Ubuntu Wiki

Agenda

• • ubuntu-core-dev application for Soren Hansen
• • impromptu tracker review

Members Present

• • Matthew Garrett
• • Mark Shuttleworth
• • Matt Zimmerman

Summary

After a wandering discussion about the challenges facing Ubuntu Server, community and Canonical developers and the project in general, Soren's application was approved. Please welcome him as our newest core developer!

Mark Shuttleworth raised Tracker for discussion, which is the subject of an ongoing email exchange between the Board, upstream developers and Ubuntu developers. There remain some bugs affecting Tracker which make it unsuitable for release with Ubuntu 7.10, but the developers involve hope to have them resolved in time to keep it in the release. As a contingency plan, Tracker can be easily disabled by default via gconf if the problems are not corrected in time.

Log

09:16   sabdfl  https://wiki.ubuntu.com/TechnicalBoardAgenda
09:16   mdz     MootBot: hello?
09:17   sabdfl  ah, that's why soren was pinging ;-)
09:17   soren   :)
09:18   mjg59   Hm. Go ahead without mootbot?
09:18   mdz     yes, just pulling up soren's application
09:18   mdz     https://lists.ubuntu.com/archives/motu-council/2007-September/000360.html
09:18   soren   https://lists.ubuntu.com/archives/motu-council/2007-September/000371.html
09:19   mjg59   Hm. Shall we try to grab pitti and kees?
09:19   mdz     thanks
09:19   soren   kees is around.
09:19   soren   Hang on.
09:19   mdz     sponsor feedback was received from pitti, kees and ScottK
09:19   soren   That's not right.
09:19   soren   ScottK is not -core-dev.
=== ScottK commented, but not as a sponsor.
09:20   mdz     oh, I see
09:20   soren   ScottK: Thanks though :)
09:20   mdz     soren: so you've had a lot more time to spend on Ubuntu recently, by virtue of joining the Canonical team
09:20   ScottK  You're welcome.  You've been a lot of help to me.
09:20   soren   mdz: Sure have.
09:21   mdz     soren: how has that changed your views on Ubuntu?  do you have a different perspective on how the project works?
09:21   soren   mdz: Than before joining Canonical? Not much. I've gotten more aware of deadlines and such, though.
09:22   soren   I've always had a good feeling of the distro as a whole, I think.
09:22   sabdfl  do you have any new insights on the interaction of Canonical and volunteer contributors?
09:22   keescook        quick note from me: I've sponsored a few of soren's main uploads, and they've always looked good to me.  he's actually taught me a few things about some of the server packages in main.  :)
09:23   soren   sabdfl: It's certainly been interesting to get a first hand view of the inside of "the mother ship".
09:24   mdz     soren: it's difficult to know how things look to part-time contributors when living and breathing Ubuntu as a day job.  is there anything you've learned from this experience that you think would help MOTUs and prospective MOTUs to contribute more easily to Ubuntu?
09:25   soren   mdz: Well, as part of the day-to-day environment and also as part of the motu-uvf team, it's getting more and more obvious how we view it differently.
09:25   soren   When you only have so much time to spend on Ubuntu, it can be hard to see the big picture and only focus on the few packages you have time to worry about.
09:26   sabdfl  i need to commute to a home network. +1 from me for soren, on the back of excellent feedback from MC and all who commented, I'll be back online within 15 I hope for trackerd discussion
09:26   soren   I haven't spent a lot of time reflecting upon that, though.
09:26   soren   sabdfl: \o/ Thanks for that!
=== ogra cheers for soren
09:26   mdz     sabdfl: I don't have anything further to say about tracker; the email thread is...
09:27   mjg59   soren: How do you feel we're doing on the server?
09:27   soren   I think it might be a good idea to more thoroughly communicate what the purpose of the different slots in the development cycle is for.
09:27   soren   mjg59: Not good enough :)
09:27   mjg59   soren: Heh. Ok, can you expand on that?
09:27   soren   mjg59: It's interesting, really:
