Sex, Drugs & Unix

Home » Archives » April 2005 » Debian is dead

[Previous entry: "ESR as financial adviser"] [Next entry: "HTML O' the Day"]

04/26/2005: "Debian is dead"


In an interview with InternetNews, Debian's new project manager, Branden Robinson basically says nothing.
He "won't get in the way" (of sarge).

More telling is this fun little pull quote from Bruce Perens:

"The biggest challenge Debian has to face is to go to regular releases rather than 'when it's ready,' which has turned out to be 'once in three years,' " Perens told internetnews.com. "This will happen with or without Branden, and he has stated that he'll keep out of the way of the release managers."


But wait, wasn't one of the fundamental tenents of CaTB
Release Early, and Release Often":

Early and frequent releases are a critical part of the Linux development model. Most developers (including me) used to believe this was bad policy for larger than trivial projects, because early versions are almost by definition buggy versions and you don't want to wear out the patience of your users.


Debian has several failures, which I won't repeat here. Ubuntu isn't much better, and the development process for both distros is flawed in the extreme.

So, the flock, with the exception of the truely dedicated, is abandoning Debian for Ubuntu. RedHat seems to have finally pissed-off the assembled crowd, who's immediate reaction was "Fedora Core", a distro with more inherent value in its left pinky than all of RedHat's $2B market cap.

Still, for the hobbist or professional developer, all these distros are flawed in the same way, they make an excuse for source code availability. For a development desktop, Debian is deeply flawed. Its package manager likes to get in the way, pulling in header files and "development libraries" that are completely out of touch with the rest of the system. For the true depth of the depravity, try compiling redboot on a debian machine.

Still, debian's deepest flaw (one it shares with Redhat/Fedora as well as Ubutu) is that its doesn't allow an EZ rebuild from a known, stable source tree. Imagine the state of the linux kernel if huge bits of it shipped as binary objects that you downloaded from one of several webservers. Thats a Microsoft/Sun ideology!

Why in the world, in this day and age when 2GHz CPUs and 512MB of RAM are common, would anyone prefer a system that they can't build from scratch? Being able to easily build from source is the sine qua non for a unix, or unix-like system. Anything that gets in the way is flawed.

In this regard, Gentoo continues to hold together. If you gotta run linux, its G2 for my money, and the debian folk could learn a lot from portage. If you've attained any level of linux modicum, gentoo is your "distro" (as an aside, the fact that Gentoo is really an "undistro" is likely the reason to keep it out of the hands of young and impressionable minds. Better to learn *nix somewhere else first, then graduate to a distro that won't get in your way.

However, for my money, if you can run it, FreeBSD still holds together much better.

For those of you about to write-in about OpenBSD, don't bother. I don't associate with rip-off artists. And speaking of theft, check this little bit of larceny from Y2K:


CL: You can buy VA cheaper now.

ESR: It's a buying opportunity, yes :-) And I can say that, because I'm not in the United States. If I were in the United States right now, the SEC wouldn't let me say that


For anyone who groks even the slightest edge of SEC rules, the SEC won't let you say it if you're out of the country, either. (Not as a US citizen or resident, anyway.)