Congrats to Sarah on her new acquisition. Horner, welcome to the club.

So how does one go about asking Google for the name of the triple-breasted whore from Eroticon VI in the Hitchhiker's Guide to the Galaxy?

I suppose the only answer is carefully and out of visual range of anyone sensitive.

(It turns out her name was Eccentrica Gallumbits)

I put in a best effort attempt at driving off a close family friend yesterday evening. A friend not of my family, but rather of Claire's.

We had some friends over for dinner and at some point the conversation swung dangerously close to religion. I'm normally quite happy to let these thing slide. To each his own, and there's usually not much point in debating these things because, ultimately, they come down to a matter of faith. Either you believe, or you don't.

But every now and then people either throw out ludicrous statements which beg to be "questioned", or they try to justify their faith using something that sounds vaguely like science. The latter drives me particularly batty and last night I just couldn't let it slide.

By the end of the "discussion" we'd covered a range of topics including what the term "God" means, whether or not any religion can claim to be better in some way than any other, mediums (of the channeling kind), the number of the beast, the decimal number system, whether or not irrational numbers are necessarily unrepresentable finitely in a given number system, finite and infinite state automata, Kirlian photography, the bible code, the religious right in America, the Qur'an, the validity (in a statistical sense) of drawing conclusions from a single event, and a dozen other topics I can't remember.

No blood was shed but it almost certainly came closer than it should have.

Stumbled across numberspiral.com a while back and have been meaning to post it ever since.

The basic idea is to think of all the positive integers as being evenly spaced points along a curve and then to lay that curve out so the perfect squares line up along a straight line heading to the right.

All sorts of interesting patterns start to emerge. The inset picture is one where all primes on the spiral have been marked.

It turns out there are some interesting relationships between selected classes of curves. The "curve" containing the perfect squares is really the set of integers given by the formula x(x+d) with d=0. The curve highlighted in the inset picture is the set of primes given by Euler's prime-generating formula: x(x+1)+41.

Ah crap. User error.

Don't laugh, or I won't tell you.

I had a test for a timeout condition in our DBAT scripts, at which point the Xen domain was hard-killed. The timeout was set to 30 minutes. Introducing the Xen build pushed our build time up to around 28 minutes. Throw in some time to copy and boot the image and, well, yeah ... egg all over my face.

In my defence the reason for having a timeout at all manifested within hours of "correcting" the situation: a new Xen kernel build stopped midway through prompting for some unset kernel options. DBAT is supposed to be automated, so the last thing you want is something like this requiring manual intervention before you pick up and build fixed sources.

As an aside (an as a lousy attempt at distracting you from my daftitude) the kernel build seems to prompt repeatedly. By the time I'd noticed the build was "stuck" the log file containing the build output was 2.1GB in size. Eish.

Our DBAT (daily build and test) runs on top of Xen. Basically we have a chunk of disk space put aside formatted with a filesystem with everything our build process requires. A cron job fires every hour or so, copies this image over, mounts it as a loopback device and boots it using Xen. An init script fires at boot time, sucks our DBAT framework out of our Subversion repository and hands control over to it. This proceeds to check out, build, install and test each of our projects. The results are mailed to the team and the image is discarded so the next run is clean (nothing magical here).

Or at least that's how it works when things go well. Of late we've started seeing failures. Initially they were intermittent. One out of every dozen or so runs would fail. And it was only the failures where we built Xen itself along with a few custom kernels (we only did this once a day because the build is much longer than most processes and the project isn't changing very quickly). So it looked to be load related. So we've enabled the Xen build on every run and the failures poured out. We're still getting the occasional successful run through but 9 out of 10 runs fail outright. Never quite in the same place but always midway through the Xen build. This may be statistical (Xen consumes by far the bulk of the time spent running the DBAT) or it may really indicate something load-related (The Xen project build is pretty intensive).

Today I set things up to boot the DBAT on a raw device rather than mounting it as a loopback device (nothing like a combination of sudo, cron and dd to make me break out in a cold sweat). I had high hopes for this. But, alas, no joy. Same behaviour (although in a slightly different part of the process).

I have one more card to play: building a bog standard domU kernel for the DBAT (we're booting a slightly customized kernel). If that proves to be as joyless as the last few attempted fixes then I may have to do some kernel deep-diving with kdb.

(Updated: 20:01 - Thanks to Sam for the reminder) I know this sounds like hardware. I thought so too so I subjected it to 6 hours of testing of memtestx86 and then, after it survived without any errors, rebuilt a separate machine from scratch only to watch it do the same thing.

