Sponsored Link •
My impressions of the Python Conference 2007 in Dallas, Texas, written up on the plane home while it was still fresh in my mind.
I'm writing this on the plane back home from Pycon. I'm exhausted, but it's been a great week. The conference exceeded all my (and everybody else's) expectations, with a 40% attendance increase, excellent keynotes, and an incredible "buzz".
Apart from the keynotes I didn't see many presentations, but I heard that there were many great ones. Pretty much the only talk I attended was Jim Hugunin's talk where he showed the progress of the IronPython project. (Ironpython is Jim's Python implementation for Microsoft's CLR platform, which he created as an open source project before he joined Microsoft, and is now developing with a small team as one of the few open source projects at Microsoft.) The wizard Merlin was thankfully absent; his place was taken by a much more interesting demo, where Jim controlled a Lego robot via a Python script. IronPython 1.1 beta has been released; it features Python 2.4 compatibility, plus partial 2.5 compatibility enabled by a command-line switch.
Friday morning's keynote speaker Ivan Krstic gave an overview of the non-profit One Laptop Per Child project, which aims to produce a hundred million laptops at a cost of $100 a piece, to be distributed to children ages 6 and up in poor countries (with the cooperation and at the expense of participating countries). The first laptops (for relatively small trial projects in places like Brazil and I believe Thailand) are expected to ship this August. The project, started at MIT and with many industry partners (including Google), combines existing and custom hardware and software components into an educational platform, dubbed the XO. While some of the XO's hardware is proprietary, all the software is open source -- the basis is formed by a stripped down version of Linux (based on Red Hat's Fedora), on top of which they are developing a new GUI and window manager written in Python, and educational applications also written in Python. The XO's hardware design focuses on extreme ruggedness, reliability, and low power consumption; the expectation is that there will be no realistic way to service the laptops in the field, and the environment will often be harsh: temperatures up to 50 degrees Celcius, and irregular (and often absent) power sources.
Krstic gave away a few computers (the result of an early production test run), and I was one of the lucky recipients. I'll probably frame it and hang it on the wall; a bug in the mouse pad firmware causing it to omit some "up" packets makes it pretty hard to use (this will be fixed in the next production run of course). It's definitely no competition for today's commercially available laptops (which are 10x the price). Almost everything about it is small: the keys (just right for children's little fingers), the screen (plenty of resolution though, and dual-mode for viewing in bright daylight), the memory (120 MB on my copy, half that for the final product), and the flash disk (500 MB). The processor runs at less than 400 MHz. On the plus side, it runs a full Linux kernel with many utilities available, has self-configuring WiFi wireless, three USB slots, a camera, microphone, stereo speakers, audio in/out ports, and (once they tune the power management) it is expected to run for 22 hours on one battery charge. It weights less than a kilo!
The software is far from finished. An early version of the GUI and window manager are available, and a few small demo applications: chat, video, two games, and a web browser, and that's about it! The plan is to write all applications in Python (except for the web browser), and a "view source" button should show the Python source for the currently running application. In the tradition of Smalltalk (Alan Kay is on the OLPC board, and has endorsed the project's use of Python) the user should be able to edit any part of a "live" aplication and see the effects of the change immediately in the application's behavior. (A versioned document store will make it possible to roll back disastrous changes.) This is where Krstic wants my help: he hopes I can work magic and implement this feature for Python. I got started right away during the conference, with a reimplementation of python's reload() function that can patch classes and functions in place. Even this small component still has a long way to go; a checkpoint of the work in progress is checked into subversion as part of the Py3k standard library. That's not where the rest of my OLPC work will show up; they use GIT for source control, so I will get to learn that.
The keynotes had a strong educational theme: on Saturday morning Adele Goldberg gave a passionate plea for improvements to the USA's educational system. 40 years ago, US education was #1 in the world; today it is #19. The public school system is stuck in a complete political gridlock; changes are nearly impossible to make due to the many constraints imposed on schools by federal regulations, fearful and litigious parents, bullying, lack of funds, and many other depressing factors. It also seems that most uses of computers in the classroom have turned into disasters: the well-meaning geeks behind many school computer experiments don't understand the situation in the average classroom. For example, computers show up without sufficient power, are likely to be stolen, or locked up in a safe rather than being used! Schools are in total fear of the internet, which is seen as a source of pornography and influences of the devil. I hope Adele will put her slides on line, there was much interesting data.
On Sunday morning Robert Lefkowitz entertained us with a literate history of literacy throughout the ages, and (as is his habit) more questions than answers.
For me personally, this conference signified the coming together of the Python 3000 project (a.k.a. Py3k or Python 3.0). While in last year's keynote about this topic I mostly presented proposals, process, and plans, this year I could reveal many finished (as well as some unfinished or controversial) features, a concrete timeline with an alpha and a final release date (June 2007 and 2008, respectively), and, most importantly, a well-defined migration strategy. Migration was also the topic of several heated discussions I had at the conference, for example with a group of Twisted developers (lead by Glyph Lefkowitz and moderated by Steve Holden) about the best way to approach the transition for a project like Twisted, which is bound to support several different Python releases (currently 2.3, 2.4 and 2.5). The best strategy will be to start by adding 2.6 to the list of supported versions when it comes out, and then gradually drop support for older releases and add support for 3.0. Using 2.6 (to be release around April 2008, i.e. before 3.0) will be key for this strategy.
Python 2.6 will contain various compatibility features, such as optional run-time warnings when a feature is used that will be dropped or changed in 3.0, as well as the ability to import certain 3.0 features from the future (e.g. from __future__ import dict_views, which Thomas Wouters has already implemented in a branch). Once a program is warning-free in 2.6, a source-to-source translator that recognizes syntactic constructs but doesn't do type inferencing or data flow analysis can be used to translate it to 3.0 syntax, and there will be a good chance that it runs correctly under 3.0, although still not quite as good as when moving code unchanged between 2.x and 2.(x+1). The recommended development mode is then to develop under 2.6, and on a regular basis do the automatic conversion to 3.0 syntax and test under 3.0. When problems are found under 3.0, it should be possible to change the input to the conversion tool in order to help it make the correct tranlation.
I'll blog more about this in future episodes.
|Guido van Rossum is the creator of Python, one of the major programming languages on and off the web. The Python community refers to him as the BDFL (Benevolent Dictator For Life), a title straight from a Monty Python skit. He moved from the Netherlands to the USA in 1995, where he met his wife. Until July 2003 they lived in the northern Virginia suburbs of Washington, DC with their son Orlijn, who was born in 2001. They then moved to Silicon Valley where Guido now works for Google (spending 50% of his time on Python!).|