These two really stood out...
Rebooting Airplanes
<"Douglas W. Jones">
Tue, 28 Nov 2006 13:29:42 -0600
In the last few weeks, I've done quite a bit of flying, and twice, now, I've been on planes where they had to reboot.
The first trip where this happened, as we were scheduled to leave the gate, there was a delay, and then the pilot said over the intercom: "We're having trouble with some of the cockpit instruments, so I'm going to force a hard reboot by switching off all the power for a bit." The lights and all other power on the plane then went off, and after a fifteen second pause, on again. A minute later, the pilot said: "That seems to have fixed the problem," and we were off.
I wasn't impressed. As far as I am concerned, this is clear evidence of a genuine design error somewhere in the system.
The second problem happened on Sunday, on a flight back from Amsterdam. On that flight, they had serious problems with the in-flight video on demand system. They tried a "soft reboot" of some kind, and it didn't work, so they then tried two "hard reboots," their term, and after the second try, it worked fine. Their instructions were "until the system comes all the way up, please don't touch any buttons." That alone suggests poor design. The system ought to come up with interrupts disabled on any devices that it's not ready to listen to, after all.
The reboot process took close to half an hour, and watching the displays in the seat backs that were visible from my seat, I could see that they were being rebooted in sequence, about one per second. Furthermore, as each in-seat display was rebooted, it showed the Linux penguin and then a Linux boot script, revealing that each seat-back display was a little Linux system, suggesting that they were all networked to a video server for the plane.
Again, the need for these global reboots is strong evidence that the systems were not well designed,
I wonder if both of these stories illustrate problems with the kinds of graduates we are turning out these days. CS programs across the country are emphasizing high-level courses in web programming, but fewer and fewer students know anything about the fundamentals of parallel programming that underly things. So, in constructing the kinds of distributed applications that show up in contexts like streaming video and cockpit instrumentation, they are working without the theoretical underpinnings needed to understand the problems they encounter.
Mascalls, Manchester, what's the difference?
<"Mark Brader"> Sat, 2 Dec 2006 04:36:21 -0500 (EST)
A British ambulance crew, transferring a patient to a hospital where they had never gone before, drove 200 miles out of their way before realizing that their satellite navigation device had given them the wrong directions.
These reports mention other incidents of sat-nav gaffes, but don't say what the actual error was this time; this shorter one says that the system showed their destination's address as being in Brentwood in Manchester instead of Brentwood in West London.
The patient was not harmed, and the crew has been told they should have known better.
No comments:
Post a Comment