This is a Heroku Buildpack for Conda, the Python distribution for scientific computing by Continuum Analytics. It is based on the lightweight Miniconda distribution, keeping the final slug size as small as possible.
To control the packages installed by conda, supply a
requirements-conda.txt file (which can be created by
capturing the output of
conda list -e for your working conda environment). Note that you MUST specify
libgfortran and libopenblas explicitly in this file for numpy to work properly. See included requirements-conda.txt, which
can be used as a template.
All other packages can then be installed with a
requirements-pip.txt file for pip
to process. In this way, you can install binary packages via Miniconda for everything you can and
still use pip for anything you can't.
If you're on a Mac and want to test the scripts locally like I did, you'll need to make your local environment as close as possible to a GNU Linux distribution using Homebrew with the following commands:
Fix issues with "date": brew install coreutils Update your path with PATH="/usr/local/opt/coreutils/libexec/gnubin:$PATH"
Fix issues with "sed": brew install gnu-sed --with-default-names
Fix issues with "find": brew install findutils --with-default-names
Copy the snippet above into CLI.