I had another cleat-related incident on the way home tonight. I thought I'd unclicked my left foot when I came to cross the road, but apparently hadn't, and I'd already rocked towards the left and was well and truly leaning that way before it became evident that my leg was still anchored to the pedal. I fell pretty heavily on my left wrist, and it made a nasty crunching sound. Doesn't appear to be broken though, just sprained I guess. On inspection of my left shoe when I made it home, it would seem that one of the bolts that holds the cleat to the shoe has come loose and disappeared. So when I turned my foot sideways, the cleat rotated on the other bolt, and didn't release properly, whereas I thought it had released by virtue of the fact that my foot moved sideways. Just as well it's forecast to rain tomorrow. Don't feel much like riding at the moment.

I did some investigation (which was very little, dak’s source is very helpful) upon the possibility to do source-only uploads to Debian’s ftp-master. In short: It’s not possible. Jennifer rejects uploads which do not contain a single binary Debian package:

if not has_binaries and Cnf.FindB(“Dinstall::Reject::NoSourceOnly”):
    reject(“source only uploads are not supported.”);

But well, there’s a way to trick it: If the package contains an arch:all binary independent package and if you upload it together with the source (the binary-indep target of debian/rules will help), Jennifer thinks that the upload contains indeed a binary package and ignores the fact that there are no real architecture dependent binary packages. The autobuilders will do the work for you.

Andrew Reid is now truly mobile. I achieved something approaching mobility Zen, where I could use my Sony Ericsson T630's GPRS connection via Bluetooth from my PowerBook and my iPaq. And then I left the country. Bit of a bugger that.

CRCnet Configuration System
Now that the CRCnet Configuration System is nearing a releasable state, we’ve had approval from the University’s commercialisation arm that we are allowed to release it under the GPL. This is good, I’ve been working on the system for the last year at least and it is nice to know that it will be available to the community at large. I think what is going to be most interesting to me is to see whether there is any interest in using it from external people and particularly how easy they find it to use. The current plan is to release version 1.0 at LCA’06 in conjuction with our talk of everything that the CRCnet Project has achieved over the last few years. Debian Stuff
Haven’t made a lot of progress with my NM application over the last month. I think both Alexander (my AM) and myself have been too busy with other things. Alexander did get the latest PHPwiki packages that I had prepared uploaded and there have been no new bug reports yet, so I think we’re making progress. This package made use of the wwwconfig-common functions developed by Sean Finney which was a great help and we’re much closer to having a robust PHPwiki package (and one that is up to date) than we were a few months ago. The next step for the PHPwiki packages will be to try and migrate to the dbconfig-common structure for database creation and management. The hold-up there is that dbconfig-common does not support sqlite or sqlite3 at the moment. We use sqlite as the default database for the Debian PHPwiki package to save having to install a heavy-weight system such as mysql or postgresql. Reading the dbconfig-common documentation it seems that it will not be too difficult to add sqlite support so hopefully I will find some time for that over the Christmas break. I wrote some scripts to nicely automate VAP creation with the new madwifi-ng drivers for Atheros wireless cards and posted them to the Atheros mailing list. Kel Modderman picked them up and pointed me towards the pkg-madwifi alioth project where madwifi packaging for Debian is being co-ordinated. It will be good to get involved with that as working with madwifi is a fairly large part of my job at the moment, so it will be easy for me to justify spending time on Debian packaging tasks for it. The key issue we need to sort out here is whether udev rules or ifupdown hooks are the most appropriate way to create the VAP devices, or whether a combination of the two is required.

I don’t think I’ve been really, truly happy for the past 4 or so months. Funny how a book like American Psycho reminds me of this fact. I need to do something about this.

