April 2009 Archives

The Detroit Free Press brings us the fabulous story of — I think, it’s not entirely clear what sport we’re discussing here — ice-hockey fans, and the multiple-arms race between them and officials. The officials, you see, are trying to stop them throwing octopus onto the rink. Read the article for absolutely no enlightenment at all, in the grand scheme of things.

(via MetaFilter. The other links there are worth exploring, too. Though be warned that the New Scientist video has possibly the most annoying voice-over ever.)

“Computer spies” (sic) have broken into the Joint Strike Fighter project and made off with several terabytes of data, reports the Wall Street Journal.

This would be the same F-35 Lightning II project of which one of the major partners is the British Government, themselves not unfamiliar with losing important data. Though, to be fair, we tend to just leave it lying around or bung it unencrypted into the Royal Mail, rather than be so unsporting as to force people to have to bother stealing it.

But hang on — a few years ago, there was a bit of a hoo-hah about this very project. It transpired the Americans weren’t quite convinced the UK understood concepts of data security (well… durr), and thus had mild reservations about handing over the complete source code for the F-35’s software. The British, not letting anything past them, realised this meant that between a British pilot flicking an insouciant British finger onto a British bomb-release button in a British plane on a British mission, and that specific (British, obviously) bomb, there could conceivably be some dastardly Yankee code along the lines of:

if (bombReleaseTriggerDepressed) {
    switch (isBombGodFearingAmerican) {
        case (hellYesBlowThemAllToHell) {
            releaseMunition(YeeHaw);
            break;
        }
        case (heckNoLimeyBastards) {
            errorMsg(insufficientPermission, escalateUACLevel);
            break;
        }
    }
};

Understandably ticked off, the British threatened to pull out of the project and… oh, I don’t know, build TSR.2 or something. In the end, some sort of fudge was reached whereby Blair could announce (Wikipedia again):

“Both governments agree that the UK will have the ability to successfully operate, upgrade, employ, and maintain the Joint Strike Fighter such that the UK retains operational sovereignty over the aircraft.”

…which is one of those odd phrasings that means less the more you think about it.

Ironic, then, that it turns out the easiest way of extracting the long-sought-after F-35 source code may have been to partner with the Chinese all along. Indeed, there are two possibilities here. If MI6 were being clever they could have just outsourced their industrial espionage, presumably at a huge saving to the British taxpayer.

Alternatively — and back to the Wall Street Journal —

“Foreign allies are helping develop the aircraft, which opens up other avenues of attack for spies online. At least one breach appears to have occurred in Turkey and another country that is a U.S. ally, according to people familiar with the matter.”

I think we all know which other country that’s likely to be.

Oh, arse. They got the code, then left it on a bus, didn’t they?

Inspired. Danny MacAskill from dave on Vimeo.

Crivens.

Geek post warning: feel free to skip this, folks, it’s intended mostly for Google.

As of April 2009, eSATA for Macs is a bit flaky. Well, it probably doesn’t have to be, but cheap eSATA is flaky. Specifically, products based on the Silicon Image 3132 chipset cause random but repeatable kernel panics on Mac OS X systems. Here’s what I know so far:

Until recently, I’ve been editing off a bunch of 1Tb Firewire 800 hard drives. They’re Lacie RAID 0 units, and they work well enough. 65Mb/sec read or write is enough for the sort of editing I’m doing (ProRes 422). They’re a bit big and heavy to be truly portable, but it’s reasonably easy to bung one in a bag with my MacBook Pro and edit HD footage away from base.

However, they’ve a reputation for being unreliable, and indeed one of mine is starting to get a bit flaky. They’re also full, or at least ‘more full than I’d like.’ So a few months ago I started looking for alternatives.

I could have pushed old stuff into a Drobo, but they’re very slow indeed. Not just ‘too slow to edit off,’ but ‘too slow to move a typical 250Gb project onto.’ An hour’s archiving activity is fine, but a day is ‘deep backing store’ territory. I might explore the new DroboPro, but the proprietary nature of the filesystem concerns me. Hardware RAID 5 and swapping out entire disk sets may be more cost-effective.

