Potential pitfalls when going from old-style to new-style classes

Chris Kaynor ckaynor at zindagigames.com
Wed Nov 19 20:34:44 CET 2014


On Wed, Nov 19, 2014 at 11:14 AM, Skip Montanaro <skip.montanaro at gmail.com>
wrote:

> I discussion on the code-quality list got me thinking. Suppose I have
> an old-style class in a 2.x app:
>
> class Foo:
>   def __init__(self):
>     blah blah blah
>
> I still use 2.x exclusively, but anytime I run pylint over a bit of
> code and it complains that Foo is old-school, I make the obvious
> change to inherit from object. While I do this mostly to shut up
> pylint so I can focus on more serious messages, I must admit I have
> never experienced any problem changing from old-style to new-style
> classes. Under what circumstances might this be problematic?
>

Under most common cases, its as trivial as just subclassing object, however
there are some oddities:

   - The main case errors may pop up is with multiple inheritance (MRO was
   changed).
   - If you are accessing or have defined magic methods.
      - For example, overriding __call__ on a specific instance will no
      work with new-style classes, but will work with old-style
classes. Defining
      it on the class (the common case) works the same in both cases.
      - A number of new magic properties were added, __slots__ and __mro__.
      So if those are being defined (they shouldn't - all dunder names are
      reserved by Python) it could change behavior oddly.

It looks like
http://stackoverflow.com/questions/54867/old-style-and-new-style-classes-in-python
has some decent answers to your question.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-list/attachments/20141119/716122cd/attachment.html>


More information about the Python-list mailing list