eriksmartt.com/blog

The last day of PyCon

Day 3 went by rather quickly. The Keynote was given by Greg Stein, covering the topic of “Python at Google.” Greg’s been using Python for a LONG time and had some nice stories about using Python in corporate settings (including Microsoft.)

Google’s use of Python is mostly in their build- and server-management tools, but they do have a few web toys (like http://code.google.com/ ) that are built with it. He said that Google uses Python because it’s “highly adaptable,” meaning that software requirements and computing environments can change without drastically hurting the development process.

In describing the time it takes for programmers to learn Python, he said something to the effect of, “[you can] read it in 2 hours, program in 2 days, [and] contribute to the company in 2 weeks.” Fabulous.

For the web projects using Python, the meat of the work is still done in C , which is the most common programming language at Google. However, when the C code is built, the build system automatically uses SWIG to generate Python bindings for prototyping.

After the Google session, Jukka and I got ready for our presentation on Python for Series 60. After talking to people at the conference for a couple days, we decided that we should actually start the presentation by describing what Series 60 actually is. In hind-sight, we probably should have titled the presentation “Python on your phone.” I was a little disappointed that we didn’t have enough time to cover everything we wanted, but fortunately we had a lot of really good conversations in the hallways after the presentation.

After lunch I headed back to the Lightning Talks, and once again, there were some great projects presented. My personal favorite this time was the demo of rlcompleter2. Amazing!

To wrap up the day I went to Ted Leung’s presentation on PyBlosxom. I’ve been thinking more lately about moving my blog over to PyBlosxom, and thankfully, the session answered a few questions I’d had.

After the session I had to run to the airport, but amazingly enough, I’m actually on a plane (right now), and we’re on schedule. A delay out of D.C. made me almost miss my connection, but I made it right as they were closing the gate. So I’m 1-in-4 now for getting home on time this month… we’ll see how next week goes.

Day two at PyCon

Day two at PyCon started with a cold shower. Not by choice, mind you, but apparently the One Washington Hotel hasn’t been investing their profits into an adequate water-heating system. At least it stopped raining — it’s colder, but not wet.

The conference keynotes take place first thing in the morning in a mid-size auditorium. But surprisingly for the tech conference, you can’t bring your coffee (or any drinks) into the room — so it’s an early morning, sleepy crowd. This is also the one room that lacks wi-fi and power outlets, which is good or bad depending on how you look at it.

The keynote this morning was Guido’s “State of Python” address. He briefly talked about the ‘decorator’ decision process (mentioning that “Perl isn’t all bad”), then went on to discuss the formation of new Python Security Response Team in reaction to the first official Python vulnerability alert that happened recently (for SimpleXMLRPCServer.)

Guido then moved on to the more controversial topics of Python 3.0 being incompatible with Python 2.x (for example, old-style objects won’t be supported anymore, which is a good thing), the addition of the any() and all() standard methods, the likely removal of map(), filter(), and reduce(), and the potential demise of lambda. He then talked about potential changes to how divide-by-zero can be reported, and the very heated topic of static typing.

After the early sessions, we held an informal Python for Series 60 BoF, where we got some good dialog about networking on Symbian, but not too many people were there to hear it. That’s the nature of holding BoF sessions though, so hopefully we’ll have a larger crowd at the real session tomorrow morning. We were luck enough to get the main auditorium for the presentation (I say ‘we’ because Jukka’s joining me for the more meaty, technical bits), but unfortunately, the sessions are rather quick (20 minutes) and we’ve had to drop a lot of things we were hoping to unveil.

After lunch I went to the Lightning Talks, which I think wins my vote for the most interesting session. The format was a quick, 5-minute soap-box window for presentations about random projects, ideas, and questions. It was long enough to get the gist of what people were up to and jot down some terms to google later, but not so long that as to get boring. It’s hard to pick a favorite, but I really enjoyed Ka-Ping Yee’s demonstration of his new scrape module.

I’ve once again gathered the URL’s I wrote down during the day, but if you’re just looking for more detail on the sessions, Matt’s been doing a much better job then I at posting regularly.

