Coding Guidelines

Getting started

Before you can start making changes to Quod Libet you have to set up a development environment. See “Creating a Development Environment” for further details.

Source Overview


Things in the View menu


Extensions to QL / EF (i.e. the plugins)


File format support


Library management code


Base classes and structural enabling plugins


Operon, the QL CLI tool


GTK+ widget subclasses/extensions


General utility functions and setup code

If you want to get a full overview of QL’s code, good places to start are browsers/, formats/, and library/

Code Guidelines

We try to keep Quod Libet’s code in pretty good shape; when submitting a patch, it’s much easier to get it included quickly if you run through this checklist of common-sense code quality items. Make sure your patch:

  • Passes existing tests. You can test this by executing ./ test

  • Is commented.

  • Adds your name to the copyright header of every file you touch. This helps you get credit and helps us keep track of authorship.

General Guidelines

We prefer Python to C. We prefer ctypes to compiled C wrappers. A good way to get a new feature applied is if you include tests for it. Stock strings and string reuse are awesome, but don’t make the interface awkward just to avoid a new string.

Unit Tests

Quod Libet comes with a lot of tests, which helps us control regression. To run them, run ./ test. Your patch can’t break any unit tests, and if you change tests in a non-obvious way (e.g. a patch that removes an entry point and also removes a test for it is obvious) you should explain why.

It’s possible, indeed encouraged, that a changeset was for no other purpose than to improve the testing / test coverage, as there have been plenty of bugs that have slipped through. As usual, any fix associated with a confirmed bug should include tests that would have found the original bug, where possible.

Printing Text

All terminal output should go through the print_, print_w, print_e, or print_d functions. These will handle Unicode recoding. They also let us capture all output for debugging purposes.


All text that could be visible to users (with debugging mode disabled) should be marked translatable.

You can do this by simply using the _ function which is globally available (through __builtin__):

print_w(_("This is translatable"))

To handle plural forms use ngettext:

text = ngettext("%d second", "%d seconds", time) % time

It is good practice to add a comment for translators if the translation depends on the context:

# Translators: As in "by Artist Name"
text = _("by %s") % tag