Jeff gave a presentation/overview of functional programming in Python
Eli talked about:
- Google's MapReduce - a software framework implemented by Google to support parallel computations over large (greater than 100 terabyte) data sets on unreliable clusters of computers.
- HardOCP - Hadoop implements MapReduce, using the Hadoop Distributed File SystemHDFS) MapReduce divides applications into many small blocks of work.
- glusterfs - a clustered file-system capable of scaling to several peta-bytes.
Jeff gave a short presentation on anecdotes about optimizing Python code.
- Premature speed optimizations are the root of all evil
- Enhanced readability is an optimization
- Follow pep8 and use structured code — you get an automatic speed up by just keeping your name spaces clean and local
- Know where your code is spending it's time and optimize their first — when in doubt, optimize the inner loop first
- Remove the dots - short circuit lookup intensive operations. i.e. _lstappend = lst.append
- Check out shedskin — although it can compile entire programs, it is perhaps most useful at speeding up a targeted area of code. ShedSkin makes writing an extension module very simple — see "optimize the inner loop first"
We enjoyed some decent pizza and shop talk between talks.
The door prize, "Python in a Nutshell" was won by Burch. A big Thank You to O'Reilly for the door prizes at our meetings. Thanks to you to Marsee!