A long time, the common way of working around MSIE CSS bugs was to exploit parser bugs and missing features in it's CSS parser. Recently I read about a cleaner approach; I tested it recently and it seems to work quite well. Judging from the W3C validator, it's compatible with XHTML, too. It's nor perfect, but it probably is the best we can do currently. (And definitely cleaner than the parser-bug-exploits listed before.) The approach is simple: given the string <!--[if IE]> Foobar <![endif]--> in an HTML file, almost any parser will treat this as an comment. Except for Internet Exploiter, which will treat "Foobar" as if it was part of the regular text. Instead of "Foobar", you can load an override stylesheet to work around Internet Explorer bugs. Note that as a side effect, MSIE is not a standards compliant parser for XHTML. ;-) Now for the drawbacks:

• The part hidden from every parser except IE is not validated by tools
• The converse construct (remove --), to hide stuff from IE only, is not valid XHTML
• The conditional needs to be in every HTML file, it's not in the CSS and not cached with the CSS, so sligthly less performant than other solutions.
• It doesn't nest, you can't use comments in this construct
• They are, after all, a hack

For comparison, a few of the older tricks in CSS:

body        background-color: red;  
html>body   background-color: green;  

This is 100% valid CSS, and works because Internet Exploder doesn't support CSS 2 child selectors. But you cannot always use this, since you need to know the parent elements name. Note that this can give you trouble with the upcoming Internet Explorer which will likely have support for this selector, but will probably be "compatible" to some of the known MSIE 5/6 bugs.

body  
  voice-family: "\" \""; /* some browsers have a parsing bug */
                         /* and will ignore the following rules */
  voice-family: inherit;
  background-color: green;
 

This abuses an attribute to confuse the CSS parser of some browsers. A real hack, strongly discouraged.

I've still been extremely busy. Unlike my last fairly incoherent brain dump, this time I've been keeping notes on the things I wanted to write about, so hopefully I won't forget things. Here we go...

Squirrels

Squirrels are absolutely everywhere. They're very fast and very cute. They're kind of like possums, except diurnal. They run up trees with ease, and leap from tree to tree. It's quite cool to watch them. The damn things don't stand still long enough to be photographed (properly), and they don't let you get too close without scampering up a tree or running off. I'm told they're the number one carrier of rabies, so I shouldn't be trying to get too close anyway.

Sales tax

Definitely one of the more annoying aspects of consumer life is the fact that prices quoted are not inclusive of sales tax. This means you always need to part with more cash than you expect. Furthermore, there are some places (some fast-food outlets) that do quote the prices inclusive of tax, so it's really hard to know what you're in for.

Speed limits

I find speed limit signs to be either infrequent or harder to notice, or a combination of both. They're just a black text on white background sign. I'm used to them having the red circle. That said, no one seems to drive at the speed limit anyway.

Ethnicity

Canberra's a very white town compared to say Sydney, but I reckon at least the San Jose area of California blows the socks off what I've seen of Sydney for multiculturalism. You really notice it over here. I've been at the mall in Milpitas, and you can play "spot the white European" (a variation of "spot the Caucasian").

Different responses to thankyou

I think the fairly standard response to saying thanks that I'm used to is "You're welcome". Over here, you get "mm hmm" a lot, which is just strange.

Fractions vs decimal

They love their fractions. Petrol gets quoted at 221⁹/₁₀, which took me a while to realise was 221.9 cents per gallon

Alcoholic beverage labelling

Australia has the concept of a "standard drink". I remember as a kid the advertising campaign "rethink your third drink" (or for women, "rethink your second drink"). The idea being that men could drink two standard drinks in the first hour, and then one standard drink every hour after that, and be at 0.05% blood alcohol, which is the legal limit for Australia. And every bottled alcoholic drink is labelled with how many standard drinks it is. An average beer is about 1.2 standard drinks for example. No such labelling over here. You get the percentage of alcohol per volume, and that is it, which makes self-regulating very difficult. The limit is apparently 0.08%, but I don't think I want to be driving around on the wrong side of the road at that level (or any level really).

Christmas vs "Holidays"

Everyone and every piece of advertising goes on about the "Holidays". Retail people are always wishing us "Happy Holidays" instead of "Merry Christmas". I think this is because there's a large chunk of people who don't necessarily celebrate Christmas.

Tipping

I haven't really come across tipping in everyday life as much as I'd expected, but when I think about it a bit, I'm at work most of the time, so that's probably why. The general rule seems to be to tip "double the tax", so for a bill at a restaurant, or where you get personal service, such as a haircut, you tip. But you don't tip at McDonald's, nor Subway (where I would have expected you would).

