23rd March 2008

Posted in

Blog :: Baypiggies, Post PYCON edition

I went to baypiggies Thursday night for the first time in several months. The crowd was actually pretty small (some people were still at PyCON) but I had a really good time. JJ, Drew Perttula, and Wesley Chun presented on their PyCON experiences.

JJ has issued a flood of posts about his PyCON experiences. He said Martelli's talk on Callback Style Programming was the best (I think baypiggies is going to get an extended version of that talk sometime) but he also said the Raymond Hettinger talk on Python containers(lists, sets, dicts) was his favorite :)

Drew has a page up with links to the things he talked about. Based on his talk I have a couple of pieces of software I need to check out; particularly trestle (doctests for web apps) and supervisor (process monitoring)...

Wesley talked about a couple of things but ended up showing us some demo code that used the Google Spreadsheets API It looked really simple to read and write to the spreadsheet and especially paired with publically editable spreadsheets (or the new and awesome "make this spreadsheet a web form" functionality) I could see using this for some cool notification/cross machine storage capabilities. Wes also pitched his upcoming book a bit (NOT Core Python 2; that will be out in another 3 or 4 years he said). Wesley is writing a Django Book with the estimable Paul Bissex and Jeff Forcier. Django's getting a plethora of books all of a sudden and there was actually a bit of talk about Django at PyCON as driving some of the increase in interest Python. Apparently somebody asked how many people were programming in Python because of Django and got quite a bit of response.

This leads me to my own take away from the evening, however, which was to notice that there was a bit of Django-bashing as well. JJ contrasted Django's ORM unfavorably with SA (which is correct, but perhaps not as important as you might think.) I have hopes that as the admin app is decoupled from models and there's some movement to make a Django ORM->SA translation layer (see Michael Trier's blog) and have to admit that scanning the SA talk slides from PyCON makes me interested in playing with SA a bit.

More seriously, however, several people expressed great displeasure at the release state of Django. It's good that version 1.0 is coming (and check out the feature list: I'm really looking forwards to newforms-admin and model inheritance, but there are lots of other goodies there as well.) However! Nobody uses 0.96 which is a year old and the advice you'll get from everybody (including me) is to use trunk.

This does not tend to promote the impression that Django is stable, solid, and mature. I gave a little pitch for the impressive quality of the trunk code in my experience, but several people were fairly contemptuous about a project that never has releases. It also makes book writing hard (What are you supposed to say: use at least rXXXX or newer for the examples in the book but things may have changed already...) I've heard some of the arguments against 0.97 but I really think it needs to be done. Land one or two items that are done or almost done and roll it out, then head towards 2.3 or whatever we're calling the next release :)

Final and unrelated thought. Somebody was fumbling for the name of the "Consuming HTML" presenter at PyCON and a fellow behind me said "Ian Bickering, wasn't it?" I like Ian's work (<sarcasm>and I'm sure he's never heard that one before</sarcasm>) but I cracked up...

Posted on March 23rd 2008, 01:21 AM