Pandas and SciPy Python libraries needs glibc headers in order to be installed
through pip. In latest Alpine distributions, the needed glibc header has been renamed.
Furthermore, the latest NumPy release (1.14.4) breake the installation of Pandas.
At the moment, the PyPI community didn't fixed this stuff, so this is a repo to build
a Docker image based on Alpine containing the workaround for the installation of
NumPy (1.14.3), Pandas (0.23.0) and scikit-learn (0.19.2):
- proper link to the needed
glibcheader - preventive build of
Pandasdependencies in order to avoid the latestNumPyversion
As parent image, we considered tiangolo/uwsgi-nginx-flask:python2.7-alpine3.7
(https://github.com/tiangolo/uwsgi-nginx-flask-docker/tree/master/python2.7-alpine3.7), so
the resulting image will host a Python 2.7 Flask application based on uWSGI and Nginx with
scientific libraries installed.