Build Process#

The py5 library is different from the typical Python library in that the code itself is the output of the meta-programming project py5generator. This approach is necessary because of the size, depth, and complexity of Processing.


You will need to download and install Anaconda. You’ll also need build tools such as make and ant, as well as the source control program git.

Clone Source Repository from GitHub#

You will need to clone the py5generator repo from GitHub, or better yet, fork the py5generator repo and clone your fork. You’ll need to create a fork if you intend to make a pull request.

If you don’t already have the Processing 4 application on your computer, you’ll need to download that too.

On my computer I keep all of the py5 repos together in one directory.

mkdir pythonprocessing
cd pythonprocessing
git clone

Install Java 17#

You will need to have Java 17 installed on your computer and available on your path. Relevant information about how to install the correct version is available on the py5 install page.

Create Anaconda Environment#

Next you will need to create the Anaconda environment. This will install Python packages that go beyond what is necessary for regular py5 use.

cd py5generator
conda env create -f environment.yml
conda activate py5

The conda activate py5 command will activate the py5 Anaconda environment. You will see (py5) in the terminal prompt to indicate that you are working in this new environment.

If you have difficulties creating the Anaconda environment, try setting the channel priority to strict with this command:

conda config --set channel_priority strict

Locate Processing Application#

When the build process runs, the code will search for and copy jars from the Processing PDE application directory into a specific location in the destination directory. It is important to download the correct (latest) version of the Processing application. On my machine, I keep the Processing PDE application in ~/INSTALL/processing-4.

The Java classes need to be compiled with debug information because the meta-programming code parses the output of javap to inspect the Processing jars and gather information about the relevant public fields and methods. Currently Processing jars are compiled with debug information enabled. If this were to ever change, the build process would need to change to adapt.

Run the py5 Makefile#

Finally, run the Makefile. The processing_dir parameter needs to point to the location of the Processing application. The py5_build_dir parameter is where the generated code will go.

make processing_dir=~/INSTALL/processing-4 py5_build_dir=../py5code skip_black=true

The Makefile uses realpath. You may need to install the “coreutils” package if you don’t have it on your machine already.

The optional skip_black argument will skip black formatting of the output code, accelerating the build process. Use this during development.

A full build takes under a minute to complete. When this runs you’ll see a lot of logging information appear on the screen. There will be a few warnings about skipping typehints for some Matrix functions, but everything else will be info messages. You should familiarize yourself with the logs so you can spot changes and identify when something goes wrong.

The final Makefile step installs the generated code into the active Anaconda environment. At this point the new code will be available for you to use.

When the build process detects changes to Processing’s methods and fields, a message will appear in the logs. This process is designed to adapt as the Processing library evolves. Be aware that signature changes to known methods will result in automatic updates to documentation files in the py5_docs/Reference/api_en directory.

There is an additional make command for generating reference documentation for the website.

make generate_py5_docs py5_website_dir=../py5book py5_api_lang=api_en