Python3

Porting to Python 3

As of version 1.4, beets supports both Python 3.x and 2.x. Sometime soon, we'll drop support for 2.x.

Coding conventions

• calls to map() should be wrapped in list() when a list is expected as it returns an iter in python 3

PY3 initial support TODO

• add python 3 to trove classification
• check documentation for 2.x, 3.x references
• check wiki for 2.x, 3.x references

PY3 only TODO

• remove u' (and u") prefixes from code
• remove u' (and u") prefixes from documentation
• remove u' (and u") prefixes from wiki
• delete https://github.com/beetbox/beets/wiki/String-literals page
• run 2to3 to make sure we got everything
• audit all calls to open() (or any other I/O) to see if we can read/write them as text vs binary
• remove any PY2/3 version checks and the related code
• remove py2 specific tests and packages in .travis.yml
• remove any usage of six
• remove # -*- coding: utf-8 -*- from all python files
• replace buffer with memoryview
• replace codecs.open, io.open with open
• use explicit keyword arg names in function definitions - PY2: def foo(.., **kwargs) -> PY3: def foo(.., arg1=None, arg2=None, ...) as per https://docs.python.org/3.3/tutorial/controlflow.html#arbitrary-argument-lists
• use str.format style ({}) in beets.logger or drop it completely https://docs.python.org/3/whatsnew/3.2.html#logging
• Remove F405 from flake8 ignore configurtion if beets.logging is replaced
• change flake8-coding configuration to no-accept-encodings
• use more 3.x only functions from shutil to replace some of the functions in beets.util https://docs.python.org/3.4/library/shutil.html (TODO: be more specific)
• drop mock dep since it's part of the stdlib (py >= 3.3)
• replace deprecated inspect.getargspec
• Remove unnecessary traceback passing in HumanReadableException