Prehistorical Python: Patterns past their prime

by Lennart Regebro

There are many idioms and patterns that used to be a best practice but isn’t anymore, thanks to changes in Python. Despite that they often show up even in new code, and some of these patterns are even explained to be Good Ideas at stackoverflow and similar.

This talk will bring out the most common of these patterns so you know what they are, and why you should avoid them.


# python 2.5
from collections import defaultdict
data = defaultdict()
data[key] = value

# Python 2.5-
# Exists still in Django 1.5.1
# django/db/models/sql/
if key in data:
    data[key] = set([value])


  • Unique values
  • Unordered
  • Fast lookup
  • Python built-in in 2.4

Sets before sets

d = {}
for each in list_of_things:
    d[each] = None

list_of_things = d.keys()


# new way - missed the old ways
retval = set()
for tn in template_nmes:
    retval.update(search_python(python_code, tn))
retval = sorted(retval)

Conditional Expressions

# old way
# django 1.5.1 django/db/models/
first_choice = include_blank and blank_choice = []

# new way
first_choice = blank_choice if include_blank else []

Constants and Loops

# outside vs inside
# PyPy is 33x slower on this one!
each * 5 ** a_var


Thought: Evaliate your constants outside the loop

String Concatenation

# The 'fastest' way
self._leftover = b''.join([bytes, self._leftover])

adding is faster than using the .join() method used above. WTF?!?

Explaining the ‘WTF?!?’

  • Looping over a list of strings and adding them together is slow.
  • Using .join with a list of strings is fast.
  • If you add just two strings, adding them is faster.