Versions Compared

Old Version 23

changes.mady.by.user Nathan Wailes

Saved on

compared with

New Version Current

changes.mady.by.user Nathan Wailes

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Table of contents

...

Criticisms of Python

  • 2016.12.25 - ZenHack - Why Python Is Not My Favorite Language
    • Posted by Yang on his Twitter
    1. Lack of Encapsulation - Python makes it more annoying than it should be to wall things off behind an interface.
      • My summary: There's no protection / warning if you unknowingly try to use a private variable name (like "_bar") in a class you've created that subclasses another class that also uses that same private variable name.
    2. Weak Lambdas - Python's lambdas are seriously hamstrung: They can only be a single expression. No statements. Python has a bunch of built-in language features that would be totally obviated by adding a decent syntax for multi-statement lambdas.
      • The 'with' statement's functionality could be done very simply with lambdas, and wouldn't need to have special code within the core Python codebase to handle just that feature.
      • I couldn't make total sense of his decorator example. He seemed to be saying that, when using Flask, it's 'silly' to have to give your view-functions names when they're already uniquely defined by their string routes, and that in a language with strong support for lambdas you could just refer to the view-functions as unnamed/anonymous(?) functions, which would make the code easier to understand(?) / cleaner(?).
    3. Duck typing isn't very useful for production code - it is all too common in Python to either use nominal types anyway, or pass around mystery poultry.
      1. duck typing: If it looks like a duck and it quacks like a duck, it’s a duck. Or, as a more concrete example, if it’s got a next() method that returns the next element and raises StopIteration when you’re done, it’s an iterator.
      2. I didn't fully understand his example, but he seemed to be saying that, when using duck typing, you need to be sure to clearly document what your interfaces take as input and what they return, and once you're doing that much work, you might as well write the code in a slightly different way that'll do error checking for you, at which point there is no longer duck typing.
    4. Lack of a Type System - There are a few things that I really want a type system for: 1. Catching bugs earlier, 2. Refactoring, 3. Documentation

Learning resources

Information on miscellaneous tasks in Python

...

  • Add logging to your code via a single decorator:


    • Code Block
      def run_with_logging(function):
    @wraps(function)
    def wrapped(*args, **kwargs):
        logging.basicConfig(filename='logs/' + function.__name__ + '.log', level=logging.DEBUG)
        logging.info('Starting at ' + str(datetime.now()))
        try:
            function(*args, **kwargs)
        except Exception as e:
            logging.exception(e)
            traceback_message = traceback.format_exc()
            send_an_email.send_email_with_error_message(traceback_message)
        logging.info('Finished at ' + str(datetime.now()))
    return wrapped


    • For your main function (that kicks everything off), just add "@run_with_logging" above the function declaration.
    • For anything you want logged in any submodules or subfunctions, just add a "logging.info(message)" to the code, and import "logging" into the module.
  • Add info to an AssertionError:


    • Code Block
      try:
    assert(thing_you_want_to_check_for)
except AssertionError as e:
    e.args += ("Invalid company domain", company_domain)
    raise


Optimizing Python programs

  • Optimizing Python programs, PyPy to the rescue
    • 7:35 - First focus on optimizing the complexity of your code.
    • 12:30 - If switching to PyPy doesn't provide enough of a speed-up, you can use VMPROF to profile your program. He explains how to use it.
    • 16:06 - I stopped here.

Packages

...

Virtualenvs