One of the things that was up in the air yesterday was the possible creation of an "enterprise" Debian distribution. An eWeek story last week cited sources "close to Mandriva, Progeny and Turbolinux" that indicated the Linux Core Consortium (LCC) members would be doing a new enterprise distro based on Sarge. Murdock confirmed today that there's some truth to the rumors, but the details are a bit off.
Yes, indeed, work is being done on a enterprise Debian distro by Progeny and a few other Linux players -- but Mandriva and Turbolinux don't appear to be on board at this time. Murdock confirmed today, however, that "we have been talking with Debian companies, basically in pursuit of the original LCC mission -- to build a common core that had both Debian and RPM components, so you could build a Debian or RPM distro using it."
If the startegy is to get this common core into as many distros as possible, given that the majority, even vast majority are based on Debian, had to get Debian and Debian derivatives involved. Simultaneously, it became clear that there wasn't a lot of excitement around building another RPM core... which seemed to be the way the LCC was going.
Murdock also mentioned that this new Sarge-based distribution would be LSB 3.0 compatible, and RPM and Red Hat Enterprise Linux (RHEL) compatible. Since there are a number of differences between RHEL and Debian, not least of which is the actual kernel shipped by each distribution, I asked Murdock how they could hope to accomplish this and still remain compatible with stock Sarge.
We are still evaluating precicely what RHEL and RPM compatibility means, we do know that it's important, because RHEL is widely used, RPM is widely used. In order to make Debian a stronger commercial platform we have to deal with these kind of compatibility issues. that's one of the aspects of the story, which we're still working through. As I said to Steven, we are going to have inital compatibility in the short term, but for precicely the reasons that you said, we're still trying to figure out what it means in the long term, but suffice it to say that we know it's important.
Murdock also noted during our conversation that he hadn't planned to be talking about any of this just yet. "We weren't expecting to be talking about it (the DCC) early, we were planning to announce at Linux World Expo" until the eWeek story appeared. (At least Progeny isn't trying to sue anyone for leaking some info to the press ahead of time...) I'm looking forward to further details on how they plan to remain compatible with Debian and RHEL.
I asked Murdock whether the Ubuntu folks had been approached about this, and whether they were going to be part of this effort. Murdock said he'd "love to see Ubuntu involved in this. They're a pretty important player in the Debian world."
However, Murdock said Ubuntu is not involved in the DCC at this time. While prepping for yesterday's blog, I asked Murdock about the fact that the Ubuntu Foundation was launched without any mention of Debian, Murdock said that he viewed it as "a sort of declaration of independence from Debian."
Debian is increasingly just another upstream source for them. Personally, I think this is a huge mistake on their part--sure, they have lots of momentum, but that's largely because Debian seemed to be faltering for a little while. But now that sarge is out there, the real momentum is behind Debian again, though Ubuntu still has momentum on the desktop side. If I were them, I'd continue focusing on that. I certainly wouldn't be so eager to unhook from the Debian train just yet.
Murdock also disagreed with Benjamin Mako Hill's assertion that "Debian is not even binary compatible with itself" in reference to Ubuntu's reasons for not pursuing binary compatibility with Debian.
I have no idea what Mako means when he says Debian is not even binary compatible with itself. When the core packages are the same (as they were between sarge and sid for a long time while sarge was moving toward release), I mixed and matched between stable, testing, and unstable all the time, and so did a lot of other people. Of course, when testing or unstable are making a transition to new versions of core packages (as they are now), it's not quite that simple, and you have to do a recompile. But that's because those are the development branches. There's a good reason for the fact that they aren't compatible--it's called progress. I see no similar technical reason behind Ubuntu's decision to introduce incompatibility. The only rationale I can come up is business related.
We already have source-level compatibility, namely configure, make, make install. We moved beyond that a decade ago when we introduced dpkg, and APT takes that many, many steps further, such that we can potentially have binary-level compatibility only dreamed of in the UNIX-era dark ages. It's absurd to go back to that kind of thinking given the tools we have right in front of us.
Murdock also said that he's willing to give Ubuntu the benefit of the doubt. "I've been critical, but it's been difficult for them to respond with tangible actions, but now that Sarge has been released, it will be interesting to see what they do."
Finally, I asked whether members of the Debian project had been approached about the DCC. According to Murdock, "the nice thing here, to get Debian involved, all we have to do is say, 'you lead, we'll follow.'"
We're saying that Debian Sarge is the center of gravity. We are tackling some problems, filling some gaps that Debian does not currently fill by itself. One of the centerpieces is LSB 3.0 compliance. Think about it this way... think of this as the commercial interests around Debian aligning with each other and getting involved in the existing Debian community to help fill some of these gaps in a way that is done as part of Debian, not going off and reinventing Debian. Just simply getting involved in the community, and like any member of any community we're working on problems that are important to us but doing it in the context of the exist community.
When the Linux Core Consortium was announced, I wondered whether it would be successful. Efforts to unite players in the Linux community around a common "core" have floundered twice now -- previously with UnitedLinux, and now with the LCC.
There were some serious problems with the UnitedLinux effort, and I'm not just referring to the involvement of SCO. The players in UL were all, more or less, pursuing the same market. Each player had its own distinct distribution, and most of the development burden was placed on SUSE.
The LCC also had a major problem -- namely, the lack of a common heritage. The mission for the LCC -- to create a "core" distro that they could all use -- was a bit complicated since each company had its own distinct distribution that would have been hard to mesh with the others.
The DCC, should it get off the ground, has a distinct advantage. One, the vendors are not pursuing the same audience. Progeny, Xandros and Linspire, for example, really aren't competing for the same customers. While Xandros and Linspire are both desktop-oriented, they're aiming for different customer bases. It's unlikely that they'll be fighting over the same customers -- at least in the very near future.
Secondly, and perhaps most importantly, they all share a common heritage: Debian. There may be adjustments necessary for the companies to share a common core, but they'll be minor compared to the adjustments that would have been required for Mandriva, Turbolinux and Progeny.
I've said it before, and I'll say it again -- I think the best strategy for Linux is to unite behind a community-based project like Debian, where each vendor can build a solution that meets their customers' needs while remaining compatible with the core distribution. The market also needs a platform that is not controlled by a single vendor.
With any luck, Ubuntu and the DCC can work together to make this happen. There's a lot of momentum behind Ubuntu at the moment, and I think both projects would benefit greatly by working together.