Curious Venn

Twenty people came to the MPUG meeting at Inspire 9 this month for an evening of Python and pizza.

PyCon AU 2013

The dates for PyCon AU 2013 have been set: July 6 – 7, 2013, which will coincide with uni holidays. There is a plan for miniconfs, such as Django and OpenStack, on the 5th, hopefully with another CodeWars, and then sprints on the 8th and 9th.

The parse module

Richard Jones briefly covered the syntax for new style string formatting before launching into parse(), which is essentially the inverse of format(): parsing strings with format() syntax. There are 20+ types/conversions available and a plugin system to add your own types.
There are additional (obviously named) functions: search(), findall() and compile(), and while this may seem very reminiscent of re, the regular expression module, it offers a lot of the power without the often accompanying complexity.

parse is available on pypi and Github.

Message queuing

Richard also presented his PyCon AU 2012 talk on message queueing by an MQ noob, covering some of the popular MQ solutions, with a focus on RabbitMQ and Pika. The summary was that RabbitMQ (and AMQP) is enormously powerful but also challenging to configure and debug, especially if you do not know Erlang (which RabbitMQ is written in).

TCPCatcher was mentioned as a useful tool for testing networks. It is cross-platform and allows you to monitor, record and modify TCP packets.

I recommend checking out Richard’s slides.

Python 3.3

As everyone hopefully knows, Python 3.3 is officially out. Rather than covering what is new in Python 3.3, Ryan Kelly highlighted what he found exciting in Python 3.3. It was a long list!

• Syntax for delegating to a subgenerator, as per PEP 380
• Explicit unicode literals, to simplify porting legacy 2.x code
• Flexible unicode representation
• Reworked I/O exceptions
• A script launcher for MS Windows
• Removal of a GIL (but this is not the GIL you were looking for…), the Global Import Lock
• Tracebacks on segfault
• Easy namespace packages
• virtualenv support in core, via PEP 405
• unittest.mock
• A new type: types.SimpleNameSpace

Could 3.3 be the release that is the tipping point for the adoption of Python 3?

Next month

Despite the public holiday for a horse race, we will still be meeting on November 5, the usual first Monday in the month.

These are my notes from last week’s Melbourne Python Users Group meeting at Inspire 9.

We had a great turnout, probably due to the combination of last month’s PyCon AU and Richard’s guest spot on Byte Into It the week before. We also had a puppy come along — I think that’s a first for MPUG!

Richard kicked off the meeting with an overview of PyWeek, the Python game programming challenge, which started on Sunday.

Tennessee gave us three short talks:

• a snippet of code that allows you to connect the Python debugger to a signal handler, so (for example), you can use Ctrl-C to drop your program into the debugger at any time. There are more details in Tennessee’s blog post.
• the links to the PyCon AU talk videos which are now up at youtube.com/pyconau (and Adam MacLeod has started a list of presentations and slides that are now online)
• an introduction to GitLab, the self-hosted Git system for those who don’t or can’t use GitHub. It was worth noting that it took a day to get GitLab (and the supporting Ruby on Rails infrastructure) set up on a Fedora box, with various installation issues. Along the way, we detoured into ReviewBoard, the web-based code review tool.

Benno gave a great presentation on deploying web applications to Heroku and the
“twelve factor app” methodology that I had not heard of before.

Javier is teaching at Monash and is switching his first year Data Structures and Algorithms course from Java to Python. He presented a way of extending the dict class with the __missing__ method to create a caching fib[] dictionary and his maht module (and that’s not a typo!). This led to an open discussion about the benefits of looking at the pure Python versions of various data structures and Peter Norvig’s Infrequently Asked Questions.

Noon Silk showed us how to use pycallgraph to generate live call graphs of web applications using CherryPy. The demo source code is available and has plenty of potential for extending the code to graph arguments as well as methods, along with additional filtering and logging.