[00:15] "Enlightenment E17 Running On Wayland" ooh [02:47] bryceh, RAOF: hrm so theres not much use spending significant effort getting this lts backports stuff into the archive before got 12.04? [02:47] since its going to be the crazy backporting 100 packages with suffixes deal [02:49] we cant have xserver-xorg-video-quezacotl if we dont know what its actually called :P [02:49] xserver-xorg-video-intel-quezacotl that is [02:50] Sarvatt: Right. We just need to ensure that we *can* get quezacotl into the archive. [02:53] well the kernel does it and we were told this was how it has to be done so i guess thats how it's going to be? :P [02:54] imagine we'll have some metapackage stuff to work out, would be good to get it working and tested before release at least [02:54] i can probably integrate that into edgers, make a whole new stack an optional install [02:54] xserver 1.12 for instance [02:54] except yeah, libs.. [02:57] Sarvatt, yeah anything that can be practiced in edgers would likely help [02:59] well i guess at least libxi, libxrandr will have to be updated too, whatever the server needs [03:00] i thought about the upgrade issues the other day.. [03:02] but now is not the time to remember what it was :) [03:02] +try to [03:04] tjaalton: very much agreed about the time issue :) [03:05] libs will have to be upgraded, thats a given [03:05] its only the subtle releases like 1.9->1.10 where you dont, that wont happen for long [03:05] 1.6->1.7, ugh [03:07] i'm slightly worried about pixman because it has a history of affecting things like notify-osd without fixing it and is required quite often [03:12] ok so the issue is that you can't "upgrade" to Q from 12.04.3 without cleaning up the backports first [03:12] since it would be a downgrade X and kernel-wise [03:12] is that an issue? [03:12] not to me [03:12] packages could be versioned lower than Q i mean [03:12] just something that should be made clear [03:13] ~Q [03:13] ah [03:13] well you lose functionality [03:13] that's not a problem, because the package names will be different [03:13] so if you need the hw support you'll regress [03:13] and there's no way to detect that [03:13] tjaalton: what are you doing up at 5am? :) [03:14] so, installations made with .3 should not offer the upgrade to other than the next lts imho [03:14] Sarvatt: youngest giving trouble for the past 2+h [03:14] oh once P is out yeah that'll be an issue since it'd offer to upgrade to Q but versions would be higher, think i get what you mean [03:15] then it was my turn and it took 15min to get her back to sleep :) [03:15] yeah the upstream version part, doesn't matter what the package revision is [03:20] err i meant R instead of P there, sheesh [03:20] anyway, if we can make our lives easier by adding policy, then lets do that :) [03:22] hey yeah the 12.04.3 -> Q issue might be worth mentioning, can one of you add it to the blueprint? [03:22] on the subscription? [03:23] also makes me wonder about problems relating to 12.04.3 -> P (skipping Q) [03:23] uh [03:23] description [03:23] s/P/R/ [03:23] :) [03:23] in the linked wiki page [03:23] oh [03:24] hadn't noticed that [03:24] 12.04.3 -> P should be fine, right? [03:24] because precise's kernel will be newer than any of the 12.04 backports [03:24] well it is P :) [03:25] it's just, e.g., 12.04.3 -> M that have potential to be problematic [03:25] hmm [03:25] versions messed up there? [03:26] yeah I meant 12.04.3 -> R :-) [03:26] err...right :) [03:26] bryceh: should I add a "questions" headline for drive-by comments like this? :) [03:26] 12.04.3 -> R isn't a valid upgrade path, though [03:26] 12.04->Q is fine, thats the only one thats fine, 12.04.5->R will be the problem because it'll offer Q [03:27] you can only upgrade 6 months at a time or 2 years at a time [03:27] * Sarvatt is totally slow at responding [03:27] * RAOF is confused. [03:27] holy crap, me too [03:27] .4 is the latest there'll be [03:27] 12.04.5 is R X stack [03:27] newer than Q [03:27] tjaalton, yeah [03:27] .2 (Q stack), .3 (R stack), .4 (S stack) [03:28] And then we drop everyone on the T stack. [03:28] then the next one is another lts [03:28] tjaalton: .3 should be Q still, .4 R? [03:28] Sarvatt: .2 