09:28   mdz     soren: we want it to be easy to contribute to Ubuntu in simple ways, like maintaining a few packages, and yet it's essential that that work be in harmony with the overall movement of the project, especially releases.  I'm wondering if there's anything you think we can do to improve that, in terms of communication or documentation for example
09:28   soren   mjg59: We're sometimese viewed as too much of a desktop OS, while the distro we're based upon has the exact opposite problem, and in the server area we really differ very little from Debian.
09:28   mdz     soren: pitti has been working just this past week on cleaning up exactly that, and I think the current documentation is miles ahead
09:28   soren   mjg59: There's definitely some work to be done to change that perception.
09:29   mjg59   soren: So it's mostly a perception problem, and not a technological one?
09:29   mdz     (different phases of development)
09:29   soren   mdz: Agreed. The MOTU team and especially the hopefuls seem to feel UVF as a hindrance in some way.
09:29   soren   mdz: Like an annoying, pointless obstacle.
09:30   soren   Sorry, that came out wrong.
09:30   soren   Not all, certainly.
09:30   soren   "some of".
09:31   soren   But I totally understand where it's coming from. I shared that feeling when I was "only" working on Ubuntu on a volunteer basis.
09:31   soren   mjg59: Well, Debian has a pretty decent reputation on servers. We really ought to have inherited that.
09:31   mdz     soren: do you think it is a practical problem or one of perception?
09:31   mjg59   How much of that is because universe tends to be less integrated to begin with?
09:31   soren   mdz: Both, but mostly perception.
09:32   soren   mjg59: Good question.
09:32   soren   mjg59: ...for which I don't have a good answer :)
09:32   mdz     it's a difficult thing to generalize whether an upstream release, with potentially a huge amount of new code and bugs, is appropriate for inclusion in Ubuntu, and UVF is one primitive way in which we try to assess that.  Do you think we could change it to be less burdensome without sacrificing its effectiveness?
09:33   mjg59   In main, it's pretty clear that we have UVF to concentrate on stability and integration for the remainder of the cycle
09:33   soren   mjg59: Definitely.
09:33   soren   mdz: Hm... I'm not sure.
09:33   mjg59   I guess there's an argument that the integration aspect of that is less applicable to universe
09:34   soren   mdz: The motu-uvf team is a good idea. It keeps the worst crack out, surely. :)
09:34   mdz     mjg59: perhaps somewhat...I don't want to promote an image of universe as a set of packages which don't matter as much
09:34   mdz     mjg59: because that downplays the valuable work that developers put into it
09:34   soren   mdz: ...but someone made the point earlier on that any motu ought to be able to make the call that motu-uvf does.
09:35   soren   Some of it stems from the fact that universe is "not supported", so it doesn't matter if it doesn't really work.
09:35   soren   Allegedly, of course.
09:35   mdz     what inevitably happens without such a limitation is that someone uploads something at the last minute, which is broken, and results in either releasing with a broken package, or a big headache for the release managers, archive admins and others to try to push a last-minute fix through
09:36   soren   Because most of our millions of users don't care where it comes from. If a package in universe doesn't work, Ubuntu doesn't work.
09:36   mdz     yes, there's something to be said for user expectations here
09:36   mdz     our policies for maintenance and support are far beyond what most desktop users expect
09:37   mdz     folks coming from the Windows world are used to getting their applications on CD or by download from a website, and never getting updates
09:37   soren   Interesting point.
09:37   mdz     in which case the distinction between main and universe doesn't matter much to them
09:38   mdz     I certainly use plenty of software from universe, because it meets my needs and doesn't put me at significant risk
09:38   soren   Sure.
09:38   soren   Are you suggesting we don't provide 18 months of support for non-lts releases?
09:39   mdz     not at all
09:39   soren   If it's way beyond what "the random users" expect, and we expect the "corporate users" to use LTS's anyway, it seems like a lot of work for little gain.
