Fosdem '05: GCC 4 + Java

At the pre-party drink I met up with Matthias Saou, who maintains freshrpms.net and is fairly knowledgeable about the current state of the Linux kernel and GCC. He told me that GCC 4 will soon be able to compile Java direct to native code, and not just as a kludge, but in a real, competent implementation.

This is absolutely fantastic if true, 'cos it means that I can directly compile my Tomcat to run on my VM. Right now I have like 6/128MB memory free 'cos the JVM allocs so agressively. This is very, very cool, especially if you consider that automatically deploying native compiles to GNU/Linux from an Ant script is no more complex or involving than doing the same to class files.

Fosdem '05: Gradient

So, I presented Gradient to about 8 people + 2 friends of mine (thanks for the moral support, Gael and Torrey!), which was interesting.

For a couple of minutes we had no video cable, so I chalked up a bit of stuff, like the URL for the presentation, and waved the laptop* under people's noses so they could read SVG and stuff.

It seemed to go OK. I didn't stutter, freeze up, stumble over my words or anything like that. However I'm not convinced I explained Gradient as well as I could have done. Gael was a great help in showing me how to explain my ideas properly, but I rambled a bit, and the demo's could have been better.

Nevertheless, it was a great opportunity, and it was good to meet Ralph and Heiner (author of Lluna) in person.

* generously provided to me by the company I work for.

Fosdem '05: RMS

So at the Friday drink, someone told me that RMS has only two speeches: one on Free software, and one on software patents. This isn't true. The subject of this speech was how the four freedoms of the GPL are applicable to other types of information.

1) There's a whole spectrum of creative activity between creation and copying - commentaries, coppendiums and the folk process were mentioned.

2) An early point is that most laws are theoretically made to achieve a certain (desirable) outcome for society as a whole in the current zeitgeist. Modern-day copyright of course is now increasingly anachronistic, and is getting worse at the behest of the content cartels.

3) The subject of software patents was noticeable in its absence.

4) He proposes a system whereby works are split into three categories - practical works of reference, expressive, attestational works (e.g. bills, receipts, contracts, articles, autobiographies), and artistic, creative works.

He then goes on to define what is and is not in society's interest in each case as regards copyright law, and does make a very good case for applying the four freedoms directly to reference works, and making no-derivs on attestational works. For the third type his solution is a limited term copyright, but I think that this is a bit of a compromise. A neat solution to this has yet to be found.

RMS is an idealogue. He's a good one too - The GPL, and Free Software are both Good Things. However, what was conspicuously absent from his speech was any hint of what could actually be done by people like us to help change things from what we have now to what he sees as the ideal situation. It was a good speech, but there was no call to action, and he's perfectly positioned to provide one.

I wanted to ask him an inflamatory question, like this one:
"given that you believe our supposedly democratic governments are not acting democratically, or in the best interests of our society, what forms do you think direct action and civil disobediance would take if we wanted to combat this?"
However, (a) there was no time, and (b) the thing was being videoed, and I didn't want to (unfairly) label myself as a firebrand info-anarchist in front of the whole Internet.

Oops, too late :-)

Fosdem '05: Jimmy Wales & Wikipedia

His was a really good talk, and had many good points.

1) Although he couldn't talk about what he was discussing with Google, I suspect that they'd like to link to them for definitions etc. at some point in the future.

2) He presented two explanations/perspectives on Wikipedia; firstly that it's the emergent behavior of hundreds of thousands of tiny actions, secondly that it's the coordinated actions of thoughtful users. He prefers the second, but IMHO it's both: the second is a result of the first.

3) Apparently, at a certain moment they face the prospect of the 'site growth outpacing the growth of the development and sysadmin team. This is another thing Google has under it's belt that nobody else has successfully replicated - their major advances in large-scale administration of clustered/distributed computer farms. Of course, this is a major strategic asset for them, so they probably won't apply this expertise to Wikimedia's server farm for free.

4) Kudos to Jimmy Wales for having the courage to stand up and say that love and respect is an essential part of what makes Wikipedia tick.

5) He made a really good point that things like the Slashdot moderation system are inherently mechanical in nature, and compensate a little for a lack of community. In contrast, wikis are totally freeform, and don't dole out reputation or XP, which means that all moderation/structure/regulation occurs as a result of community action.