is released after Q [03:28] .1 just before [03:28] https://wiki.ubuntu.com/X/Blueprints/LtsPointUpdatesForXorg [03:28] Sarvatt: the point releases are offset from 6-month releases by 3 months [03:28] it's all there :) [03:29] broder, right, so from 12.04.3 you could go to Q (downgrading your X stack), or to R (correct X stack, but not a valid upgrade path) [03:29] Ahem. Then after .4 we transition everyone to the T stack, because P is still supported for a long while after .4 [03:29] RAOF: yep [03:29] As the various stacks drop out of security support. [03:29] I put a table with all these into the wiki blueprint page (and dates!) [03:30] so basically only vanilla P and maybe P.1 is upgradable to Q, the others -> T [03:30] yeah it's just something we'll have to document [03:31] I can't imagine someone who's already opted into 3 point releases is suddenly going to wake up one day and decide to go non-LTS ;-) But users are crazy, anything could happen. [03:31] bryceh: if it was possible to retain the backport-R stack when you upgraded from 12.04.3 -> Q, the user could reboot and upgrade to R immediately [03:32] but i don't know if that's going to be possible. it depends on how the packaging magic plays out [03:32] bryceh: users can still do that, just that update-manager shouldn't offer it. no way to protect them from sed & apt-get :) [03:32] That'd be different. [03:32] broder, right, the only problem would be if they installed 12.04.3 because the Q stack was faulty on their hardware or something. But we get into weird corner cases at this point. [03:32] broder: Hm, but you raise a good point; there needs to be some update-manager trickery to ensure that things work correctly. [03:33] bryceh: but if they kept using backport-R stack after they upgraded to Q, in the same way that you can have old kernel versions lying around... [03:33] yeah they just should reinstall at that point, it'll keep /home intact anyway [03:34] There's a policy question here - on a 12.04.x → 12.10 upgrade, what stack should the user end up with? I'd like to say they should just end up with the 12.10 stack, regardless of which 12.04.x they upgraded from. [03:35] * Sarvatt was seriously hoping we could just upgrade via -backports and include that on the livecd and not do package name mangling [03:35] * broder 's backporter head hurts from that idea [03:35] RAOF: sounds like something update-manager could handle, removing the old stack before upgrading [03:36] Sarvatt: Right, that's one way of doing it; we could also have the 12.10 stack declare breaks/replaces on the appropriate bits. Either way, it needs deciding before the 12.10 release. [03:36] but then they risk losing hw support [03:37] in fairness, that's always a risk on upgrade, regardless of whether you're going forwards or backwards [03:38] just declare it as a non-supported upgrade path.. you're on your own if you do that [03:46] I think this is a huge corner case. The use-case we're talking about is for people who installed 12.04.3 *in preference to* 13.04, and then want to upgrade to 12.10. [03:46] very true, was about to say that too [03:47] They probably don't want to do that at all :). At worst, they *actually* want to upgrade to 13.04 or 13.10. And we'll support upgrading to 14.04. [03:47] right [03:51] so say, ubuntu-desktop depends on xorg [03:51] how does that get fixed [03:51] i'm not seeing it [03:52] :) [03:52] xorg-q Provides: xorg [03:52] install ubuntu-desktop-q :) [03:53] Which is one of the things we need to check pre-12.04: we need to ensure that nothing outside the stack has versioned dependencies on pieces of the stack. [03:53] if there are going to be other backports due to pixman changes or such [03:56] we're going to end up having deltas to debian on all of X doing this aren't we? [03:57] only on the next lts when it's cleaning up the mess :) [03:58] but don't see why the interim release packages should [03:59] The stack is reasonably self-contained; I don't think we'll need to touch *everything*. [03:59] do you really think meta packages will handle this correctly? [04:01] like right now, try to upgrade to a new X video abi version with a nvidia driver or virtualbox installed providing the