linux.conf.au

Well, the conference is nearing its end, and yesterday morning it was showing. Wednesday night finished up rather late.

Yesterday I had a look at Rusty Russell’s talk on advanced C programming. Very enlightening, although much of it went woosh over my head due to my state of sleepiness and the speed of the presentation. In particular, it introduced the Comprehensive C Archive Network (the logo and concept is a blatant copy of CPAN) which I will definitely have a look at, as I find myself coding in C more and more.

I also had a look at the Dili Village Telco, which is making a big difference in Timor Leste (East Timor). I wonder though… now that people have been saved those long 3 hour walks, is there now a looming obesity crisis in Dili? Hopefully it’s compatible with the Serval BatPhone, as I can see the two being quite complementary once the economy picks up enough for people to afford the Android devices required.

Lunar Numbat brought up some interesting concepts. In particular, the use of 8051 hardware surprised me. It seems high integration and radiation are not a good combo, unless you want to see your bits flipped randomly. I do worry about the MCP2515 CAN controller in use though — that thing gave us lots of grief in an electric harvester project (a lot of data corruption issues in fact, frames about-to-be transmitted would frequently get their data payloads overwritten by received frames before being transmitted).

I learned a bit more Perl tricks. Perl is quite a nice language really, and gets unfairly called “executable line noise” (which can be done in any language). I finished up with a discussion on how to disassemble flamewars and extract useful information.

The Penguin Dinner went down well, another late night but quite enjoyable. We didn’t do so well at the trivia, but I chucked in $50 to the flood effort. It finished again after 10:30PM, so another late night. Thankfully I wasn’t riding the bike home.

Well, this day the conference got on in earnest. Tuesday I spent almost all my time in the Parallel computing conference, which was very interesting.

The first session was, let’s just say, very interactive. Dr. Vinton Cerf giving a presentation on the implications of parallel computing on network infrastructure, wound up being as much him teaching us, as the rest of us teaching him. A true two-way exchange, with the added challenge of hearing aids deciding to turn themselves off without warning. 😉

That, and the discussions into FPGAs were very interesting, but sadly, I won’t be messing with FPGAs for fun at home. Too expensive, and sadly the companies involved seem to think that something is at stake by being truly open. (Yes fellas, release a FPGA with fully open tools, and I may consider trying out your product! Keep the status quo, and I’ll keep ignoring you!)

Yesterday started off with a very pertinent talk from keynote speaker Geoff Huston discussing IPv6 and the impending shortage of IPv4 address space. (i.e. we run out in February this year. Hint hint iTel!)

The direction that the commercial interests are pushing the Internet in is certainly a sobering thought. I myself run a dual-stack network, and I intend to code what I can to support IPv6 from the outset, however there is a hell of a lot of inertia to overcome on this issue. Non-technical minds are touting NAT as the solution, which still only gives us 48 bits of effective address space. Yes, fine for a small network, but globally it’ll be a nuisance. And NAT makes it harder to trace offenders.

We’re in for interesting times with X by the looks of things, with a lot of code moving out of userspace and into the kernel. That, and multitouch adds a whole new layer of complexity — largely because the hardware cannot tell us much about what the user is doing, we have to make an educated guess. That’s going to be fun to code for!

I was hoping to see the talk on ARM Device Trees by Jeremy Kerr, but sadly that got cancelled. So I prepared to pack up and check out the HTML5 discussions, just as I do so, a horde of people stream in. The HTML5 lecture had moved to the room originally slated for the other lecture. So I stayed put. HTML, CSS and JavaScript have come a long way in the last few years, and some of the things that are now possible make definite food-for-thought.

Sadly most of it leaves IE users out in the cold, but perhaps that’s payback for all the ActiveX and IE-only sites out there. I certainly will be making no special effort to support this browser.

Last night was a late night at the Professional Delegates Networking Session, a bit of confusion as to whether dinner was provided or not, when it started, and for some, where to go. It seems if you ask a Tom Tom GPS for directions, it’ll point you way over in the direction of West End. If you ask Google Maps, it’ll direct you to the Queensland Museum. At least a few of us locals could point the way. 🙂

The networking session lasted ’till after 10:30PM, at which point I made the long trek home on the bike, arriving at midnight. Needless to say I’m sitting in the back row of F509 now, somewhat zombie-like, and I gave the bike ride a miss today due to the late night. Hopefully the cup of tea before Rusty Russel’s lecture will wake me up. 🙂