That's about all my current observations on living over here. Tomorrow we get our townhouse in Central Park. Our air-freight was finally delivered today. (What was supposed to take about 8 days took 4 weeks). Not sure when we can expect our sea freight at this stage. We bought a couple of mountain bikes last weekend, and if the weather is okay tomorrow, we'll pick them up and ride them home. We've found a really nice off road (as in not shared with cars) bike path, which runs close to home all the way to work. I'm really keen to roller blade to work if it's feasible, and also to cycle in. I'm a bit disappointed to discover after we've been accepted for the place that we're right on top of a toxic waste dump (or something like that). If I grow a second head, you'll know why. The other piece of exciting news is that the bank sent me a debit card, when they'd previously told me they wouldn't until I had a Social Security number (still no sign if that yet). This thing (the debit card) is freaking weird. It's Mastercard branded, and in some places you use it like you'd use a debit card for EFTPOS back home. You swipe, enter your PIN and that's it. In other cases, you swipe and sign a bit of paper like a credit card transaction. In either case, money is sucked out of our checking account. The scary thing is that no one seems to check signatures in this country, and a lot of places will accept credit card payments electronically (like petrol stations where you just swipe the card at the bowser and enter the ZIP code of your billing address and that's it). Oh, and I can use this thing like a credit card for Internet purchases. What bothers me is that the level of authentication is so low, but it's sucking actual money out of our checking account, so with all the credit card fraud around, it's relatively trivial to draw actual money out of a bank account, as opposed to using someone's credit. That said, I just got the card today, and since then, it's made shopping so much easier. It's amazing how many places won't accept a foreign credit card, and I was a big user of EFTPOS back home, and this ability to use a debit card like a credit card means I can essentially continue in that habit. I wasn't looking forward to carrying lots of cash or having to use cheques.

Yes, you can surely tell I am no guru when it comes to describing sounds - maybe that's because in Spanish every letter has one, or at most two, sounds? Anyway, Christian, that's the way to say it - The O sometimes sounds a little bit U-ish, in some places you could see this same sound (coming from N huatl, Mexico's main indigenous language) as huaks. The k (in the x) is usually pronounced, but quite soft - sometimes we just refer to the place as "Oastepec", and nobody gets hurt (or worse, sent to Oaxaca (Huajaka - yes, the X is the only letter whose sound truly changes (and that is because Spaniards used it as a wildcard for S, KS, SH, J - In fact, M xico is pronounced M jico, but never (EVER) spelt that way besides what other Spanish speakers say. The original sound, though, was closer to M shico)))
And, yes, I am also one of many thousands of people longing to ride through the Copper Canyon. It's supposed to be really, really beautiful. Too bad Chihuahua is ~20 hours away from Mexico City :(
And about Oaxtepec's location: It's 60km south of Mexico City, not of Downtown Mexico (that would still be out of the city, but not by that much). The best route is not straight, so from my house (in South-Western Mexico City, ~15m away from where the Cuernavaca highway begins) it's about 90 minutes. The highway is _very_ good and safe, don't worry. I do expect people coming by car from the USA. Here you have a good map of Morelos state. Oaxtepec is on the third column, second row, near Cuautla. Just for completeness sake, just North of Morelos you have Distrito Federal, where Mexico City (partly) lies. The airport, where most of you will arrive is again in the second row, third column (hmmm... A pattern starts to emerge! Lets see what that means...) - We are working so we don't have to explain to you all how to get from there to Oaxtepec, I hope we can provide buses covering the route, at least in the days with most arrivals.

As it seems people are arguing about the revision systems hype, I just wanted to point that CVS rulez. Sure, it does not $some_obscure_witty_feature but, hey, is this feature really useful? I've been using CVS for ten years, coded around 500k source line of code and honestly, it suits all my needs. To tell the truth, I don't even use all of its capabilities. So, CVS can't keep history of a file when you move it. Big deal? After all, renaming a file is just like creating a brand new one. Leave a commit message stating previous name of foo is bar if you really need history and that is it. So far, I never stumbled upon a case where this "feature" limited me or made me loose more than 15 seconds. Same for distributed RCS. Sure, on the paper they look good. Call me a moron but I never managed to use one properly. Even a large but well managed project does not need such monstruosity. Matthew is right on one point: RCS are like languages, geeks advocate languages because of a specific subset of features. But they don't realize 90% of functionality of a language is present in all other languages and the 10 other percents can be achieved with method if you really need them. I personally find very discouraging to learn a brand new RCS for every project I contribute to if upstream has decided this one is cool. If you want users to contribute, use a standard RCS as well as a standard mailing-list system, common language, etc.