09:39   mdz     only agreeing that the distinction between main and universe is artificial for a certain (common) class of desktop users
09:39   soren   Oh, right.
09:39   mdz     consider ubuntu-backports
09:40   mdz     people who use backports are most interested in having fresh applications
09:40   mjg59   soren: How well is main serving our server users? Are most of them ending up using universe as well?
09:40   mdz     they accept the tradeoff in maintenance and support
09:40   soren   mjg59: Currently, yes.
09:41   soren   mjg59: We're reevaluating a lot of software to find out if it should be promoted.
09:41   mjg59   soren: Do you feel that that's something that needs changing?
09:41   soren   mjg59: Yes, indeed.
09:41   mdz     mjg59: even if they use packages from universe, I think the distinction is much more relevant to them, since they need to be informed about which software will receive security updates
09:42   soren   mjg59: I realise it puts more strain on the security team (and the rest of us, too), but there's a lot of really useful software in universe that really deserves proper support.
09:42   mjg59   I think that's pretty inevitable, yes
09:42   mdz     sure, but that's one area where a distinction needs to be drawn between Canonical and Ubuntu
09:42   soren   mjg59: I think it scales fairly well, though.
09:42   mdz     in large part, it's Canonical who backs the commitment to security updates by providing dedicated resources
09:43   mdz     and so those decisions need to be made with a view to what's best for Canonical and its customers, more than the inherent merit of the software
09:43   soren   mdz: Of course.
09:44   mjg59   I agree that we're failing to communicate the different levels of support. People enable universe and then never think about it again.
09:44   soren   mdz: My point just is that there's a lot of really useful software in universe that Canonical's paying customers would like to use.
09:44   soren   mdz: ...but if they're not properly supported, they might go elsewhere.
09:44   mdz     yes, there are some specific cases where we should make adjustments on that basis
09:45   mdz     but I'm wary of saying that software "deserves" support and therefore should be promoted to main
09:45   soren   Right. Bad choice of words.
09:46   soren   I mean that it's software that's in a good enough shape to not cause too much of a burden and also useful enough to make a difference to paying customers.
09:46   mdz     mjg59: no more questions from me, say when
09:47   mjg59   soren: So, arguably the two main issues facing us on the server are (1) people not taking Ubuntu sufficiently seriously, and (2) people not understanding the differing support levels, and this reflecting badly on the distribution as a whole?
09:47   soren   mjg59: No, our main issue, i think, is hardware support in LTS releases.
09:48   soren   mjg59: That's really what most people point out to me as our major shortcoming.
09:48   mjg59   Ok. How can we deal with that?
09:49   soren   mjg59: Well, the obvious solution is more kernel developers.
09:49   soren   However..
09:49   mjg59   Heh
09:49   soren   It would be much more interesting to get to the point where we can make the hardware vendors do it themselves.
09:50   mjg59   Hardware vendors are moving towards not being interested in doing so
09:50   mjg59   Intel have been making it pretty clear that they're not planning to
09:50   soren   Really? Becuase we're too good at it without them?
09:51   mdz     I do not think that a shortage of kernel developers is the main issue.  they can't do much about enabling arbitrary hardware without access to it, and can't efficiently do testing across a wide range of platforms
09:51   mjg59   No, because they end up with a large set of develoeprs doing nothing but backports, with most of the benefit going to the Linux vendors rather than them
09:51   mjg59   Though I suspect backporting hardware support is a discussion better suited to Boston than here...
09:52   soren   Agreed.
09:52   soren   mdz: That's also true.. It's a tough problem to tackle.
09:52   mdz     the only reasonable long-term solution to this problem is to enable portable (across Linux versions) drivers to be written
09:52   soren   I'm not sure that will happen anytime soon.
