[00:25] hey [00:26] anyone at uds/ [01:13] mlankhorst: sure thing, just got back from a shopping spree [01:29] im below, still need to go to a computer store this week [01:31] buy from newegg or amazon, get it shipped to the hotel :) [01:38] hmm something weird happened, compiz is dog slow [01:40] ha, fast again [01:49] I spent a couple of hours yesterday walking around trying to find a grocery store and a decent pub. today I noticed they were both on the block next door.. [02:29] no us credit card sadly [02:43] aren't they all :) [17:13] ha, got the xserver crash with sigabrt in raise [17:14] ...isn't that where sigabrts always come from? [17:14] might well be :) [17:14] getting a signal is kind of the point of raise() [17:14] sounds like an assertion failure [17:15] resume from suspend, mouse works but nothing else doesn't, then when I kill the session I got this crash popup [17:17] nice thing is that this should be backtraceable from a virtual console.. === yofel_ is now known as yofel [20:05] tjaalton, i'll be attending the hybrid graphics meeting remotely. i don't intend to say much aside from the current state of 3rd party solutions. just want to take notes and report them to the hybrid graphics mailing list. [20:06] is that cool? [20:09] LLStarks: sure [20:50] tjaalton, thanks. will there be a transcript and recording of the audio available afterwards? [20:50] *chat transcript [20:53] there's a hybrid graphics mailing list? [21:11] LLStarks: not sure, could be [21:49] oh, nice https://blueprints.launchpad.net/ubuntu/+spec/foundations-q-session-management [21:52] jcristau, yeah: https://launchpad.net/~hybrid-graphics-linux [21:53] page needs to be cleaned up, but the list is at the bottom: https://lists.launchpad.net/hybrid-graphics-linux/ [21:57] LLStarks: the session concentrates on the work airlied has been leading upstream, and for 12.10 the blocker seems to be whether xserver 1.13 will get drvmodelv3 or not, since kernel 3.5 appears to be ready on the drm front [21:58] seems rather unlikely [21:58] and wth, I thought the session was on friday, but now it's scheduled for wednesday morning.. [21:59] jcristau: yeah, WIP all over the place on the branch [22:34] tjaalton, drvmodelv3 is the first branch to build successfully in months. i think it has a chance. [22:34] but it's worthless without the i915 uxa and nouveau ddx patches [22:35] the i915 ddx is problematic since airlied is only interested in uxa. ickle is committed to sna. [22:37] also, there are no proposals for acpi management of prime. right now, the card is always on unless explicitly disabled through pci and acpi calls [22:37] LLStarks: thanks for the info [22:55] somehow had overlooked the ddx support, duh [23:00] the ddx won't build against the drvmodelv3 xserver. months ago, i got nouveau to build, but never i915. iirc, there was a special abi needed. other than airlied, i'm probably the person who has tried to compile everything. it was really mindblowing to see the xorg.0.log that was produced. [23:00] LLStarks: any other stuff missing from the blueprint blurp? [23:01] i'd recommend talking to lekensteyn about adopting or upstreaming bbswitch for power management [23:01] bumblebee is not worth the time or effort, but bbswitch is essential for power saving [23:01] right [23:02] it would work with hw that don't have the bios options? [23:02] yes [23:02] great [23:02] the card will run with full fans unless the nvidia/nouveau module is active [23:03] heh, ok [23:03] also, canonical should harass nvidia to resubmit their proposal for blob drivers to use dma-buf prime [23:03] it was narrowly shut down for the 3.3 window and nvidia never bothered with it again [23:03] I think there was some concensus in allowing that [23:04] there was, but nvidia has proactive about it. [23:04] *has to be [23:04] right [23:06] do you know what's going on with the EA and valve stuff for UDS? i would hope that the x team wouldn't be blindsided by anything crazy [23:07] dunno about any valve stuff, and wouldn't worry too much about EA either :) [23:07] wednesday will tell [23:07] if its origin, i'll be shocked [23:08] i could live with a random game or two, but not the flood gates of a desura competitor on linux [23:16] wouldn't steam be a competitor as well?-) [23:18] well yeah, but gabe plays around with linux for servers and left 4 dead 2. steam is inevitable. [23:18] ea is just the evil empire [23:20] heh, ok [23:20] I've not used any of these :) [23:23] a good graphics stack and gaming options are the key to the linux desktop [23:23] i just hope the linux ecosystem is ready [23:23] well, if nothing else, having more of the prime/blah stuff for 12.10 will make it somewhat easier to test and work on the hybrid stuff [23:24] it'll make the buffer sharing possible, but there'd still need to be a way to display everything [23:24] I will be testing it for sure, since I have the hw and should have the time as well this cycle [23:25] 3.5 is a definite for 12.10, right? [23:25] even 3.6 is on the table, right now on the session :) [23:26] kernel session right now?!?! [23:26] yes [23:27] grand ballroom a [23:31] is the woman ogasawara? [23:31] yep [23:31] 3.6 would be cutting things really close. is there anything in particular that 3.6 would bring ? [23:31] i don't think it's even been defined [23:34] hw enablement, and yes 3.6 would be released roughly by late october/early november [23:34] there were other features as well, didn't catch what it was [23:35] kernel freeze is oct 4th, so they'll re-evaluate once 3.5 is released late august [23:36] There'll be improvements in 3.6, most interestingly in haswell support [23:36] kernel releases usually don't line favorably for ubuntu+1 to be kernel+3 wrt current ubuntu release. [23:36] *kernel+4 [23:37] they'd probably be able to pull the final release as an sru [23:38] but that assumes release is an rc5 or something [23:38] something stable enough [23:38] and it wouldn't be zero-day [23:39] don't know really [23:39] Why couldn't it be a zero-day? [23:39] but yes it would be rcsomething [23:40] let's say 12.10 ships with rc5. linus says we need another week or two due to an unforeseen circumstance but everything is stable. [23:40] *week or two for the next rc [23:41] 3.3 went all the way to rc7 [23:41] Oh, it wouldn't *necessarily* be a zero-day. [23:41] so zero-day would be another rc and then another one few weeks later the final release [23:42] But the obvious precondition in going 3.6 is that it be (a) sufficiently stable, and (b) sufficiently early. [23:43] that makes sense, but i don't remember any precedent for deciding a kernel so late and shipping with an rc. [23:44] haswell would be nice, but aren't most of the bits landing in 3.4 and 3.5? [23:44] true, never done that before [23:45] releases after lts tend to be wacky [23:45] so i dunno [23:47] heh [23:47] "rolling release" :) [23:48] always usable [23:48] or u(n)s(t)able, depending who you ask [23:52] if my hearing was correct, i heard the term come up during the chatter [23:53] i've come to dislike uds though. they never revisit stuff that people keep asking for. where are delta debs, systemd, btrfs, and a sid-synced repo? === LLStarks is now known as erappleman [23:59] systemd won't happen (got blogged about), btrfs is unstable, dunno about the rest