I see that plans start getting ready. Bubulle is coming, Wouter is not, and people will soon start getting their plans ready. I am quite excited about it! Of course, I need to devote more time to it, which is hard to me - but I will do it. Oaxtepec rules, you will _love_ the place. I went there again to spend a weekend with Nadezhda - Took some more photos, added them to the album. Debconf guys: Yes, I owe you a full report... It will come. Soon. Promise.
Anyway... I'm excited. Debian friends, Mexican interested friends: What are you waiting for? Join us! Register!
[update] Well, I published this entry, but then realized I wanted to answer to Wouter's question: Why in May? ;-)
First of all, weather: The weather is plainly better in May. Somewhat warmer... The rainy season starts in May/June, but believe me, you don't want to have your laptop out during Morelos State's well known thunderstorms!
Second, I fear of political instability. Next July we will be having our federal elections. We don't anticipate instability such as what we had in 1994 (which, please you all be quiet, didn't even mean violence in the streets), but still... It's better not to have Debconf too near the elections.
Third, prices - Most places are noticeably more expensive during Summer.
And, yes, I know some people will not be able to attend... I am sorry to learn you are one of them. Still, I'm sure some other people would not be able to make it in July.

I have just realized that it’s harder, bureaucracy-wise, to become a Debian Developer, than to obtain a firearms license in Spain. By the way, which one is US English: “realize” or “realise"? (Ah, and darcs rules. At least it’s easy to use.)

When I had almost finished packaging belpic 2.3.13.full, I contacted upstream to notify them of this fact, also mentioning that I'd replaced their build system (which used simple makefiles with many bugs) with an autotools-based one. To which I got the reply that they'd already prepared a belpic 2.40 in the mean time (available at readers.eid.belgium.be), and that they'd switched to SCons as their build system in the mean time. Since I'd spent quite some time on 2.3.13.full already by that time, I decided to continue to prepare that package, and focus on 2.40 afterwards Now of course I'm happy that they chose to use something different from plain buggy MakeFiles. I'm less happy that they chose something based on python, a language which I haven't liked, ever, and don't know much about either, but, well. That's their choice, right? I'm also less happy with the fact that it doesn't properly clean up after itself (requiring me to do ugly stuff in my debian/rules clean target), and that it doesn't work. Grmbl. The build currently fails on a file that hasn't changed, because it can't find a symbol which is defined in a file in the same directory that hasn't changed, either. When I rerun the gcc call with the -save-temps argument and review the preprocessed source, I see that that include file is mentioned, but not included. Something is clearly wrong here. Guess I'll have to start learning python... hrmpf. If you care, belpic is the software required to read out the information and keys in a Belgian electronic ID card. Not of much use if you don't need to read Belgian ID cards, though :-)

