Kanotix seeks stability, sidux follows unstable

Posted Dec 15, 2006 8:05 UTC (Fri) by codermattie (guest, #42239)
Parent article: Kanotix seeks stability, sidux follows unstable

Debian is a great distribution, I ran it for years. In contexts where
risk is unacceptable it is a good choice still.

However for a developer or user who wishes to selectively track the upstream for both features and bugs the protracted stabalization on a branch model has been fraught with difficulty. The success of the 2.6 series in comparison to the previous 2.4 and 2.2 kernels is a good distro-agnostic example.

When using debian it is difficult to diverge from the distribution mainline without learning a complex package format/system. Forking debian unstable to stabilize and release as a new distribution is now classic. Without a better generalized solution for merging it will have the same problems
as debian, just with newer versions.

Merging technology is fundamental to the linux development model. Git
IMHO points towards a sane future. I can maintain my system kernel as
a private branch from a tag such as 2.6.19 and then pick single changes
that fix or improve the kernel without importing a large body of changes
where it is difficult to predict or analyze the effects and interaction of those changes.

It would be interesting to see a distro that was based on the git ideas.
Gentoo IMHO has moved the farthest towards flexible branching while integrating mainline updates.

Kanotix seeks stability, sidux follows unstable

Posted Jan 8, 2007 0:49 UTC (Mon) by cleary (guest, #41669) [Link]

"Forking debian unstable to stabilize and release as a new distribution is now classic"

Just to clarify, sidux does not "fork" debian unstable, it sticks as close as possible to pure debian sid, including using the same repositories.
What sidux attempts to provide is timely patching/package holding/highly visible warnings for problem packages in dist-upgrades, current kernels - stable and rc (for the game) as well as a community to help when things go awry.
It provides open source livecd-from-scratch building tools, as well as various other live cd related utils.
It (will) provide a fast hd installer in a similar way to Kanotix, which at best I've managed to boot and do a full install in under 10 mins (using the toram cheatcode).
It's aimed at desktop users who want bleeding edge software with the least amount of pain.

The points you raise about merging technology/Git look very interesting, my experience with code repositories is currently very limited so I'm not in a position to comment in any detail :(