Wow, that's some kind of thread you started... :) I'll respond in general to a bunch of stuff on this list by topic.
COUNCIL MEETING On Sat, Nov 17, 2012 at 10:29 PM, Greg KH <gre...@gentoo.org> wrote: > > So, that's a nice summary, but, what is the end result here? > Speaking as somebody who was there, but not for the council, the summary was the end result OF THE COUNCIL MEETING. The council was asked to set a deadline for everybody with a separate /usr to adopt one of the proposed mitigation solutions, like using a script, initramfs, or whatever. That is ALL that it was asked to decide on, and that was all it did decide on. The whole business about some devs wanting to fork udev came out about a day in advance, and speaking personally it only had a little influence on my vote. The reason I agree with chainsaw's proposal to defer the decision one month was that there seemed to be enough blockers on this that nothing was going to happen for almost another month anyway (best-case), and getting people to move to initramfs or mdev or [nu/eu]dev[-ng]/whatever wasn't actually going to be holding anything up for a while. I'd also have been willing to approve a plan to set a target for something like 90 days after all the necessary tools (like genkernel) were stable and news was sent out. Based on my questions for williamh I did not get the sense that delaying a month was actually hindering the udev project (the established udev). They were encouraged to continue working on their blockers, preparing news items, and so on - everything but having a deadline/go-ahead to break systems that didn't follow the news. So, a bunch of ideas were floating around in the meeting, and I embraced the wait a month option since that seemed to have the most support of any of the options out there. If williamh had identified some actual impact of delay on the udev team I'd have probably pushed for setting the deadline now, but just putting it far enough out there (90 days from genkernel/etc being ready) that all the various teams would have a shot at it. If the udev team gets their news items all worked out and perhaps even sent out (sans deadline) and all the blockers cleared before the next meeting I'd be supportive of setting the deadline around 60 days, but that would be just moral support since I'm not on the council. OFFICIAL UDEV PROJECT I have nothing to do with the new udev project, but I did pass the staff quiz with much help from calchan. :) Read the GLEPs - any Gentoo developer can start a project at any time. That's how things work around here. If I wanted to start a linux kernel fork as an official Gentoo project I could do so tomorrow. That doesn't mean that the new udev will become the default udev, any more than Gentoo hardened will ever become the default experience for new Gentoo users. Gentoo is about choice, and if we have devs interested in maintaining something new then we'll offer that choice to our users for as long as somebody takes care of it. If anybody wants to change the defaults/etc, I'd expect that to get a lot of discussion, and almost certainly a council vote. COPYRIGHT I think this issue is best dealt with on the side - it has no bearing on any of the really contentious points here. I note that the owners of the copyright on udev have announced to the world that (emphasis mine): You may modify your copy or copies of the Library or ANY PORTION OF IT, thus forming a work based on the Library, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions... None of those conditions included keeping the copyright line intact. Anybody can therefore alter the copyright line as they wish, as they have been given explicit permission to do so. They need only comply with the other terms in the LGPL to do so (the most important being licensing it under the LGPL and making the source available. In fact, (L)GPL v3 has an optional attribution clause, and the fact that they made this explicit is because some projects might not want to give out this authorization. So, if you want an official ruling from the trustees we would need to meet/vote on it and perhaps discuss with counsel, but my thinking is that anybody distributing work under the (L)GPL has waived their right to be named on the copyright line of any copies distributed by others, and as far as I can tell I have found nothing to the contrary from any authoritative source. The only way I think you could argue that removing copyright notices for a (L)GPL work is illegal is if you argue that an author doesn't have the legal power to license that right to another. However, I'd still think that promissory estoppel would probably interfere with any kind of recourse - you can't give somebody permission to do something, and then sue them for actually doing it. So, legal or not anybody with standing to sue over this has likely given up their rights to do so. Again, that's my two cents and not a license for anybody to do anything. This topic did come up recently with regard to accepting some other kind of outside work into Gentoo, and as I recall there was some debate over whether the copyright notices could be changed. I'd have to dig up the details - I think the issue might have been mooted before any kind of formal decision was reached... IS THE NEW UDEV A GOOD/BAD/UGLY IDEA Seems like this is the main point of this whole thread, and I don't find it terribly useful to harp on. If people want to start a udev fork more power to them. I'm supportive of that, just as I'm supportive of having systemd in the tree and unit files for as many packages as possible. As projects mature I'd be all for offering them as options in the handbook. Gentoo is about choice. Ditto for a /usr move or whatever else. I think we should have a reasonable default behavior, and as others have pointed out we could use profiles to control a bunch of these behaviors globally (like library install location, and so on). Again, offer the user the choice, and generally be conservative with the defaults. Will all these projects go the distance? Hard to say. I agree that projects inspired by "hate" or whatever often fizzle out, but several high profile forks have stuck around - usually because of conflict over the major goals of the project. The next few years should be interesting, as the amount of vertical integration seems to be creeping up, and if a major Gnome release is systemd-only or whatever that could really bring things to a head. This almost seems like an androidification of the traditional linux distro - where choice still exists but the ability to swap out layers starts to go away unless you stick with a lighter desktop environment. Should make things fun for the toolkit developers when a system might or might not have dbus/systemd/udev/X11/wayland/linux/bsd and who knows what else. Rich
