Conda

CMI Python Meeting

27/3/15

Anaconda

  •  Anaconda is a distribution of Python developed and maintained by Continuum Analytics.
  • It bundles over 195 scientific python libraries as binaries that are very difficult to build, such as MayaVi
  • It also provides the conda environment and package manager.

How to install Anaconda

Miniconda

  • Whilst installing large numbers of packages may be useful, some may only want the bare bones.
  • Miniconda is Anaconda, without the packages.

How to install Miniconda

Using Conda

  • Conda is both a package (like pip) and environment (like virtual envs) manager
  • However, it is designed to manage the versions of C and C++ extensions, to prevent incorrect library linking.

Conda as an environment Manager

Create a conda environment: 

conda create -n {env_name} python={2/3} {packages}

For example:

conda create -n py3k python=2 rdkit

Changing environments

source activate py3k
source deactivate

Switch to an environment:

Exit an environment:

conda env list

List environments:

As a package manager

Install a new package:

conda install {package}

Uninstall a package:

conda remove {package}

Update a package:

conda update {package}

List installed packages

conda list

Works well with pip

pip install requests
conda list

List installed packages

conda list

Doesn't really work well with virtualenvs and virtualenvwrapper :(

Why it's better than virtualenv?

conda create -n py3k python=3 boost=1.56 rdkit
conda create -n py2k python=2 boost=1.56 rdkit

RDKit now works with both Python 2 and Python 3!

This is incredibly difficult without conda.

Made with Slides.com