09:53   soren   AFAIK the kernel developers don't believe much in "API stability at all cost".
09:53   mjg59   The only way that's likely to happen is if the distributors cooperate on a porting layer
09:53   mjg59   It's unlikely to change upstream
09:54   mdz     I suppose it's also possible that Linux will one day stop regressing massively, and allow new kernels to be provided for older releases
09:54   soren   Ironic, really.
09:54   mdz     but I think portable drivers are more likely
09:54   soren   Since a considerable percentage of kernel developers (ie. upstream) work for distributors.
09:55   mdz     (as unfortunate as the current opinions upstream may be)
09:56   mdz     I don't know whether any of the current userspace driver work makes this more feasible
09:56   soren   I wouldn't know.
09:56   soren   I try to steer clear of anything in the kernel that deals directly with hardware. :)
09:56   soren   It's scary.
09:57   mdz     and we're getting a bit off-topic for the meeting
09:57   ogra    yeah, what about soren :)
09:57   soren   Oh, right.
09:57   mdz     mjg59: any other questions regarding soren's application?
09:57   mjg59   No, I think I'm happy now
09:57   soren   Wow.
09:58   mjg59   Vote?
09:58   mdz     ok, votes then
09:58   mjg59   +1 from me, packaging skills sound good and has a solid understanding of where we are and where we want to be in the server market
09:58   mdz     <sabdfl> i need to commute to a home network. +1 from me for soren, on the back of excellent feedback from MC and all who commented,
09:58   mdz     +1, demonstrated understanding of project policies and rationale, positive feedback from core sponsors and MOTU
09:58   mdz     soren: congratulations and welcome
=== ogra cheers
09:59   soren   \o/
09:59   soren   Woot!
09:59   ogra    welcome soren :)
09:59   soren   mdz: Thanks!
09:59   mdz     I've updated Launchpad
09:59   mdz     mjg59: do you know what sabdfl was talking about regarding tracker?  I'm not aware of any decision before the TB or a need for discussion outside the ongoing email thread with TB, Ubuntu developers and upstream represented
10:00   ScottK  soren: Congratulations.
10:00   soren   mdz: He's back now.
10:00   soren   mdz: He just joined #u-d
10:00   soren   ScottK: Thanks!
10:00   mjg59   No, I wasn't aware of any further tracker discussion being scheduled
10:01   sabdfl  hey
10:01   mjg59   sabdfl: Good timing
10:01   mdz     sabdfl: you mentioned something about tracker?
=== sabdfl needs to file a bug about suspend/resume on X60 w/ gutsy
10:01   sabdfl  first, a story
10:01   sabdfl  it took me a while to find a cab
10:01   mdz     sabdfl: you may or may not be aware, we discussed it is some detail at the previous TB meeting
10:01   sabdfl  when i climbed into it, he had a screen in there, where they usually show adverts and comedy
10:01   sabdfl  his said:
10:01   sabdfl  ERROR
10:01   sabdfl  Corrupt file
10:02   sabdfl  \WINDOWS\SYSTEM\etc etc
10:02   sabdfl  mdz: want to discuss today's email exchange re tracker here, or are you happy with the decision options?
10:03   mdz     sabdfl: I'm satisfied with the direction of the conversation, and it (unlike this meeting) includes representatives from upstream and the Ubuntu kernel team
10:03   sabdfl  ok
10:03   sabdfl  nothing more from me then
10:03   mdz     sabdfl: when we discussed it at the last meeting, we presented our concerns to upstream and interested developers
10:04   mdz     upstream acknowledged our specific concerns, and was confident they would be resolved in time
10:04   sabdfl  ok
10:04   sabdfl  are they excited to have their bits in the spotlight?
10:04   mdz     we accepted this, with the proviso that if things didn't go according to plan, tracker could be trivially disabled by default with a gconf key
10:05   mdz     I wouldn't want to speak on their behalf, but they do urge us to stick with it and believe it is a good option for us and for our users
10:05   sabdfl  ok
10:05   mdz     the kernel issue seems to be the main one at this point
10:05   sabdfl  that's me then
10:05   mdz     because we don't know the cause yet
10:06   mdz     ok, that's it for agenda then.  is there any other business?
10:06   mdz     ok, adjourned then.  thanks, all
10:07   sabdfl  thanks, and good night