foss.in, day 1: kernel hacking, hadoop, qtwebkit, plan 9
I sat in James Morris talk about how and why you should be a kernel hacker. This was a talk you should have attended, rather than bothering to read the slides, because much of it was spoken, rather than presented in text. James shares that you must have genuine interest, common sense, and the willingness to work hard to succeed in a FOSS project (he was referring to the kernel, but lets just generalise here). In the getting started sense, he touched on the basics: fixing bugs, scratching that itch you have (i.e. angry programming – something doesn’t work for you? fix it), and trying many ideas (if an idea doesn’t work, try again). Its true that the initial learning curve to submit to most open source projects is steep, so persistence is important. Alas, follow your interests (don’t do something that you’ll dread), and seek mentors.
The part about being wrong. Its quotable: “If you find yourself arguing with Alan Cox, you’re probably wrong”. Memorable. But the advice on admitting wrongness, and moving on – this is crucial, to any FOSS project. Take it as a pearl of wisdom, and heed this advice. Its true for any FOSS project.
James carries an Apple Macbook. He had problems with displaying X. I found it amusing, that throughout the conference, a lot of folk had problems with X, WiFi, and resuming from suspend. In fact, resuming from suspend occasionally stopped working for me, too. Is Linux ready for the laptop? I’ve asked it before, I just keep wondering. Most of the KDE hackers (high number of KDE folk), actually seem to be using Macbooks. As are most of the Sun folk. After all, with virtualization, you can run other OSes (including OpenSolaris). KDE, with Qt libraries, are also generally native (with X11) on OS X. At some stage, I have got to try out GTK/OSX.
Personally, I think my time would have been better spent at Sreekanth’s talk (he’s from Mahiti, Sunil Abraham’s company), which was titled Natively supporting RDBMS in Plone/Zope for storing Content. The slides could do with more expansion (or I get that this is one of those talks that I needed to be in, to listen and take notes), but its something I’m surely going to look into, in the near future.
After lunch, I decided that I needed to learn more about Hadoop [Meet Hadoop! (Open Source Grid Computing)]. Devaraj Das gave a great talk. I had already learned a bit about Hadoop from watching a Yahoo! video, so this was expanding my understanding further. Hadoop requires a blog post on itself, because I have notes from the talk and from the video, sitting in Tomboy.
QtWebKit by Simon Hausmann was the bomb. Demos alone, took the audience by surprise, and made me want to try QtWebKit out. I found that even though they have mailing lists, most of the WebKit work is done via IRC – they’re on freenode #webkit. I found this to be odd – where’s the logging, and is this really setup for external contributors? Not everyone can be on IRC 24/7, and alone, is not the best communication method. Clearly, from this talk, my interest in GTK-WebKit is quite clear and apparent. As a rendering engine its great – maybe we should be out with Gecko. I only managed to check the code out later in the night at the hotel, but after 1.4GB of traffic (via svn), and a compile, my WebKit directory is now sitting comfortably at 2.1GB. Its, not exactly small, even though its just a rendering engine, eh?
Expecting myself to get bored at the talk on Plan 9 from Bell Labs, by Anant Narayanan (having seen a talk on Minix at linux.conf.au by Andrew Tanenbaum), I was actually really impressed. The design behind Plan 9, and the concepts, seem to make this an OS worth trying. Sadly, I never did get to download the ISO from their official site (I did get it from Anant’s site though). Its small (weighing it at 79MB), and apparently doesn’t really run on much “real” hardware, so I’ll give it a twirl on QEMU or KVM to see how it goes. I learned that UTF-8 was created for Plan 9 (and now we get the benefits of it!). This is clearly a project that benefited from the Summer of Code – by the looks of it, Anant got interested in this, after GSoC 2007. And kudos to Google then, for allowing others to realise the potential of Plan 9.
All in all, great day 1. I was planning on visiting the lightning talks, but they were cancelled for Day 1.
Technorati Tags: foss.in, foss.in 2007, kernel, hacker, rdbms, plone, zope, hadoop, qtwebkit, plan 9, plan 9 from bell labs
I don’t even try suspending my Thinkpad. The battery lasts long enough for it to keep running for most trips that I make. I just have to make sure that I don’t bump it…
APM was more useful than ACPI.
FWIW – 2/3 of the WebKit source tree is the regression test suite (the LayoutTests directory) and a significant part of the rest is other things that do not go into the actual compiled library (for example the webkit.org web site and various scripts and tools).
The directories that are actually part of WebKit as such are WebKit, WebCore (excluding manual-tests) and JavaScriptCore (excluding tests). This adds up to 117M of SVN checkout (so the source is about half that). That may still sound big, but it’s impressively small for a modern feature-rich web rendering engine.
[…] foss.in, day 1: kernel hacking, hadoop, qtwebkit, plan 9 By byte I sat in James Morris talk about how and why you should be a kernel hacker. This was a talk you should have attended, rather than bothering to read the slides, because much of it was spoken, rather than presented in text. … Colin Charles Agenda – http://www.bytebot.net/blog […]