Licquia Blog (old)

Got the latest Dr. Dobbs Journal just now. It comes with an eval copy of Windows Server 2003. I saw the slogan for the ad campaign (see the title) and about lost it. That phrase can be abused so many ways:

• "...after paying for it, you'll need to!"
• "...'cause your server just won't be the same afterwards."

OK, I'll stop now. But I still want to know why Microsoft does this to themselves. First "Where do you want to go today?", now this.

Just so everyone's aware, I had nothing to do with a spam from some Korean outfit. They just decided to use one of their harvested addresses as the Return-Path for the mail, and I was the unlucky sap to get the honor.

This isn't hard to do, actually; if you want a demo, let me know, and I'll send you some mail from yourself.

I suppose a lot of us "amateur journalists" get to make weird proclamations regarding journalism, so why not?

Nick Petreley, tech journalist extraordinare, has solved the software installation problem Debian people (and certain other apt users) don't know they have. It's simple, really: just write a packaging system that allows software to install on any distribution of Linux, quickly, without conflicts or "package hell" or special dependency resolvers like Debian's apt.

This miracle is to be created by somehow combining some of the ideas of the current dynamic linker (except better), some of the ideas of Gentoo (without the time-consuming build step), and some of the ideas of the "ltdl" library from GNU libtool (except, again, better). He drops a teaser at the end of the article that implies that he'll have all the answers in the next article.

Wow. Who needs engineers when you have journalists?

This article is kind of like me commenting on reporting the news by saying something like "hey, if only journalists would just report the truth, they'd have far fewer problems". (Hi, Mike!) Well, duh. Reporting the "truth" is the hard part.

Similarly, Petreley reduces the software installation problem to its hard parts and then punts: "just solve these little problems, and we're cool". It's really easy to write articles that imply that we engineers are too stupid to think of such an easy thing, especially since no one will expect him to actually produce working code. Sorry, he can't; he'd love to raise the dead and heal the sick, but he's got this column deadline, you understand?

Just to be fair, the problem I want to see him solve is this: how does he write a dynamic linker that can either predict the future or write missing code on demand?