Compiling Python from source

Things to consider

Before you compile and install Python there are a few things you should know and/or consider:

Unicode

Python has a long and complicated history when it comes to Unicode support. Unless you have very specific reasons you should configure Python 2.7 to enable UTF-32 support. This increases memory usage but improves compatibility. In Python 3.3+ the Unicode support has been completely rewritten and strings are automatically stored using the most efficient encoding possible.

You enable UTF-32 in Python 2.7 by passing --enable-unicode=ucs4 to the configure command.

Shared library

You should compile Python as a shared library by passing --enable-shared to the configure command. All modern Linux distros ship with Python compiled as a shared library. It reduces memory usage if more than one Python process is running, and there are third-party tools that might not work properly without it. To make sure the executable can find its shared library you also need to pass some additional flags to the configure command (LDFLAGS="-Wl,-rpath /usr/local/lib").

If you do not have sudo or root access you will probably not be able to compile Python as a shared library. If someone knows how to solve this please leave a comment below and I will update this text with instructions.

Use “make altinstall” to prevent problems

It is critical that you use make altinstall when you install your custom version of Python. If you use the normal make install you will end up with two different versions of Python in the filesystem both named python. This can lead to problems that are very hard to diagnose.

Preparations – install prerequisites

In order to compile Python you must first install the development tools and a few extra libs. The extra libs are not strictly needed to compile Python but without them your new Python interpreter will be quite useless.

Execute all the commands below as root either by temporarily logging in as root or by using sudo.

Download, compile and install Python

Here are the commands to download, compile and install Python.

# Python 2.7.14: wget http://python.org/ftp/python/2.7.14/Python-2.7.14.tar.xz tar xf Python-2.7.14.tar.xz cd Python-2.7.14 ./configure --prefix=/usr/local --enable-unicode=ucs4 --enable-shared LDFLAGS="-Wl,-rpath /usr/local/lib" make && make altinstall # Python 3.6.3: wget http://python.org/ftp/python/3.6.3/Python-3.6.3.tar.xz tar xf Python-3.6.3.tar.xz cd Python-3.6.3 ./configure --prefix=/usr/local --enable-shared LDFLAGS="-Wl,-rpath /usr/local/lib" make && make altinstall

After running the commands above your newly installed Python interpreter will be available as /usr/local/bin/python2.7 or /usr/local/bin/python3.6. The system version of Python 2.6.6 will continue to be available as /usr/bin/python, /usr/bin/python2 and /usr/bin/python2.6.

You might also want to strip symbols from the shared library to reduce the memory footprint.

# Strip the Python 2.7 binary: strip /usr/local/lib/libpython2.7.so.1.0 # Strip the Python 3.6 binary: strip /usr/local/lib/libpython3.6m.so.1.0

Install/upgrade pip, setuptools and wheel

Each Python interpreter on your system needs its own install of pip, setuptools and wheel. The easiest way to install or upgrade these packages is by using the get-pip.py script.

# First get the script: wget https://bootstrap.pypa.io/get-pip.py # Then execute it using Python 2.7 and/or Python 3.6: python2.7 get-pip.py python3.6 get-pip.py # With pip installed you can now do things like this: pip2.7 install [packagename] pip2.7 install --upgrade [packagename] pip2.7 uninstall [packagename]

The packages will end up in /usr/local/lib/pythonX.Y/site-packages/ (where X.Y is the Python version).

What's next?

If you are using Python 2.7 I strongly recommend that you install virtualenv and learn how to use it. Virtualenv makes it possible to create isolated Python environments. If you are using Python 3.3+ then you don't need virtualenv because that functionality is already built in.

Each isolated Python environment (also called sandbox) can have its own Python version and packages. This is very useful when you work on multiple projects or on different versions of the same project.

Create your first isolated Python environment