Click here to code!

Learning from Python mistakes

Posted by Bob on Wed 28 December 2016 in Learning • 2 min read

There are some great free Python O'Reilly ebooks. In this post some useful tips from Mike Pirnat's How to make mistakes in Python:

  • Use virtualenv / pyvenv to isolate your environment.

  • I am still doing this: using the default REPL which leads to a lot of arrow-up repeating, use IPython or Jupyter Notebooks.

  • Always return a value from your functions / methods.

  • Use PyLint (I am also relatively late in this, future post, promise ...)

  • Read, use, abide by PEP 8, the Python style guide.

  • (not Python per se) name your variables wisely. There is a whole chapter in clean code dedicated to this.

  • Don't abuse lambdas, I almost only use them for sorting. Ditto for list comprehension, too much nesting makes it hard to read them. Same goes for decorators, too much nesting makes for complex code.

  • Avoid long if/elif/ blocks, wrap options into classes or dicts, use Enums, look at the Replace Conditional with Polymorphism pattern.

  • Leave extensive getters and setters for Java, use properties (future post).

  • Write small methods and (decoupled) modules, I wrote about this generically here.

  • Avoid the global scope.

  • Be specific in your imports, from time import * is asking for trouble. Use time.time() instead and you won't have name clashes.

  • "Explicit is better than implicit" (import this): don't use pass in except, handle the error (log it). And be specific what exception to catch (e.g. 'except IOError' is better than just 'except' which catches everything).

  • Don't re-invent the wheel: PyPI is full of great modules you can pip install.

  • Never use mutable default values for methods, it leads to weird behavior because each time you call the method you expect a new object (say list), yet you modify an existing one, use None instead. This and other common gotchas are described here.

  • Overeager Code: keep your constuctors (dunder inits) methods lean, watch out for modules doing a lot of operations when being imported.

  • Test your code! There are great frameworks apart from unittest, checkout pytest.

  • Logging is cheap, use it! Config is tricky (future post ...)

Let's thank Mike Pirnat for sharing all these great insights in this freely distributed ebook of just 80 pages.

Keep Calm and Code in Python!

-- Bob

See an error in this post? Please submit a pull request on Github.