FireWIre 800 is pricy, in part because it’s hard (in the UK) to source bare enclosures and fill them with one’s own disks, which is what I used to do with FireWire 400. It’s also not that quick, in the grand scheme.

No, the future is eSATA, the external version of the connector that hooks up our (internal) hard drives these days. Quick, cheap, reliable. Well, pick any two. Or maybe one. Read on:

A couple of months ago I bought a Sonnet eSATA expresscard for my MacBook Pro. They sell two, and I bought the cheaper of the options. I also bought a remarkably cheap trayless eSATA drive enclosure, an Edge10 DAS401. For the money it’s really not bad. I’m unimpressed by the shielding around the (internal) power supply, and it could be a little quieter, but it’s a solid enough box and easy enough to use. Thumbscrews everywhere are a nice touch. Cheap, but cheerful, and it has a port replicator so all four drives hang off a single eSATA cable.

The card and the enclosure talk to each other absolutely fine. Software RAID is out of the question: I managed to brick a pair of drives by trying it, having to reformat them via Windows before I could connect them to OS X again without triggering a kernel panic. But a single drive (Hitachi 1Tb units) pushes about 95Mb/sec read/write, which is ample for my purposes. The same drive mounted internally in a Mac Pro pushes 105Mb/sec, so the external overhead is acceptable.

It’s also trivial to pull a drive from the enclosure and shove it on a shelf, in a WiebeTech Drivebox. This appears to be my new equivalent of ‘putting the miniDV tapes in a shoebox.’

There are, however, some problems. Specifically, the eSATA card drivers are flaky. They mostly work, but mounting or unmounting a drive, or connecting or stopping the card, can cause kernel panics. Boff! Gone. Game over. Once it’s all running it works just fine, but hot-swapping is a no-no. In fact, the best approach is to shut everything down, configure the system, then power it all up. Which is fine, if mildly inconvenient.

However, since I bought the Expresscard Sonnet have posted a dire warning on their site:

“Temporarily not recommended for Mac systems with greater than 2Gb of memory.”

Yikes. My MacBook Pro has 4Gb.

The problem, I believe, is the chipset, which is based on the Silicon Image 3132 product. Most cheap eSATA systems seem to use this design, and the drivers, whether your card is from Sonnet or Lacie or anyone else, come directly from Silicon Image. And they stink. Luckily, for me, they seem to work OK. Ish.

At least, they do in the MacBook Pro. I recently bought an end-of-line 2008-model Mac Pro, and that’s a whole different animal.

The Edge10 box came with a bundled PCIe eSATA card, a tiny little thing with a dashing red board and a chip inscribed ‘Silicon Image 3132.’ Oh, crap. Sonnet’s equivalent card looks identical to the one I have, and also carries the dire warning about Mac systems with more than 2Gb RAM; my Mac Pro has 10Gb.

Sure enough, the Mac Pro throws a kernel panic at the merest hint of hooking up drives in the Edge10. It does, occasionally, work. Sometimes it’ll push data solidly for 20 minutes or more, but usually — and certainly ‘eventually’ — it panics. Hard lock. Reboot. Bastard.

Frustratingly, the way out of this appears to be to invest in a Sonnet Tempo E4P card, which is another £250 but uses some chipset that doesn’t involve the dreaded Sil3132.

Alternatively, Silicon Image might update their drivers. They haven’t touched them in almost two years, but you never know.

Until they’re fixed, however, the only sure approach is to stay away from Sil3132-based setups with Macs. They either don’t work at all, or aren’t reliable.

Forest light 1, originally uploaded by Quernstone.

Have a picture.

Easter weekend consisted of flat cleaning (hurray for council dumps!), Sunday brunch for eight, and a jolly jaunt to Puck’s Glen, in the Argyle forest park. Followed by a two-hour traffic jam on the way back past Loch Lomond.

Ah well, can’t have it all.