Python White Magic

Python is a powerful language. Beginners appreciate its surface of simplicity and ease to use, where (almost) everything “just works” (TM) as expected. However, under the hood, there is a whole world of rules and layers which can be (ab)used by experienced programmers to tweak the language in unexpected ways.

Often, these techniques are labeled as “hacks” or “black magic”, to indicate that they should not used in production or that you should not look at it unless you are “initiated” to some not better specified sorcery.

However, if used with care and in the right places, these techniques can lead to better, faster and/or more readable code, or can be extremely useful during debugging, and thus deserve the definition of “white magic” which programmers can use to improve their code. Moreover, looking at how they are implemented is a good way to learn about some of the deep corners of the language.

This talk presents some of these advanced techniques which I have seen in use in real code, e.g. in PyPy or pdb++. Including, but not limited to: creating new functions by reusing existing code objects, instantiating multiple copies of the same module, playing with metaclasses in various ways, changing the __class__ of an object on the fly to get specialized behavior, automatically entering the debugger when a certain event occur, etc.