Now for the links:
- http://www.elementalsecurity.com/
- http://www.infoworld.com/article/04/09/24/39FErrdev_1.html
- http://unicon.org/
- http://www.cs.luc.edu/~anh/gallery
- http://www.scipy.org/
- http://numeric.scipy.org/PEP.txt
- http://svn.webdav.org/repos/projects/ezt/trunk/
- http://codespeak.net/svn/user/argigo/hack/misc/
- http://codespeak.net/svn/shpy/trunk/dist/
- http://www.ludumdare.com
- http://pygame.org/

Day one at PyCon

I checked into the One Washington Circle Hotel last night, where I was given a complimentary upgrade to a one-bedroom suite. It’s ridiculous. No one needs a hotel room this big, and I’ll feel like I’ve wasted it if I don’t have a large gathering of drunken debauchery before the weeks’ up. Too bad I don’t actually know anyone in D.C.

The flight in was uneventful, but the descent path to Reagan Airport brought the plane right passed some of the more recognizable memorials. Having a left-side window seat gave a nice view of it all.

Unfortunately, the weather is crap today — slight rain and a tad bit chilly make it exactly what you don’t call walking weather; Fortunately the conference is all indoors. PyCon is taking place at the Cafritz Conference Center at George Washington University. The conference area itself is rather small, but there are only 400 or so people attending, so it works out alright.

The opening sessions of the day started with a quick look at the work the Python Software Foundation (PSF) has been doing to revamp the Python Web Site. The new site (and logo) isn’t live yet, but they will be changing the branding and feel of the site to be a bit more “business friendly.”

The backend system for generating the site is being updated as well. The new site uses YAML and reStructuredText for content storage. The thinking is that this format will make content authoring easier and more accessible.

Next up was Jim Hugunin (Microsoft) from the IronPython project. Jim started the presentation by asking (by a show of hands) how many people in the crowd were developing on Windows, and then, how many were deploying on Windows. It turned out, surprisingly few. I sit in the OS X / Linux camp as well (I guess I deploy on Series 60 too), but I wasn’t expecting that from this crowd.

Even though I don’t use Windows, Jim’s demos were quite compelling. With access to the .NET framework, he was able to interact with native applications and libraries from an interactive Python session. And according to pystone, Python actually runs faster on the CLI then it does through CPython! The performance gains apparently come from the CLI’s JIT compiler, which converts the IL bytecode to native x86 instructions.

One question from the crowd left Jim without an answer though. The issue was whether Microsoft will be patenting parts of IronPython that could eventually come to harm the Python community. The concern is that innovations coming to Python from Microsoft might be restricted to the platform (which obviously isn’t good.) Jim couldn’t comment on the issue; However, the IronPython work is at least being released under a BSD-like license.

After lunch I hit the Mac scripting sessions. The first covered appscript; the second and third were on PyObjC. I’ve looked at appscript before and found it a bit convoluted, so the first presentation was helpful in explaining “why” doing AppleEvents from Python seems so awkward. Apparently much of it is due to how the OSA environment works.

After appscript, Bob Ippolito gave an “Introduction to PyObjC” and “PyObjC Hacking” session. Both were interesting, but the latter was most inspiring. Bob gave a demo of doing PyObjC code injection into running processes. He focused on the Preview.app application, first inserting a class browser to explore private methods, and secondly to insert an actual Python console into the running application to open it up for exploration. The implications of this are pretty severe. It fundamentally explained a procedure for hijacking a running application and modifying it’s environment at runtime. The most abusive application I can think of would be to hijack an application like iTunes that does DRM encoding, and re-write the DRM methods in real-time to bypass the content signing mechanisms. Wickedly cool. Bob’s presentations can be found here.

After the lunch break I heard Michelle Levesque’s session on PyWebOff — a project to document Python web application frameworks in a hope to identify the best and encourage the Python community to focus on improving and adopting a small set of web frameworks. The motivation for the project comes from two sides; first being the actual problem Python has doing web development (it’s a little cumbersome), and second, the press attention the Ruby on Rails project has gotten.

After Michelle I heard Ian Bicking’s presentation about WSGI Middleware and WSGIKit, then Donavan Preston’s presentation on “Responsive GUI Web Applications” using Nevow. Donavan’s presentation focused on the LivePage functionality of Nevow that implements an XmlHttpRequest framework for Python. While the capabilities for this aren’t new, the success of GMail and Google Maps has brought it some new attention lately.

