Thursday, July 26, 2007

OSCON 2007

OSCON 2007

I went to OSCON this year. I'll tell you all about it in a moment, but first let me apologize in advance about this report. There's an ongoing debate in some circles about whether bloggers are "journalists" or not. I don't have a really set opinion about that -- I don't really like journalists very much, but being one gets you certain traditional legal protections. What I can tell you for sure is that I'm not much of a reporter. I like to think I'm ok with a camera, but I'm also not in any way a news photographer. And I'm a tad on the antisocial side, so I'd make a very poor interviewer if I was inclined to try it, which I'm not. Also, I don't have the chutzpah to try to use my status as a Z-list blogger to get special treatment, evne though I manage to be ignored by billions worldwide every single day. No, I attended in my professional capacity, and never breathed a word to anyone about my personal media nano-empire. Not this blog, not even this blog's geeky and obsessive sibling.

I'm not even a very good conventiongoer, to be honest. My PHBs were fine with my going to OSCON this year, since it's just across the river, and it was still possible (barely) for me to shuttle frantically back and forth between OSCON sessions and random meetings with the PHBs. My PHBs are also cheap bastards, so they were fine with my going, so long as it didn't come out of their budget, so I had to resign myself to the free admission option, which means you get keynotes, exhibit hall admission, a small subset of the total convention sessions, and a few random other goodies. You don't even get one of those nifty OSCON 2007 laptop bags. Not that I need another laptop bag, it's just that it's fancy schwag I'm missing out on. It's the principle of the thing.

So anyway, I went to a few keynotes and attended a couple of free sessions, and picked up a few t-shirts and other schwag. I took a few photos while I was there -- the full set's on Flickr here. Here's one photo you might find entertaining, from a talk by Bill Hilf, the guy in charge of "open source strategy" or some such up at the Beast of Redmond.

OSCON 2007

The one really newsworthy bit from his talk is that MS is supposedly working with the OSI to get their various "shared source" licenses approved as genuine Open Source (tm) licenses. I'm pretty sure he just said "working with", so there's no ETA on when -- or if -- this might be finalized. So I don't know how much meat there really is here, but it was an effective PR move to take at this particular venue.

Hilf seemed like a reasonably personable guy, cracking a few jokes at the start to warm up the crowd. The talk was primarily an attempt to convince us that M$ isn't the Great Satan, and there wasn't a lot about technologies I use or care about. (Ruby on top of .NET? Okayyy...). An audience member asked about software patents, and Hilf mentioned the controversial MS-Novell deal while answering, but the answer was a sort of managerial non-answer. Apparently the big problem with their recent patent FUD is not that it was incorrect, or that they're playing predatory monopolist again; it's just that they mishandled the messaging about it. As if there was any way to accuse people of violating your super-secret patents and not offend anyone.

So possibly you might gather from this that I remain firmly unconvinced. But I was polite, of course, and clapped where appropriate.

The MS guy was sandwiched in between an economist (who wanted to tell us all that we aren't objective enough and need to overcome our biases), and the guy who started the Swedish Pirate Party. A photo from that talk:

OSCON 2007

As with the preceding talk, I remain unconvinced. I can understand the visceral appeal of the absolutist approach, arguing that there just shouldn't be any such thing as copyrights or patents. But think about it for a moment: You can't have a GPL unless you have some sort of legal copyright framework in place, some way to get people to comply with your release terms. Some might say that's a good thing, but I disagree.

There was also a claim made that you can either have civil liberties or copyright, but not both. The argument is that effective copyright enforcement means the government has to read everyone's email, to check whether people are mailing illicit MP3s to each other. You could make a similar argument that all traffic laws should be abolished, because the only way to catch every last traffic violation is to put government tracking/reporting devices in every car on the road. Come to think of it, catching every last violation of just about any law would require a police state, and would also be extremely expensive. Hence it's rarely attempted, and it never succeeds.

Don't get me wrong, I hate DRM and the DMCA at least as much as the next geek. The current laws on the books go wayyyy overboard and need to be scaled back, either by Congress or the courts. And don't get me started about dumb software patents and patent trolls.

Yesterday's keynotes were somewhat more technical: A couple of talks about threading and concurrency issues, including one from Intel about their newly-GPL'd ThreadingBuildingBlocks library. I picked up a CD at their booth in the expo hall but haven't had time to check it out yet. So I don't have a lot to say about it, other than that it sounds promising. I've dealt with my share of ugly threading issues over the years, and if there really is a magic bullet out there, or even a shiny but nonmagical bullet, I'd love to find it. Also, I got a cool t-shirt along with the cd.

Intel obviously released this out of enlightened self-interest. Multicore chips are the shiznit right now, and you can't use a multicore cpu efficiently unless you do stuff in parallel, and do it reasonably efficiently. You're not getting the full effect unless you can keep all those cores busy, and you aren't likely to pop for a shiny new 16-core Xeon a couple years down the road if your apps grind to a halt from all the inter-CPU overhead. The point is that they have an economic incentive to address the problem. Whether they have addressed the problem is a question I can't answer right now.

There was also a much more academic presentation by a guy with M$ Research, over in Cambridge, UK. He chatted about something called transactional memory. The idea is to have memory accesses work like database transactions (hence the name), so you can define a block of code such that either all the related changes to memory occur, or none of them do. This has a certain aesthetic appeal to it, but when someone with a major corporation's research arm tells you there's a performance impact and does sort of a handwave about it, it's quite often a sign you'll either need to buy a much faster machine, or get used to molasses force field mode.

So here's a bit of the big expo hall. Longtime readers may notice that the place looks a lot like the beer festival I went to back in April, except with less beer. I did see some people walking around with beer, so it wasn't entirely absent, but I think you had to get it in the paying-attendees-only area. And it didn't look like very good beer anyway, so phooey on that.

OSCON 2007

I wandered around the expo hall for a while, picking up a few t-shirts and other misc. schwag items. The highlight was actually at a booth run by Oregon State University's CS department. They're writing some media player software for the OLPC, the long-awaited ~$100 Linux laptop, and they had a few units on hand to play with. Intel was showing off its semi-competing (and also Linux-based) Classmate PC. My quick-look impression was that the OLPC is a very cool, clever gadget, the Classmate somewhat less so. Neither is exactly a speed demon, but the Classmate seemed slower. I think that's because it comes with a bunch of off-the-shelf open source apps rather than the OLPC's new custom-written ones. You click "new text document" or the equivalent and then wait a looong time for the editor to start up, and then you realize you just launched OpenOffice. It's cool to see OOo on a box that size, of course, but it would help a lot if the box was a touch faster. My real complaint about both devices is the keyboard. I think I now understand why both are aimed at kids; my clumsy old grownup fingers had all kinds of trouble. Touch typing just wasn't doable, although you could probably adapt to that given enough time.

Here's a photo of the Intel booth. They had the large booth in the center of the hall, surrounded by transparent mesh "walls" covered with nature photos. Not the most informative photo, but I thought it turned out kind of cool.

OSCON 2007

I don't have my pile-o-loot in front of me, so I suppose I should take some more pics and post those, just to give a sense of what sort of schwag there is to be had at these things. And I've probably forgotten something I wanted to cover, so there may be a second OSCON post Real Soon Now. But right now I'm going to go check off another item on my busy social calendar, the Oregon Brewers Festival. Yes, OSCON happens to run concurrently with one of the nation's premier beer festivals. Just try to tell me that's a mere coincidence.

No comments :