I've come back from foss.in in Bangalore, India which I attended as a representative of the Debian project. Foss.in is the event formerly known as Linux Bangalore. It has broadened out to cover all aspects of Free and Open Source Software and is the largest such event in South Asia. I'm told almost 3000 people attended this year. I arrived on the morning of Monday the 28th. The conference organizers put up myself and other speakers at the guesthouse of the Indian Institute of Science. The accomodations were spartan but comfortable. They got considerably less spartan by the next day which was when I figured how the hot water tap worked. I shared my room with Kartik Mistry of Gujarati l10n fame. The first day of the conference was on Tuesday the 29th. The conference was taking place for the first time in a new venue, in several large canvas tents in the grounds of the Bangalore Palace. There were some technical difficulties; when we arrived on site in the morning, there was a line of people atleast several hundred long, waiting to register. Undoubtedly, some of them must have given up and gone home, but the fact that so many people persevered is a testament to the high level of energy surrounding the event. The talks began late too but everyone I spoke to seemed to take it goodnaturedly. I had planned to give two talks. One my standard half-hour "What is Debian?" song and dance. If you went to Debian Day at Helsinki, you already know what that's about and a 90 minute workshop on "Creating Debian Packages." Both were scheduled for Tuesday afternoon. Owing to the earlier delays, the first of these which was slated for 2:30pm didn't start until 4pm. The organizers actually asked me to roll both talks up into one supertalk which was not a problem for me. However, I made one fatal mistake. At lunch time I had noticed one of the food stalls was selling an inviting-smelling cheese pizza. Now, I've been to India enough times to know, "phirangis" simply don't eat dairy-based food in India unless they have some time to acclimatize to it first. I even said as much as I was eating the pizza. Somehow I rationalized that the cheese was pasteurized and well-cooked so the rule wouldn't apply to me but boy was I wrong. I first noticed signs of distress three-quarters of the way through "What is Debian?" I struggled valiantly to the end but then had to excuse myself to head off to the latrines and perform a hard reset of my gastro-intestinal tract. A couple of minutes of vomitting later, though a bit wobbly, I felt well enough to return and start talking about creating Debian packages. I didn't have slides for this one, I just used my laptop to show the innards of binary and source packages, all the various files in the debian directory, debhelper, debconf, and the lifecycle of package maintainence. Unfortunately wi-fi didn't seem to be stretching as far as my location so I couldn't show such nifty things as apt-get source and pbuilder but hopefully my descriptions of these things were impressive enough. There were lots of good questions from the sizable audience and I think we went over 90 minutes. Apart from some BOFs, I didn't really have any formal commitments for the rest of the conference so I spent it holding court at a table in the expo area. I was constantly busy demoing Debian, answering questions, and signing keys. I met Vaidhyanathan Mayilarangam who is a Debian developer, though inactive at the moment. He is going to start getting involved again soon. On Saturday the 3rd, I met the two currently active DDs, Ramakrishnan Muthukrishnan, and Ganesan Rajagopal. Throughout the days of the conference I also met many other figures on the Indian Free Software scene and made a lot of new contacts. That's it for now; in future blog entries I'll write about some of my impressions of the prospects and problems of Debian and Free Software in India. It remains for me to thank: Branden for authorizing funds for my plane ticket, Atul Chitnis and the other organizers and staff of foss.in, Mahesh Pai and Prasad Kadambi who assisted me in navigating Bangalore, and all the other friends of Debian, new and old, I met during my trip.

* 1. General Public License. This program is free software, and may
* be redistributed or modified subject to the terms of the GNU General
* Public License (version 2) or the GNU Lesser General Public License,
* or (at your option) any later versions ("Open Source" code).

* 4. Notices. User hereby agrees not to remove, alter or destroy any
* copyright, trademark, credits, other proprietary notices or confidential
* legends placed upon, contained within or associated with the Software,
* and shall include all such unaltered copyright, trademark, credits,
* other proprietary notices or confidential legends on or in every copy of
* the Software.

Mmhmm.

6. Each time you redistribute the Program (or any work based on the
Program), the recipient automatically receives a license from the
original licensor to copy, distribute or modify the Program subject to
these terms and conditions. You may not impose any further
restrictions on the recipients' exercise of the rights granted herein.
You are not responsible for enforcing compliance by third parties to
this License.

NOTHING TO SEE HERE. NO GPL INCOMPATIBILITY. NO PROBLEM. ENTIRELY HAPPY. NOTHING WHATSOEVER TO BE CONCERNED ABOUT, CERTAINLY NO COPYRIGHT ISSUES WHATSOEVER

#ifdef __cplusplus
extern "C"

#endif

(Note: kernel module code)

#if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,0)
#define get_zeroed_page get_free_page
#define try_inc_mod_count(mod) ((mod && !(mod->flags & MOD_DELETED)) ? __MOD_INC_USE_COUNT(mod), 1 : 0)
#endif // LINUX_VERSION_CODE < KERNEL_VERSION(2,4,0)

#if defined(CONFIG_ALL_PPC) && !defined(CONFIG_PPC_PMAC) && (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0))
#define CONFIG_PPC_PMAC
#endif

ENTIRELY COMPETENT CODE WRITTEN ENTIRELY IN THE LINUX KERNEL CODING STYLE

Note to Fedora Core 2, 3 and 4 users:The default stack size in original FC2, FC3 and FC4 kernels is only 4K. Because some drivers require more stack space to operate properly, we recommend using a kernel with a larger stack size if you experience problems or crashes. You can either rebuild your own custom kernel, or get one of our replacement "stk16" kernels provided for your convenience through the distribution-specific Fedora Core 2i, 3 and 4 links above.