This is important: wikis provide space in which communities grow, and their structures grow and evolve with them, and are inherently human in nature. Contrast that with the /. moderation system, and the games/controversies/pending rewrites that surround it.

What I find interesting is to see where else on the Internet we can move from mechanical regulation to simply providing a substrate in which real human communities/societies can grow, unfettered by mechanics imposed by a centralised authority. Something to sleep on.

Fosdem '05: impressions

So yeah, Fosdem '05 and the traditional drink was great fun and really interesting. I've never seen so many geeks in one place. Met a bunch of cool people, had a great time. A friend of mine went crazy with a cash card at the O'Reilly stand, but hey, I might get to read the books too :-)

1) Geeks know how to party.
2) Anti-social my ass.
3) RMS is probably the only person on earth who would give a good, intelligent talk
a) without slides,
b) commanding fixed attention from so many semi-ADD geeks,
c) absentmindedly scratching his nether regions in front of more than a thousand people.
4) Giving a decent speech/presentation is quite a skill.
5) Until Friday, it never quite sunk in that there's an enormous society of people to whom Free software is a real, tangible matter of principle. RMS is on to something, and I'm not talking about the itch in his pants.


Slideshow presentations

The reason people put all that stuff in bullet points...

When you create the thing, you think you have to say what needs to be said at the presentation, with your slide show.

Not so. You will be there, and you will say what needs to be said. The slideshow or PPT or whatever is not there to say things, but to show things.


CodeCon '05: misc.

Incoherence was incredible, something I will start showing to people.
OzymanDNS is totally mind-bending. Stuff like this is going to take back the 'net. TCP VPNs over SSH over IP over DNS...
Note to self: check this out.

CodeCon '05: SciTools

SciTools had a very, very cool demo. Not sure what extracting DNA from chick-peas with common household materials actually had to do with the presentation, of which I understood very little, but still.


- Molecular biologists and geneticists face the same issues as hackers when talking about the uses to which their work could be put. Breast cancer detection and Anthrax - almost all mentioned in the same breath concerning the same tech.
- Another issue they face is that they don't want to deal with the GATCC stuff any more. That's low-level assembly thinking. As far as I understand, nature already has it's own HLL's, but I'm not entirely straight on that point.
- Cool sparklines and vast databases of stuff is online, e.g. LocusLink and OMIM.
- The human genome fits in 250 meg of RAM.
- Quote: "Biologists don't index from 0".

CodeCon '05: Wheat

Wheat is very cool. Building an object model in URL space is a great idea. One endpoint is to publish and extend from objects on other servers. Strange stuff: new language, which is understandable given the 'research project' nature of the beast, but still a little regrettable. References are a little strange, but this is to be expected. A little polish needed, but very, very cool.


- look at TinyTemplateEngine as a replacement to JSP. It looks niiice.
- check out Zope web-objects
- Check out Rails and Ruby, 'cos it's the next best thing, baby!
- Check out the Wheat Wiki.

SF '05: Memorable characteristics and events

1. When you ask for "coffee" in a diner or something, you get some terrible drip-filtered gack. No wonder Starbucks has stomped all over the USA. By way of comparison, even the smallest eateries in Brussels have coffee machines that make a passable espresso and decent coffee.

2. Buying public transport tickets is a great way to get rid of even the smallest loose change. Over here, this stuff accumulates to levels at which you could store them as ballast in a ship's hold and use them for shrapnel in it's cannons[0]. On the buses and trams in SF you can pour fist-fulls of the stuff into the machine and it'll sort and tally it all correctly.

3. The SF tagline should be "the Yoga and Tai-Chi capital of America". Get out early enough and you'll see any given green space has a cadre of people doing their exercises.

4. The place is surprisingly bike friendly, provided you stick to the costal areas. Enormous great SUVs give you half the lane to play with, as if they'd rather suffer a head-on collision with another motorist than a mental health lawsuit from a mildly claustrophobic cyclist.

5. Everyone is friendly. Cool, interesting conversations were had on diverse subjects such as the psychology of addiction, the concept of currencies, long term geographic effects of global warming on property prices and purchase strategies, the LT, WTFDNU, etc. This stuff is slowly permeating through into the general conciousness.

