Warning

This documentation is for an unreleased version of MPF!

This is the developer documentation for MPF 0.54, which is the “dev” (next) release of MPF that is a work-in-progress. Use the “Read the Docs” link in the lower left corner to view the developer docs for the version of MPF you’re using.

Setting up your MPF Dev Environment

If you want to work on the core MPF or MPF-MC code, you have to install MPF and MPF-MC a bit differently than the normal process.

Why? Because normally when you install MPF and MPF-MC via pip, they get installed as Python packages into your Python/Lib/site-packages folder, and that location is not too conducive to editing MPF source code since it’s in a deep random location. Also, if you ever ran pip again to update your MPF installation, you would potentially overwrite any changes you made.

Instead, you need to install MPF and MPF-MC in “developer” (also known as “editable”) mode. This mode will let you run MPF and MPF-MC from the folder of your choice, and will allow code changes or additions you make to be immediately available whenever you run MPF.

1. Install a git client

MPF is cross-platform and runs the same on Mac, Windows, or Linux. So any changes or additions you make should work on all platforms.

If you’re on Windows or Mac, the easiest way to get a git client installed is to use the GitHub Desktop app. This app will also install the git command line tools.

2. Clone the MPF and/or MPF-MC repo(s)

Clone the mpf repository and its submodules :

git clone --recursive https://github.com/missionpinball/mpf.git

Same thing for the mpf-mc repository :

git clone --recursive https://github.com/missionpinball/mpf-mc.git

If you’re using the GitHub Desktop app, you can also browse to the repos on GitHub and click the green “Clone or Download” button, and then click the “Open in Desktop” link. That will pop up a box that prompts you to pick a folder for the local codebase.

Then inside that folder, you’ll end up with an mpf folder for MPF and mpf-mc folder for MPF-MC.

3. Install MPF / MPF-MC in “developer” mode

Create a “virtualenv” for your MPF development in a mpf-env directory (Note : if you don’t have virtualenv installed, you can get it via pip by running pip3 install virtualenv.

Using virtualenv lets you keep all the other Python packages MPF needs (pyserial, pyyaml, kivy, etc.) together in a “virtual” environment that you’ll use for MPF and helps keep everything in your Python environment cleaner in general.

Create a new virtualenv called “mpf-venv” (or whatever you want to name it) like this:

virtualenv -p python3 mpf-venv

Then enter the newly-created virtualenv:

source mpf-venv/bin/activate

On Macs, you will need to uninstall and reinstall kivy in the the virtual envirment to avoid ambiguous kivy library errors.

1) pip uninstall kivy
2) pip install kivy -no-binary :all:

Each time you’ll work with your MPF development version you’ll have to switch to this environment.

source mpf-venv/bin/activate

Note: in this environment, thanks to the “-p python3” option of virtualenv, the version of Python and pip is 3.x automatically.

Next you’ll install MPF and MPF-MC. This is pretty much like a regular install, except that you’ll also use the -e command line option which means these packages will be installed in “editable” mode.

Install mpf and mpf-mc like this:

pip install -e mpf
pip install -e mpf-mc

You should now be done, and you can verify that everyething is installed properly via:

mpf --version

Note : you could also install mpf and mpf-mc in your global environment using sudo pip3 install -e mpf and sudo pip3 install -e mpf-mc, or in your user environment using pip3 install --user -e mpf and pip3 install --user -e mpf-mc.

4. Make your changes

Be sure to add your name to the AUTHORS file in the root of the MPF or MPF-MC repo!

5. Write / update unit tests

We make heavy use of unit tests to ensure that future changes don’t break existing functionality. So write new unit tests to cover whatever you just wrote, and be sure to rerun all the unit tests to make sure your changes or additions didn’t break anything else.

More information on creating and running MPF unit tests is here.

6. Submit a pull request

If your change fixes an open issue, reference that issue number in the comments, like “fixes #123”.