Or you could fix your drivers. Oh, except you can't, because they're actually chunks of the Windows one wrapped in code like the above.

This is truly an excellent piece of work.

Remember, kids, you're only paying $19.99 for code of this quality!

Linuxant are possibly worse than Orkut. And Fujitsu Siemens.

Most workstations are well-served with firewall rules like this:

*filter
:INPUT DROP
:FORWARD DROP
:OUTPUT ACCEPT
-A INPUT -i lo -j ACCEPT
-A INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT
COMMIT

Yes, this is just two rules. Use iptables-restore to load. No need to fire up a complex firewall builder tool for that. ;-) The first input rule enabled loopback traffic, the second allows data traffic for established and related connections (ftp data channels for example). If you want to allow incoming SSH, add

-A INPUT -p tcp -m tcp --dport 22 -j ACCEPT

Oh, unless of course you want to run filesharing and such. You might want to allow incoming connections on other ports as well.

If you are responsible for a network with many "nomadic" users, e.g. students, you'll notice quite often they "forget" to shut down their bittorrent programs. Port filters only work up to a certain amount, so I wanted to try out the "string" match included in recent kernels (e.g. 2.6.14). You'll need iptables 1.3.4, which is not yet in Debian, but you can grab some temp packages from my debian directory. The filter rule I tested was

iptables -I OUTPUT -j DROP -m string --string "BitTorrent protocol" --algo bm --from 0 --to 100

(I don't know if algo bm or algo kmp is more performant, or which performance hit this is going to have on your router. The from parameter could be increased to skip the tcp header, too) Note that you might want to add high-volume "usually good" ports with accept rules in front, like port 80 (www), ssh (22) or mail (25, 465, 110, 995, 143, 993). Be careful with using "state ESTABLISHED", too: the string will be sent over an already established connection, not a new, so this is where you need to apply the filter!

Only 5 months until the next DebConf, in Oaxtepec MX this time. As I read tonight the very well done Final Report for DebConf5, and remember that excellent conference I am already looking forward to the next one. I've been thinking about driving down from Virginis to Oaxtepec instead of flying in. It's a long drive, but maybe I could stop at a few DD's houses and/or travel down with some. In fact, I hope we get a lot of American DDs who have not been able to make recent DebConfs now that it's so relatively near to home. So, time to think about any talks I'd like to give there. The absolute best talk at DebConf5, the one I would most like to emulate, was Enrico's polygen extravaganza. The key things were that a) Dr. Zini rules b) it was a simple talk introducing us to something fun c) large quantities of running code were written on the projector during the talk. I might try to emulate the parts that I can (without taking Enrico-emulation medication) to some degree in one talk, if I can just find the right subject. Since I hate writing papers for talks, I also got an idea for a way to use a talk time slot that will be impossible to write a paper for before time. Gotta hurry, since it turns out talk submission closes amazingly soon: On the 6th!

Seems I despaired too soon about not being able to get good-looking fonts in Emacs. Thanks to Duncan Mak I've been able to compile and install an Emacs with anti-aliased fonts on Ubuntu. Here's the recipe:

• Get the XFT_JHD_BRANCH of CVS Emacs, see XftGnuEmacs on the Emacs wiki, and this page from Deling Ren.
• Get a further patch, from this mailing list post
• Get the emacs-snapshot-debian directory from this arch repository and use it as the debian directory inside the Emacs source tree

Then I had to edit out the build instructions for the -nox variant of Emacs, as it didn't build properly, and to be sure, added this to the Emacs-gtk confflags section of debian/rules :

  emacs_gtk_confflags += --with-xft=yes

After that, fakeroot dpkg-buildpackage -b in the source directory, and some time later, Emacs packages with anti-aliased fonts. There are some limitations: you need to set the font in your ~/.Xresources file like this:

  Emacs*font: Monospace-8

You may also want to fiddle with the Xft.dpi and Xft.hinting X properties to get the fonts to match your desktop environment. OK, so it's a bit of effort, but worth it. Hopefully this patch will mature and make its way into future Linux distributions.