6. Geeks are everywhere. Trying to sleep on the plane when your brain is randomly plucking tech words from surrounding conversations is like trying to sleep during a Windows install. You know there's nothing better to do [except fill out your product registration card!!11 :-P], but something might come up that makes life interesting.

7. T-shirts: saw a T-shirt with "Dept of Homeland Security - since 17nn", and a picture of four Native American Indians with muskets. Hahaha.

8. The Apple store in SF: It's JUST A STORE. Why people queue up outside of it every morning is totally beyond me, nice machines though they be.

[0] Why yes, I happen to be reading The System of the World, which is a fantastic Stephenson, and which also contains the best Monty Python Joke Ever.

Various observations re: travel

1. In coach on Boeing 747's, the aisle-facing armrest hook is lethal to garments with pockets that can snag. This fact was confirmed by repeated trial and error. We honour the pants that gave their seams in the cause of science during these experiments.

2. Most airport toilets now have a dry-blower and paper towels, which prompts one to ask the question: if I wish to dry my hands after having washed them, which is more environmentally friendly? The blower directly increases the amount of heat entropy in the universe, and may be run of a grid that is feeding partially-renewable electricity. The paper will probably be recycled, but the process of collecting, processing and recycling used hand-towels almost certainly takes more energy than that used by the blower. On the other hand, paper towel collection scales fairly well. Maybe different tools should be used for a given hand wetness index.

3. My head is a strange place. I woke up to the alarm, saw "8:00", got ready for work, stepped out of my flat and realised that it was actually 20:00. I reset my alarm clock wrong, but my body rythms are so screwy that I can't tell the difference. On the other hand, this means I have lots of sleep to look forward to.


CodeCon 05: RPOWs

My introduction to hashcash, which is interesting. RPOWs are (briefly) digitally signed, unique, audit-tracable bits of hashcash. Verification and the all the sensitive stuff is done inside of an IBM 4758 crytoprocessor, which is (a) expensive, (b) very, very secure.

The spread of Guerrilla cybercash is one of the things I'm predicting will result from the dissemination of TCPA-ish technologies, as I outline briefly here. Hal Finney, being the legend that he is, has already done cybercash for the nearest thing we have to a consumer-level TCB.

One thing he alluded to in his presentation is that he's taking advantage of the 'remote attestation' features of the 4758, which in other contexts (such as DRM systems) would be seen as a bad thing. Remote attestation is unique to the 4758 right now, but if it becomes more widespread via NGSCB and TCPA, then things could get interesting.

CodeCo '05: ApacheCA

As seen here.

Originally meant for auth and id for Apache people, now has an open CA for anybody to create signed certs, groups and stuff.

Opinion.1: The Apache root cert shouldn't be dependent upon Verisign, or anyone else.

Op.2: Once live, Apache's root cert should be included in the FireFox install. People should be able to secure their websites without paying Verisign a dime.

Observation.1: Apache could provide a pluggable, secure solution to the issue of auth and id in the context of online communities and P2P networks, and one that's considerably more palatable than Passport or the Liberty project, owing to Apache being non-profit and having better whuffie than MS or the Liberty project's backers.

Obs.2: I've said elsewhere that SSO is a bad idea in principle (because it commutes credentials across security domains), but that's something else that could be done with this if you wanted to.

CodeCon 05, San Francisco:

A beautiful city. In FEBUARY. This will make Brussels a little harder to come back to.

Observations from Day 1: when the projector broke, people actually paid more attention to the presentation in general, and the speaker in particular. Re-inforcing my opinion that PPT as commonly used is plain bad for presentations.


Thank you JURI

This makes me very happy.

Software patents delenda est.

On the importance of introducing oneself

So me, being the smart-arse I am, rigged up my systems to email me when errors in my webapps occur, from a generic mailbox on our work server.

Today, I get a stack trace out of the blue, and all the systems that I know of are working fine. The email doesn't say what system it comes from, merely that something is wrong with it.

In future, I shall be more courteous. "Hi there, I'm system FOO BAR, "+e.getMessage() and so on and so forth.