Is Ubuntu a Compatibility Nightmare for Debian?

An anonymous reader submits "Following Friday's release of Ubuntu Linux 5.04, Ian Murdock, founder of the Debian project, told internetnews.com: 'Ubuntu's popularity is a net negative for Debian.' He explained: 'It's diverged so far from Sarge that packages built for Ubuntu often don't work on Sarge. And given the momentum behind Ubuntu, more and more packages are being built like this. The result is a potential compatibility nightmare.' Ian suggests a method for averting crisis on his blog."

Comments

Ubuntu is a good thing.

[natrius]

Packages made for sid don't even work on sarge all the time without pulling in extra packages from sid. That's the same thing that happens with packages meant for Ubuntu. When you have different sets of software installed on various computers, one single package isn't going to work correctly on all of them unless you're willing to mix packages from different repositories.

I don't think it's really fair to say that Ubuntu is a net negative for Debian. It's definitely a net negative for sarge, since very little, if any, of the work put in to Ubuntu has trickled down to sarge. However, it's good for Debian as a whole because when the ball gets rolling for etch, most of the work will already be done. Ubuntu puts out stable releases for three of the four release arches for etch, so I doubt much extra work will be needed there, although I don't really know that much about what additional work would be necessary.

Sure, Ubuntu's existence has various downsides, such as the proliferation of deb packages provided by developers that only work on Ubuntu, but would those people have made Debian packages in the first place? The packages are merely a byproduct of Ubuntu's popularity, and more people using Debian and Debian derived distributions is definitely a net gain for Debian. I don't see why he would write off all the benefits that Ubuntu provides while focusing on a few issues that are negligible IMO.

The packaging issue is one that's never really going to go away. On his blog, Ian cites software developers and ISVs as reasons for unifying Debian and Ubuntu packages. All free software developers have to do to get their software packaged by Ubuntu is request it. The Ubuntu packagers work fairly close with the Debian developers to make sure that the work trickles down to Ubuntu proper as well. For commercial software it's a bit harder, but that's one of the things to deal with in the Linux ecosystem. Like I said before, packages made for sarge wouldn't even necessarily work on woody. You have to target specific sets of available software, or just distribute binaries that install the software based on various LSB assumptions.

Ubuntu Sarge

[Stalin]

http://www.ubuntulinux.org/ubuntu/relationship/doc ument_view

"Ubuntu makes a release every six months, and supports those releases for 18 months with daily security fixes and patches to critical bugs.

As Ubuntu prepares for release, we "freeze" a snapshot of debian's development archive ('sid'). We start from 'sid' in order to give ourselves the freedom to make our own decisions with regard to release management, independent of Debian's release-in-preparation. This is necessary because our release criteria are very different from Debian's.

As a simple example, a package might be excluded from Debian 'testing' due to a build failure on any of the 11 architectures supported by Debian 'sarge', but it is still suitable for Ubuntu if it builds and works on only three of them. A package will also be prevented from entering Debian 'testing' if it has release-critical bugs according to Debian criteria, but a bug which is release-critical for Debian may not be as important for Ubuntu.

As a community, we choose places to diverge from Debian in ways that minimize the difference between Debian and Ubuntu. For example, we usually choose to update to the very latest version of Gnome rather than the older version in Debian, and we might do the same for key other pieces of infrastructure such as X or GCC. Those decisions are listed as Feature Goals for that release, and we work as a community to make sure that they are in place before the release happens."

So, who cares that it isn't compatible with Sarge? Is Sarge really compatible with Sid? I think not (if you are sane). Shouldn't Ian be saying that Ubuntu isn't compatible with his "componentized Linux" (http://www.progeny.com/products/components.html)?

Re:hypocrisy

[noahm]

2. Debian has changed its release architectures after Sarge so that Etch is not slowed down by unknown, exotic and/or obsolete architectures.

That is most definitely not the case. There was considerable discussion on the debian-devel list following the release team's proposal to limit the etch release to 4 architectures. While the proposal may still be implemented, it also may still undergo significant changes. People have been suggesting all sorts of counter proposals to try and keep all the architectures in sync.

