01_compile_cpython.md

Compile CPython

Note

The CPython development guide provides a very detailed guide to build CPython for all the platforms that are supported.

Feel free to check it out for more context.

Compiling Python from its source might sound like a very complicated process, and many people don't even try it due to that. Hopefully, after this tutorial, you will manage to overcome this and do it yourself!

You learned in the previous exercise that you can compile one C-file with a simple command line, but in the case of large projects, you need the help of other tools that can simplify the process of compiling each file and creating the final executable for you.

In the case of CPython, we will be using a Makefile (for Linux and macOS) and a custom bat script for Windows. Due to the process being different, we will separate the explanation.

For Linux and macOS

Before building the project, we need to first configure it. For that, you need to run:

./configure --with-pydebug

This shouldn't take a lot of time, between 20 seconds to a couple of minutes, depending on your machine resources, and you will see a lot of things happening in the terminal, but don't worry.

Note

--with-pydebug builds a debug version of CPython. It's slower to run, but it adds extra assertions and reference-counting checks that will help you catch mistakes when you start editing the interpreter's C code in the next exercises.

In case a crucial dependency is not found, you might need to install it and re-run the command. More info in the official dev guide

Tip

Building CPython can take a while (several minutes, even with many CPU cores). If you're following along in a workshop, start this step as soon as you can instead of waiting until the presentation ends — you can keep listening while it builds in the background.

Once the configuration is done, you can now run the make command to build the project. This might take time, so it's recommended for you to enable more CPU cores from your system to use while building with the -j option. Assuming you have 16 cores, you could run:

make -j 16

Change the number accordingly.

Once this process is over, you will see a python binary that you can access with ./python

Important

The ./ is very important to indicate you want to run the python you just built, and not a system one.

For Windows

There is an ad-hoc script in order to build CPython, that you can use to avoid many complicated steps. On the root of the repository, execute the following command:

PCBuild\build.bat

You will see a lot of output, and if everything goes smoothly, you should see a message with the final executable in PCbuild\amd64\python_d.exe.