Skip to content

Base package for datamicroscopes, a non-parametric bayesian inference library

License

Notifications You must be signed in to change notification settings

vishalbelsare/common

 
 

Repository files navigation

microscopes-common Build Status

The base package.

Installation

The only officially supported mechanism is to use Anaconda.

$ conda install -c datamicroscopes -c distributions microscopes-common

We realize that forcing a particular package/environment manager may be offensive to some. But Anaconda is one of the best solutions out there for distributing complicated python libraries which depend on many C/C++ libraries. Who wants to deal with library version mismatch issues, or messing around with LD_LIBRARY_PATH?

This is probably the closest to "one-click" install that we can achieve in the foreseeable future.

Building

If you simply want to use datamicroscopes, then see the above instructions on installation. If you want to actively contribute, then follow these steps for building. Our build process relies on anaconda to take care of the dependencies, and then pip to install our modified libraries.

Requirements

Note that a compiler with support for C++11 is required for building. On Linux, this means g++ >= 4.8 (4.6 supports C++11, but its headers are somewhat broken for <random> which we rely on). On OS X, this means the latest version of XCode. Currently, we only support OS X 10.7 or greater.

Conflict with system libprotobuf

Currently, our CMakeLists.txt does not give precedence to Anaconda's version of libprotobuf but instead prefers the system one if present. The easiest workaround for the time being is to uninstall the system versions. This will be fixed soon.

Setting up the anaconda environment

We recommend you to not work in the default (root) environment

$ conda create -n myenv anaconda python=2
$ source activate myenv
$ conda install -c distributions distributions
$ conda install -c datamicroscopes eigen3
$ conda install cython cmake
$ export DYLD_LIBRARY_PATH=/path/to/anaconda/envs/myenv/lib # or LD_LIBRARY_PATH on linux

Note that on OS X, instead of DYLD_LIBRARY_PATH it is also possible to specify DYLD_FALLBACK_LIBRARY_PATH. This fixes some issues with using libpng (and consequently matplotlib) in conjunction with the datamicroscopes library.

Now if you want nice git SHA1 hashes for package versions (useful for dev), install gitpython. Note this step is optional.

$ pip install gitpython==0.3.6

Building/testing the C++ library

Now that the conda environment is set up, use cmake to build the C++ library. Enter the commands below, replacing debug with either relwithdebinfo or release depending on the type of build. Note the first invocation to make simply invokes cmake with the right prefixes set up.

$ make debug # calls cmake
$ cd debug
$ make
$ make test
$ make install # installs to prefix /path/to/anaconda/envs/myenv
$ cd ..

Building/testing the python library

Now use pip to install the python library. Either

$ pip install .
$ cd test && nosetests -v

or alternatively to build in the source tree:

$ pip install -e .
$ nosetests -v

About

Base package for datamicroscopes, a non-parametric bayesian inference library

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C++ 54.6%
  • Python 41.3%
  • CMake 2.2%
  • Other 1.9%