Trapped in a frame?
Break free now!

Status of Debian Sarge

Some considerations on the present status of Debian GNU/Linux 3.1 "Sarge"

This article was first published on my Blog but has been moved here for ease of use.

Tonight I've been installing sarge at a friend's HP laptop together with my friend lefant. Both of us have some experience in installing Debian on several platforms and in convincing Mac OS or Windows users to migrate to the "Universal Operating System" of the future (for most of us of today, of course *smile*). Although the friend of us was quite astonished by the ease of installation and the power of the system, we've been facing several problems during installation and I must say that while the advantages of using Debian - apart from the political and philosophical implications of being Free Software - become immediatly clear to an interested Mac OS or Windows user, Debian Sarge is still a step from being stable - and Woody is simply too old to be recommended.

The development both of the Linux kernel and the Debian system has advanced a lot in the last few years. Kernel modules can easily compete with its propietary Mac OS or Windows competitors, a lot of printers are supported by CUPS, foomatic, ghostscript and The Gimp, most graphiccards will work under X11 without troubles and many drivers for cams, scanners and other devices are being intensively developed. Above all for a modern laptop the 2.2 Linux kernel seems really outdated. Apart from this upgrading from Woody to Sarge when it becomes stable might become a nightmare for an avarage Mac OS or Windows user - remember the change from Potato to Woody. While people using Sid (unstable) all the time might not mind, this should be considered when recomending Debian. My friend had installed Woody by himself and read the installation manual - that's what he found on the Debian site, he said. But as Windows user he's used to modern software like Mozilla Firefox - which is taking over Microsoft's Internet Explorer -, The Gimp 2.0 or OpenOffice. Both the Gnome Desktop and KDE have made big changes since Woody was finally released - and also then Debian 3.0 wasn't up to date any more, as we know.

Apart from small bugs like translation errors (German), changing the keymap back to 'us' from 'de' during installation or not updating the Grub menu.lst after installing a new kernel, there where some bugs the "avarage Desktop user" (sic!) might not be able to deal with: 1. The option "Desktop" during the partitioning process created a 5 GB partion for /home while only leaving 2 GB of space for / - that's where the main system would have been installed.

Unfortunatly, this is not enaugh for Gnome and KDE, which are installed by the "Desktop" option in tasksel. When installing Knoppix e.g. the installer complains if the partition is too small. Similarly the Sarge installer should propose the minimal necessary amount of diskspace for installing a Desktop Environment (including the most popular applications) plus something for the root filesystem instead of creating such a big home directory. Simple mathematics (i.e. formal logics) don't seem to be very usefull in this particular case. 2. Synaptic (a graphical package manager) crashed while trying to install a new kernel - no comment! 3. cfdisk crashed with a permanent beep while starting in single user mode - what made it a bit more difficult to delete the existing home partition and create a new partition for /usr and a much smaller one for /home.

As lefant has brought a copy of his Sarge mirror at home on his ibook, we expected the installation to take little time, but his mirror is already 2 weeks old and we run into troubles because of some dependency problems, the biggest one with exim4 neither wanting to install nor to deinstall. No idea what exactly happened, but after having configured the system and running tasksel again, apt-get seems to have cleaned the local apt-cache and fetched all the packages again, which of course took some extra amount of time... :-(

My conclusion is that there should be a transitional solution for Debian Sarge, as I am quite sceptical that the release date is coming closer pretty quick. Updating the website providing some pointers and advice what is the recommended procedure before Debian Stable 3.1 is released would be a first step. The Debian project consists of currently more than 2,000 developers and maintainers primaraly contributing to the free software community and computer users in general in their leisure time. We cannot expect that everything works perfect out of the box and I want to stress that I am very happy whith the status of Debian and the achievments of the last period.

Lot of effort has been done to release Debian Stable 3.1 - maybe too many. I have been already happy with Debian Stable 3.0 (aka Woody) - but in this article I wanted to mainly focus on the "pressure" that's being created by Mac OS and Windows users, as well as "the competition" (i.e. other distributions). As many I disagree with the idea that there is "one distribution for every kind of person" but would like to see one real "Universal Operating System" instead - and I believe that Debian has the power to become that kind of OS!

