This doc covers all the steps you need to take to prepare the perfect Python environment to work on Agent based integrations, from installing the interpreter to install and the dependencies needed.
Integrations are supposed to be deployed in the Datadog Agent's Python embedded environment, that happens to be version 2.7.14 at the moment (you can check this info out here. To try to locally reproduce the behaviour a check would have in production, we recommended to use the same version of the Agent but we'll eventually move to Python3, so having code capable to run with both versions can't hurt.
Most operating systems come with Python already installed so chances are you might not need to do anything.
In case your Python is too old or if you don't have one, please see the following instructions for your Operating System.
Any recent version of MacOS comes with some Python installed that might be older than the version used in the Agent and might also lack of some pieces in terms of tools and dependencies that you might need. For these reasons we recommend to install a fresh, dedicated Python interpreter you can manage without the App Store, following either of the methods listed in this paragraph.
Homebrew[http://brew.sh/#install] is a package manager for macOS that makes a
lot easier installing software on macOS, specially from the command line. There's
already an awesome guide about how to install Python with Homebrew in
the Hitchhiker’s Guide to Python
we recommend to read.
Miniconda is the lightweitgth version of Anaconda(http://anaconda.com), a Python
distribution specifically designed for data processing and scientific computing.
Miniconda maintains the awesome Conda package manager and provides a full fledged
Python enviroment along with development libraries and a tool for managing v
irtual environments, all without cargo loading any library or package you'd find
in Anaconda.
You can download miniconda from here
and install following these instructions.
Miniconda is extremely self contained at the point you uninstall it with rm -r
and might be a good option if you don't want/need a full fledged Python enviroment
installed systemwide, or if you just want to give Python a spin.
TODO
TODO
Each integration has its own set of dependencies that must be added to Python in order to run the tests or just try out the collection code; to avoid polluting your Python installation with libraries and packages that would only be used by an integration, you can use the so called "virtual enviroments". A virtual environment is a self contained directory tree that contains an isolated Python installation - when a virtual enviroment is active, any package you install goes into that directory, without hitting the system wide Python installation tree.
We recommend using virtualenv to manage virtual Python environments and virtualenvwrapper to make the process smoother. There's a comprehensive guide in the Hitchhiker’s Guide to Python describing how to setup these two tools.
If you're using Miniconda, a tool to manage virtual environments is already there. Please refer to the official guide for how to use.