Personally, even though I've run Debian on MIPS, MIPSel, Alpha, and Sparc (all of which would be dropped under the Release Team's proposal) I still support the proposal and would like to see architecture support scaled back a bit. There are those, however, who feel that Debian would be giving up too much if they were to drop some platforms.

noah

X.Org and Debian

[Overfiend]

I'm sorry Google doesn't appear to have crawled the Debian X FAQ yet.

Here's a link that may help, straight into the Subversion repository where Debian's XFree86 packages are developed:

What are Debian's plans with respect to X.Org and XFree86?

The current text of that FAQ entry follows.

Because the XFree86 relicensing came at a time when Debian was trying to stabilize its XFree86 packages for the sarge release, there was some question among Debian's X Window System package maintenance team (the "X Strike Force") -- and much speculation among Debian's users -- as to what direction Debian would take.

There was never a serious proposal to attempt to ship anything other than XFree86 4.3.0 in sarge, so work on that continued while discussion on the debian-x mailing list took place. The following represents the consensus reached by the X Strike Force, without objection from the mailing list subscribers (among whom number many interested Debian developers and users).

In June 2004, Fabio Massimo Di Nitto, the XFree86 package release manager for Debian sarge and sid, started a thread to discuss the future of X Window System packages in Debian for an open discussion between users and the Debian package maintainers.

The discussion spanned nearly one hundred messages from over a dozen participants, practically all of it constructive and very useful to the Debian maintenance team. The outcome of the thread was farly clear to everyone: Debian will move away from the XFree86 tree as soon as possible after the upcoming stable release due to its license issues (see above).

The XFree86 package maintainers are committed to providing support and assistance to the Debian Security Team for the XFree86 4.3.0-based packages than Debian will ship in sarge. That is, our abandonment of the XFree86 Project, Inc., as an upstream source of code does not mean that we will abandon our commitment to the users of our production release.

Futhermore, there was near-consensus that Debian should switch to the X.Org source tree, with the goal of migrating to the modularized tree over time. We expect that the monolithic X.Org distribution will be modularized in a piecewise fashion; as that happens, we will "switch off" the building of packages from the X.Org monolithic tree in favor of the modularized components that become available from freedesktop.org.

While moving from XFree86's monolithic tree to X.Org's is a relatively simple technical transition of itself, the transition to a fully-modularized set of packages will take longer -- indeed, an unknown amount of time which depends on the speed of upstream's progress -- but we expect the process will bring the packages' quality to a higher level, thanks to the introduction of a fast release cycle for each single component. We expect to "modularize" two parts of the X.Org distribution immediately: XTerm and Xprt (the XPRINT server). XTerm is independently maintained by Thomas Dickey, and the xprint.org version of Xprt is already separately packaged in Debian.

With these changes, it will also be easier for the Debian user community to have a broader choice in X servers. At present, the Debian XFree86 package maintainers intend to support only the XOrg X server (which is based on XFree86's). The X Strike Force does not plan to discourage other people from packaging others. Debian developers that file intent-to-package notices (ITPs) for other X servers are asked to strictly cooperate with the X Strike Force to maintain similar packaging standards, simplify the bug handling on shared components (like X libraries) and discuss future changes and improvements.

As of this writing (March 2005), packaging of the X.Org X11 distribution is underway in the X Strike Force's xorg-x11 Subversion repository.

Re:Thank you!

[Overfiend]

Now here's my concern: I have no idea who Deadbeast is (There isn't a top level page -- which is wierd), how do I know it's not just wishful thinking?

Well, I'm Branden Robinson, and deadbeast.net is my vanity domain. If you're easily amused, you may want to look up deadbeast's WHOIS record. :)

I'm glad the pointer to the FAQ helped. I admit I didn't foresee that sarge would take this long to release when advocating that Debian stick with the tried-and-true XFree86 4.3 (even if hacked up and patched to support more hardware than stock 4.3 does), but the trouble is, the longer the sarge release drags on, the more disruptive it would be to try to cut over to X.Org. So I continue to believe that the best solution to this problem, as with many others, is to kick sarge out of the nest so we can focus our full attention on disrupting the hell out of unstable for a few months. :)