Since it’s actually pretty late here I’ll call it quits for now, but I’ll end with a listing of links I jotted down today:

- http://workspaces.gotdotnet.com/ironpython/
- http://toys.jacobian.org/presentations/2005/appscript/
- http://www.crummy.com/software/BeautifulSoup/
- http://developer.apple.com/
- http://pyobjc.sourceforge.net/
- http://bob.pythonmac.org/
- http://pythonmac.org/wiki/
- http://cocoadev.com/
- http://pyre.third-bit.com/pyweb/
- http://ianbicking.org/docs/pycon2005/
- http://nevow.com/

more travel delays

I guess I’m O and 3 now. I posted about my 38-hour trip home last month, but I seem to be on a roll here. Last week a snow/ice storm in Boston shut the airport down, and this week, a mechanical problem meant that we had to de-board and wait for a replacement plane. I got home around 1 am.

A stewardesses on the flight was complaining about the situation to some of the passengers. She’d been with the airline for almost 15 years and thought that they were really going down-hill. She claimed the reason it would take a while for them to resolve the mechanical issue is that they didn’t have enough ground staff. According to her, the budget-airlines were undercutting American so much that American couldn’t afford to operate. The ground staff that hadn’t been laid off were being employed under multi-year, minimum-wage contracts with no benefits. “Is this really how we want to treat people?” she asked. Sure, airline ticket prices are a bit lower, but at what cost?

Given my recent track record, I’m now expecting a long-haul heading home from PyCon.

Travel tip of the day

If you’re visiting a part of the world with a significantly different climate then you’re used to, and the locals are leaving work early because of “the weather”… follow their advice.

travel and airport culture

Thirty-eight hours of travel later, I’m finally home; Unfortunately, the trip wasn’t meant to be this lengthy. Trouble began in northern Europe. A three-hour delay for a mechanical problem meant I missed my connecting flight back to the States. Missing that flight meant that it was not possible to get home in one day. Fortunately, the airline was somewhat willing to move my flight schedule around, giving me the chance to do my evening layover in a city where a friend could pick me up from the airport. I say somewhat willing because it took three calls to the Executive Platinum desk before one of the customer service representatives would actually help. The first two told me I had to get to an airport with an American desk and let the locals do the re-ticketing. “Uh.. sorry, no. How about you just sort this out for me instead?” Oh well. At least I got home.

Even with all the chaos, my biggest take-away from the trip had nothing to do with travel planning. Instead, it was simply the cultural difference between how customers are treated in Zurich, Switzerland vs. the JFK airport in New York. These two airports couldn’t be any different, and I really felt sorry for a non-English speaking passenger I met along the way. In Switzerland, life was met with a smile and friendly conversation. Even the security check-points were pleasant. But JFK.. well, you can imagine.

Travel tip of the day: make sure you’ve got some snacks in your carry-on bag in case your flight schedule goes very, very wrong.

a rainy day on the mountain

After a great visit to Vancouver (and thanks David for taking the time to meet), a few of us headed up to the 2005 Snowboard World Championships presented by Nokia, at Whister-Blackcomb. Unfortunately, the weather is terrible (ie., it’s raining, not snowing), so I’m online. Fortunately I found a few good links:

- RSS 1.1: RDF Site Summary initial draft (via inessential.com)
Wow. Just as Brent says, I had no idea a 1.1 spec was being worked on. And my initial reactions are pretty much the same — the absence of the items sequence section is wonderful, but not requiring unique ID’s is odd.

- Preventing Comment Spam
An explanation of Google’s support for the “nofollow” hyperlink attribute to remove the incentive to comment-spam for boosting PageRank. Adding the ‘rel=”nofollow”‘ attribute is certainly a simple way to do this, but it does take the backing of major search engines for it to work. Way to go Google! Keep in mind though, this doesn’t stop spammers from targeting your readers.. it just keeps their spam from affecting Google’s PageRank system.

- WEP: Dead Again, Part 1
A quick read on securityfocus about the latest in WEP cracking tools.

Back from FooCamp

I’m back from FooCamp 2004! It was a great event filled with amazingly talented people. I want to thank Tim O’Reilly for hosting FooCamp and Rael for inviting me. I don’t know where to begin in describing the experience. It was only two days, but such a successful experiment in self-organizing info-sharing. It’s what happens when 150 smart people hang out for the weekend and share ideas. Absolutely amazing.

I think Drew Endy’s session Saturday night takes the cake for mind blowing (see parts.mit.edu for a quick glimpse at what these folks are up to), and Saul Griffith’s session about HowToons was a great presentation on engaging children by educating them.

Thanks to everyone who made the event a success!