In case you haven't been paying attention things over in the US of A have gotten a little bizarre in recent years. The rise in popularity of the Daily Show is one of the few pieces of evidence suggestive of any light at the end of the tunnel.

Although preaching to the choir (the only people who will really appreciate this are members of the complement of the set of those to whom this is most applicable):

http://www.aboyandhiscomputer.com/Greetings_from_Idiot_America.html

The IM hijinks are pretty amusing too, but steer clear if you're easily offended.

Every now and then this guy produces something astounding. He takes a simple phrase and cuts to the core of it with a simple drawing.

He's managed to illustrate that you really got to me, but despite that you have become a ghost, haunting the dark corners of quiet evenings.

One of the hardest lessons Life has forced upon me, time and again, is that every so often, no matter who you are, you have to choose one.

Just thinking of you makes me smile.

I've noticed a pattern when it comes to getting to know a tool or technology. For a while much of what happens is like magic. But slowly, as you're forced to dig into one area after another and really come to terms with how it all hangs together the curtain is pulled back and the cogs and levers become apparent. Eventually, given enough time, I reach a point where I feel completely at home with whatever it is and it's easy to forget that someone new to the game still has that learning curve ahead of them.

I think to an outsider this often seems like magic. I've lost track of the number of times either myself, or someone else very familiar with a tool, technology or even just a large body of code, has been presented with a seemingly insurmountable problem or inexplicable behaviour only to nail it down without even blinking.

I suspect this lies behind a term I hear frequently, usually in the context of something like sed or awk: incantation. It seems particularly common in the world of Unix, probably because it's possible to put together extremely powerful , and as a result complex, command line "scriptlets". Often these end up squirrelled away on a web page or in a stuff.txt and pulled out as needed, much like incantations from a magician's spell book. But if you're prepared to invest some time in man and play a little with the tools then slowly the mystery is eroded and you edge closer towards truly mastering a tool. Oddly enough, it's usually the case that the more proficient you become with these tools, the less likely you are to describe a particular invocation as an incantation, even though what you're doing may be an order of magnitude beyond the code snippets you hauled out before you reached "enlightenment".

Arthur C. Clarke said it best:

Any sufficiently advanced technology is indistinguishable from magic.

Long ride today. Much needed. A few hundred k's to work the knots out of my back and clear my head. Some alone time to think.

A friend dropped a bit of a bombshell this weekend. And I honestly don't know how to react. When I think about it I feel angry. Hulk-smash angry. Maim-kill-destroy angry. But more than that I feel powerless to do anything or to have done anything. I'm not sure there is anything you can do, other than be an ear or a shoulder.

So Andy and I took to the open road, starting out with the Pegasus crowd for breakfast in Stellenbosch before moving on to Franchoek, heading over the pass towards Villiersdorp, cutting back to Hermanus and finally heading out to Napier via Stanford. At that point we about-faced and came back via Caledon.

The country out there is my favourite type of riding country: wide open spaces with no buildings in site and long winding roads devoid of cars.

The day was a scorcher, so hot even the sheep knew to get out of the Sun, but the roads were empty and the runs between stops gave me some much needed time to mull things over. Rides like this work wonders for me because long straight stretches give me time alone to think and sweeping curves and occasional lunatic driver force me to focus on what I'm doing which seems to distract my back muscles long enough to relax.

The Henning clan suffered an increment at 12:20am today.

Oliver Niss was born and, if his face is anything to go by, he'd also have preferred a few hours more sleep.

Roses are red.
Violets are blue.
I'm mechanically inept,
And I can't write poetry

Dropped my bike off at Peter this morning so he could take a look at it for me. I just want someone who knows what they're doing to give it the once over and I don't think I know anyone who knows bikes as well as Peter does.

Dropping my bike off at Peter is always a little bittersweet. On the one hand I know it's going to come back like a new bike. But on the other hand, Peter's a little like my grandfather was and I always feel a little ashamed that he can't see his reflection in the bike. This morning he ran through his list of "hmmmms" and I felt horrible about each one (tires looking a little soft; chain needs tightening; seat needs to be washed).

But Peter's a great guy who clearly loves what he does (which is probably one of the reasons he does it so well). He phoned just now to tell me it's done and went on and on about the the long list of things he'd done to her. And he's only charging me 200 bucks for it. I think he's doing himself a disservice (but I'm not complaining). If you need someone to do any work on your bike Peter's the guy to ask.