Donnie’s Blog

Most people that work with computers know about Moore’s Law, which basically predicts that computing power will double every 18 months or so. (I don’t know why it’s called a law when it’s a prediction, but whatever!) Generally, you would imagine that this should mean that a particular computing task should take half the time to complete every 18 months, too.

But, there is one operation that seems to have remained constant over time, thumbing its nose at Moore’s Law, and that is booting up the computer. It seems like that always takes about a minute. It has from the days when I was playing with MS-DOS 3, all the way until now when I boot up Windows XP or the latest Linux. I think this is kind of counterintuitive. Hard disks are faster, memory is faster, and processors are faster, but still, somehow, it takes a whole minute for the computer to boot up.

You might say, “But there’s all these nifty new features built into operating systems nowadays! Like filesystem journaling, and automatic hardware discovery, and clock synchronization, and …” Yeah yeah, but still there’s got to be a reason why it takes about a minute, and not, say, ten minutes, to boot up. “You see, that would just be way too long for people to wait. But a minute is okay.” So, I think that there’s something “special” about getting the OS loaded in a minute or less.

As an aside, I have it on good authority that this is still going to be the case in the 24th century. You see, there was an episode of Star Trek TNG where the Enterprise’s memory got corrupted, and “Oh hey, we’ll just reboot the whole spaceship.” It took a whole hour for the geniuses to figure that one out. If Microsoft were still around then, that would have been the first thing they would have tried. They could’ve just hopped onto a subspace link to India and gotten some good tech support. The episode would probably still be an hour too, although not as exciting.

And how long do you think it took for the Enterprise to reboot?

About a minute.

Here’s an interesting little thought about how to make LinkedIn’s endorsements a little more useful. Just like amazon.com’s review-feedback mechanism, each endorsement should also have a simple agree/disagree option. Of course, this feedback mechanism would have to be constrained to other registered LinkedIn users who also worked with the person being endorsed, to try to minimize the potential for abuse.

This would provide two interesting features:

1. Endorsements would have a rough quality-metric indicating whether they ought to be believed.
2. LinkedIn could also flag users that have a tendency to hand out worthless endorsements.

Would this be helpful for potential employers doing background research? I dunno! But it would at least relieve me of my annoyance with LinkedIn.

LinkedIn gets on my nerves. Besides the fact that it’s just a version of myspace for grown-ups, it’s basically a giant schmooze-fest. No matter where you go on that site, you bump into all kinds of kissing up.

(more…)

Who says white guys can’t jump?

(If you don’t know who this guy is, his name is Didi Senft, and he is a fixture in the Tour de France.)

C++ templates are tricky.

In C++ you can declare typedefs inside of classes or class templates, and then you can use them to make your life a little easier. Well, mainly it’s easier with classes; templates make you jump through some extra hoops.

(more…)

Note to all software distributors:

When you create a .zip file and a .tgz file for distribution, please do not assume that the .zip file does not need to include the UNIX build-files.

It is very annoying.