Most friends of mine would be unhappy with a system made "by hackers for hackers", only aimed to work for those who already know "how it works". They don't want to care too much for there system - they simply want it "to work". Although this might not come into conflict with our principles of hacker ethics, wanted to understand what's going on in order to be able to control, we must deal with those wishes. Knoppix and other similar Live-CD distributions show the way: You insert the CD; your hardware is (almost) fully detected and automagically configured; you get an impressive preview of what can be done with your new system; you can easily install it on your harddisk.

This is no criticism of the Debian project, as most of these distributions are based on Debian GNU/Linux and the Debian project itself is moving into this direction. I just think we should wait a bit with a new release, but raising the mood of both the developers and the users by offering them a kind of "transitional solution" and publicly announcing the status of Debian GNU/Linux 3.1 (aka Sarge). Of course, the decision must be taken by the project members itself and I am not sure if such a step wouldn't create a wave of discontent.

From what I can say - looking from outside - it would be a pitty if an obviously buggy system was announced being "stable" only because it has already taken more time. Because of Debians legendary long release cycles the "public eye" observes it very intensly - maybe even more than other distros - and expects a lot from the project. Hence, souch a step IMHO would be a mistake and could lead to a discreditation of the project.

I admit that the described bugs can already be fixed within this week and that even my "wishlist" could be partially satisfied, but it doesn't seem very probable. All this said I think that after a successful release and a break for everyone involved in the project, there should be a brief discussion followed by a "tendency decision" or - depending on the debate - a full decision on the Debian release policy. In software development there are clear targets - with priorities - for a next release. Fixing a release date that shouldn't only be "eagered to please" or approached, but that must be performed in time - just as the so called "method" of "Extreme Programming" (which seems to be in mood right now, but also has many "enemies" within the free software movement because of its commercial implications and a concept often being "far to reality" of the avarage software developer).

If the tasks that the project had resolved to do for the next release cannot be performed in time, the release date shouldn't be postponed, but priorities should be set and a new version released. Unlike e.g. Ubuntu developers, Debian developers usually are not payed for their work. They work for the project voluntarily, which means that they cannot be ask to finish a particular piece of work "in time". But this also means that their responsibilities are self-inflicted and their must be a minimal amount of warranty (despite all warnings provided by the Debian project and the GPL - which at least in Europe are invalid).

I don't want to point users to Ubuntu, Hispalinux or Libranet or to one of the existing free GNU/Linux derivates of Knoppix, but to be able to tell them: "Go to debian dot org and get the latest stable release there!" This might become a bit difficult for a person that is not well experienced, when confronted with the discribed (or similar) errors. On the other hand, "the avarage Desktop user" (an assumption made by myself, based on many of my friends) simply expects to get the last stable version of Mozilla Firefox, The Gimp, Gnome or KDE she/he knows from the internet or from friends or PC magazines when installing a new system.
Of course, we will never satisfy everyone's wishes and there will (hopefully) always be potential for changes, but I think this issue is one that should be adressed quickly as I have the impression that finding no solution to it may lead to a paralization of the project - or at least delay or retard its progress. Six-monthly release dates for Debian GNU/Linux!!! :-)

I know I'm breaking a taboo and many won't agree with me on this. And I know that I've never really been involved in free software development and will probably change or at least adapt the opinion expressed herein when joing my first big project. But I simply felt like writing those lines and I see no reason why my considerations should make Debian GNU/Linux (not talking about other distributions here like Debian Hurd or Debian FreeBSD) less secure, M$-style crap, unstable or unusable at all. This does not mean of course that you may not be of a different opinion - and that I would discuss this delicate and sensitive concern with you. ;-)

Now I'll be watching the academy awards... :-D


ps: this is the first time i'm using the "Mail-to-Blogger" system, so please cut me some slack if formatting is messed up! *smile*

Copyright © 2002-2005 by Pablo Hoertner. All rights reserved.
Verbatim copying and distribution of this entire article is permitted in any medium, provided this notice is preserved.
Last modified: Sun Feb 27 05:33:34 CET 2005 by Pablo Hoertner