old one, the dist-upgrade will remove a metapackage and a bunch of other crap, then things are screwed [04:01] Those drivers shouldn't provide the old one. [04:02] But, indeed, that's the sort of thing to check :) [04:02] updated the wiki [04:03] wonder if stuff depending on xvfb will get confused [04:04] yeah this is my xorg-edgers experience since xserver 1.5 bugging me, i've never seen an abi update handled correctly [04:04] always some weird quirk in apt ordering screwing things up [04:05] probably my own fault with crappy packaging or not packaging absolutely everything though [04:07] by probably I mean guaranteed I guess, still worried about it :) [04:07] the 1.8 transition in ubuntu didnt go well [04:07] but was ok by 1.9 [04:08] 1.9 was in 10.10 and 1.8 was only stopgap during the dev cycle though [04:12] tjaalton, versions-current now updated [04:13] http://www.bryceharrington.org/X/Reports/ubuntu-x-swat/versions-current.html [04:14] sucks about 6.14.3 ati :( [04:15] new git snapshot? [04:15] bryceh: thanks! [04:15] Sarvatt, yeah [04:16] oh, heh. as if they'll ever bump the minor release :) [04:17] they will, 6.14.4, bet ya a beer at budapest :) [04:17] we could ask them to do 6.15 for the lts [04:17] oh thats micro [04:17] Sarvatt: that's patch release, or micro yeah [04:17] :) [04:19] any point to going through that list yet? [04:19] i was waiting for xserver 1.11 [04:19] only did libdrm [04:20] nah, stuff will get synced from testing, and the rest can wait for the server push [04:20] though I did upload wacom already [04:20] merges can be staged in git though [04:20] rebuilds are a lot faster than syncs though thinking about it [04:23] why is it necessary to have nvidia provide xorg video abi? [04:23] bjsnider: i wish i knew, i debated that one with tseliot since the start [04:24] nvidia doesn't even care what else is installed [04:24] it works on multiple abi's, it's only useful during the development release when it might be updated to a new abi it doesnt support [04:24] artifical limitations dont make sense to me [04:25] well, if we are always going to update the stack via a ppa in the future, maybe it is unnecessary to have the blobs provide it [04:25] nvidia especially, they have new abi support publically at most a week or so after its out, the real lag is the distro upgrading the driver [04:26] though fglrx is always behind [04:26] fglrx on the other hand is guaranteed 3 months [04:26] even if nvidia doesn't officially support an abi, it almost always works fine with ignoreabi [04:26] bjsnider: yea [04:26] 3 months behind? [04:26] does it work anyway? [04:58] re: apt mangling upgrades, that's probably because the thing that's not getting upgraded is marked as manually installed [04:58] while everything else is auto installed [04:58] so apt (unfortunately) concludes that you *clearly* meant to keep the manually installed thing, even if it means you lose the auto installed thing [17:06] Is there a bug in ppa-purge ? [17:07] it says that there are a lot of packages with unmet dependencies... [17:07] and it wants to delete a lot of packages not from xorg-edgers ppa [17:13] Milos_SD, i think ppa-purge has some problems handling multiarch and get confused by it [17:14] tseliot, please upload fglrx :) [17:14] ricotz: I'm working right now on some changes for the upstream fglrx but I'm afraid I won't make it today for the one in Ubuntu [17:15] I'm sorry [17:16] tseliot, hmm, if there are only problem regarding jockey could upload it somewhere where I can grab it? [17:17] ricotz: I've fixed jockey but something else's come up today and I had to delay my work on fglrx [17:18] tseliot, i see, so you dont have a "kind of working" source package at all? [17:19] ricotz: no, sorry [17:19] I only have an untested local branch [17:19] tseliot, i cant test it myself, but i really like to push something to the edgers ppa [17:20] ricotz: I know but can't we wait until Monday? [17:21] tseliot, sure, if you think it will break the world it might better [17:21] ricotz: it should. Also, I have yet to fix an issue with hybrid graphics... [17:21] tseliot, alright, it is ok then [17:22] ;) [17:23] it would fit the cracky edgers ppa though :P [17:23] heh