PyCon 2008 is in full swing here in Chicago , with over 1000 attendees. My thoughs and notes from day 1 are below.
Chris Hagner of White Oak Technologies gave an interesting talk on “Why Python Sucks (but works great for us).” I liked the Python as a competetive advantage angle – once White Oak have installed a Python solution, their competitors (who don’t use Python) will only propose complete replacement. Of course, the customer isn’t interested in completely replacing a working solution. Alas, the growing popularity of Python will eventually cancel this advantage out. This was a good lead in to Guido’s keynote on Python 3000 (I missed the BayPiggies version of this…)
Other talks of interest I caught:
Pater Skomoroch on cluster programming with Python and Amazon EC2.
Peter has made Beowulf EC2 images with useful Python pieces available – check the Data Wrangling Blog.
Python and Flex
Bruce Eckel led an open space sesion on Python and Flex, to discuss the use of Flex as a tool for Python user interfaces. Given the Flex toolchain and capabilities, this idea might get some traction. I talked about our positive experiences with Flex and the SnapLogic designer. Diane Marsh suggested local Flex User groups as a good recruiting tool, since finding Flex developers is a challenge.
Google Spreadsheets API.
A nice session the demonstrated how easy it is to access google Spreadsheets from Python Also learned from Jeff Scudder that the Python libraries for Google Contacts API are imminent.
This was a good open space meeting on the topic of packaging in general, and issues with Python packages and distutils/setuptools versus platform installers. This is a broad topic, and many of the issues aren’t Python specific. It was good to have distribution savvy people here, since better packaging would really help getting platform specific installer support in sync with the python installers.
One of the more complex problems is what ISV’s should do to in terms of distributing packages they depend on – do you bundle your own copy, or take than chance that someone will upgrade or change packages you depend on and potentially break your installation? Theres no universal solution, especially since the next logical question is how far down the stack you recurse. It’s also clear that supplying a package with custom fixes doesn’t help getting the fixes upstream where they belong.
At the end of the meeting, I think the consensus was that we are very close to where we want to be. We can get there if we can get some additional community effort behind documentation, refined guidelines for package creation, and enhancements to setuptools. Follow up will happen in the
Best quote of the conference so far came from Tom ‘Spot’ Callaway of Red Hat – “I want an egg that contains the source to make the chicken that lays the egg.”