First off, early-bird registration ends in less than a week. This year has a fantastic lineup, and over twice the tutorials of any previous year. Not to mention the sprints which are free to the public.

We have a new registration system this year. It has its quirks, and there have been some issues. Most of these were expected and we had plans in place to deal with them. AMK has taken on the role of the Registration Manager. With things in his more than capable hands everything is running quite smoothly. I can’t help but stick my nose back in to resolve some issues. To be honest I do not have the temperament (or time) to deal with talking to actual people (code is more my thing). But some people I already know, or requests were made and I knew the person was still online. AMK didn’t help things by documenting some standard responses to some of the common problems, which made it too easy for me to ‘help out’ ;-). As of this writing we have more registrations than we had last year at the close of the early-bird period (375 in 07, and 390 right now).

So why do I have the Blues? I can’t decide which tutorials to take. Work is paying for me to go this year (and has even ponied up for a sponsorship!) That means that they have a say. This is additionally complicated by the fact that I still don’t have my XO, and it’s questionable that I will get it in time. What this means is I am waffling on whether to take ‘SWIG Master Class (David Beazley)’ or ‘Making Small Software for Small People, Sugar/OLPC Coding by Example (Mike C. Fletcher)’. I really want to take the Sugar class. We use SWIG at work, and not the simple stuff either. We don’t do any of the C++ stuff, but have things like type safety on our constants with custom repr’s and the ability to do things like pass in an array of length 0 and say its of length 100; have to test those memory error conditions after all. Callback functions are supported, oh and it’s thread safe, with all calls releasing the GIL safely. SWIG gets really painful at that level.

In the afternoon it looks like ‘Tools for Scientific Computing in Python (Travis Oliphant and Eric Jones)’ for me. There were hard choices here, I want to learn WxPython, and the generator tricks look very interesting, but in the end SciPy is the best choice both for work and for fun. With my existing experience I feel safe in skipping the morning companion tutorial. We use the SciPy packages including numpy and matplotlib quite a bit at work but my experience with the later is only cursory. I hate to admit it, but many times I use win32all to push the data into excel and generate graphs that way. I have it on my list to write an FFT and a group theory approach to pitch detection using SciPy; yes that is ‘fun’.

The evening is the toughest. This is where PyCon-Tech related tutorials come to the fore. It is down to ‘Practical Applications of Agile (Web) Testing Tools (C. Titus Brown and Grig Gheorghiu)’ and ‘Django Code Lab (Jacob Kaplan-Moss, Adrian Holovaty and James Bennett)’. I would love to go to the code lab and dive into parts of PyCon-Tech. The hard part would be selecting something small enough to discuss. PyCon-Tech has grown quite large and some of the parts are quite involved. Just giving an overview of the registration system can take 3 hours (from actual experience). There are parts of the proposal system which need optimizing (but again explaining the issues alone could take too long). The real problem I would love help with is the separation of the display (html templates) from the core of what I now think of as the PyCon-Tech framework. We have a new design, but once again the design is coupled with the implementation. Swapping out another design is easier than last year, but harder than it was just two months ago. Then there is the 10K lb. Gorilla in the corner. PyCon-Tech has 0 automated testing. Guess I will have to go for the testing tools, and hope to snare the Django folks some other time. To be honest I feel weird with the Idea of bringing PyCon-Tech stuff to a tutorial; as if I am stealing time from attendees for conference related stuff.

At some point I need to do a post on the registration system… The idea was to keep it simple. Just a minor update to the old cgi form with the data stored in a database instead of a text file. Three hour chat to give an overview…. For the masochistic, you can read the comments in the source code here and here (which are incomplete) and some other notes here and here (a little out of date).