:

• 'boost.datetime'
• 'boost.filesystem'
• 'boost.interprocess'
• 'boost.numpy',
• 'boost.preprocessor',
• 'boost.python',
• 'boost.serialization',
• 'boost.system',
• 'boost.test',
• 'boost.thread',
• 'cuda',
• 'eigen',
• 'fftw',
• 'hdf5',
• 'matlab',
• 'openblas'
• 'numpy',
• 'opencv',
• 'python',
• 'protobuf',
• 'swig'.

You can either:

import SConsChecks
# call checks directly
SConsChecks.CheckNumPy()

or:

# import these convenience methods,
from SConsChecks import AddLibOptions, GetLibChecks
# define a list of required checks,
_libs = ['boost.numpy',
         'boost.preprocessor',
         'boost.python',
         'boost.serialization',
         'boost.test',
         'boost.thread',
         'python',
         'numpy',
         'opencv',
         'eigen']
# and get a check (name, method) dict.
_checks = GetLibChecks(_libs)

# You can automatically add all command line options to configure the
# libraries as follows:
def setupOptions():
    # ...
    # Add library configuration command line options.
    AddLibOptions(AddOption, _libs)

See the attached sample project for details.

Default library configuration

The following defaults are used to configure library xyz:

• Environment variable XYZ_ROOT (command-line --xyz-dir): the libraries root folder. If specified, the following include and lib directories are used relative to the root directory.

• Environment variable XYZ_INCLUDE_DIR (command-line --xyz-inc-dir): the libraries header include folder.

• Environment variable XYZ_LIB_DIR (command-line --xyz-lib-dir): the libraries linking folder.

Contributing

You are very welcome to extend the list of available checks. Please fork the repo and send me a pull request.

License and credits

This project is heavily inspired by the build system of boost numpy (see https://github.com/ndarray/Boost.NumPy/blob/master/SConscript), which hass been created by Jim Bosch.

The source code is distributed under the Boost Software License, Version 1.0 (see http://www.boost.org/LICENSE_1_0.txt).