=== IdleOne is now known as Narwhal === emma is now known as em === Quintasan_ is now known as Quintasan === ogasawara_ is now known as ogasawara === humphreybc-ipad_ is now known as humphreybc-ipad === cking_ is now known as cking === chrisccoulson_ is now known as chrisccoulson === mdz` is now known as mdz === yofel_ is now known as yofel === MichealH is now known as MichealH[Release === MichealH[Release is now known as Micheal[Release] === Micheal[Release] is now known as MichealH === Daviey_ is now known as Daviey === nhandler_ is now known as nhandler === jcastro_ is now known as jcastro === Riddelll is now known as Riddell === jjohansen is now known as jj-afk === kirkland` is now known as kirkland [15:59] fnop [15:59] #startmeeting [15:59] Meeting started at 09:59. The chair is NCommander. [15:59] Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE] [16:00] do we have persia ? [16:00] hello [16:00] hmm, doesnt seem like [16:00] [topic] Specs [16:00] did anyone get a linklist with blueprint links ? [16:00] New Topic: Specs [16:00] * davidm says hello, 11.04 is out [16:00] yay 11.04 [16:00] davidm, oh, thats good news *g* [16:01] Indeed, too bad I was sleeping when it happened [16:01] * ogra doesnt have any idea what blueprints persia set up [16:01] which makes it a bit hard to discuss specs :) [16:02] there are a lot of blueprints and many may have to be renamed, currently there is no ARM track!! [16:02] and we wont get one ? [16:02] I'm asking for one, otherwise everything arm will shift to "other" [16:03] * ogra makes a note to tickle jcastro until he cries once a day at UDS if we dont get one [16:03] goto: https://blueprints.edge.launchpad.net/~davidm [16:03] all the current specs are at the bottom labeled -o- something [16:03] davidm: I've been in calls all morning, jono recommends putting them in existing tracks [16:03] so like, foundations, etc. [16:03] I checked the list of BP yesterday, none of arm only linaro specs [16:03] fun [16:04] (sorry to barge in, we can sort it post meeting if you'd like) [16:04] i didnt get email notification for any of the ones i'm assigned to [16:04] I did not get any email either, I think it's borked [16:04] jcastro, thats sad [16:04] jcastro, yes lets sort it post meeting, 5thanks [16:05] well, that was our meeting ... we wanted to discuss specs, if we need to rename them its possibly best to discuss it now [16:05] Also we need to review any open natty spec's and either close them or move them forward [16:05] so maybe we should go around one by one and discuss the specs we have or? [16:06] NCommander, well, we did that already in the call, i think the renaming would be important now [16:06] we know who does what (well, persia knows since he assembled a list) [16:06] ogra, the name is not important really it's either arm or other that is it [16:07] davidm, well ... i guess we have to live with that [16:07] if other it will be hard for folks to find but that is the only real impact [16:07] ogra, yea, I'm less then happy but I don't make the rules [16:08] * ogra would love to rant now about being ignored as arm dev by other devs all the time and now not even having our own track ... but i'm used to be treated as second class citizen nowadays [16:08] but i wont do that rant now :P [16:08] if we have to rename my plan is arm-o-boot-speed -> other-o-arm-boot-speed [16:08] that sounds good [16:08] so people can at least match arm [16:08] guess that works [16:09] Yep it's the only way that folks that care about arm will find them if we don't have a track [16:09] Lack of ARM track though is extremely annoying [16:09] NCommander, ++ [16:09] OK moving on [16:09] Sorry I'm late. https://wiki.ubuntu.com/OmapNetbook is up. Please check it for correctness, it was mostly a cut/paste from Maverick. [16:09] nothing to move to :) [16:09] specs was our topic today [16:09] well lest talk about specs that we have [16:10] GrueMaster, download link is wrong [16:10] first is arm-o-preinstall-pool ogra that is yours [16:10] ogra: That's where I found the images. [16:11] yeah, its about putting the contents of the ship seed and langpacks into /var/cache/apt/archives om the images [16:11] GrueMaster, http://cdimage.ubuntu.com/releases/11.04/ [16:11] davidm, i think that one is pretty straightforward [16:12] GrueMaster, dont say natty there [16:12] likely, do you have any guestimate on how much time it will take [16:12] a few days including testing [16:12] not more than one to techincally implement it [16:13] testing the langpacks actually get installed etc will be time consuming here [16:13] OK what I'd like is if all specs have a guestimate of time in them, could be wrong but it would be very helpful to know some rough idea [16:13] changing livecd-rootfs to fill the pool should be quick and easy [16:14] davidm, i know though (since we discussed it before) that NCommander had concerns doing it that way [16:14] so we should definitely discuss other ways to implement it at UDS [16:15] which might take more time [16:15] putting a time guestimate up for the specs before we discussed the final implementation feels a bit premature [16:15] that is OK lets try to get some very rough guestimates this week and next and the refine at UDS [16:16] hmm, k [16:16] oops [16:19] well, lets get this meeting moving then [16:19] NCommander, arm-o-seed-spices is yours [16:19] yep thanks [16:19] any time guesstimate ? [16:19] Would estimate about a month if we stil want to implement it [16:19] Needs some minor tweaks to LP and cdimage [16:20] that is: generic rootfs and overlay for /lib/modules and boot stuff, right ? [16:20] NCommander, if we still? [16:20] what are your thoughts? [16:21] davidcalle, it was discussed for natty already but never implemented [16:22] ogra: It was more finer grain control for seeding things onto a per image basis [16:22] err, sorry DavidLevin [16:22] GAH !!!! [16:23] NCommander, well, i think the initial idea GrueMaster had was what he called hwpacks (not to mix up with linaro hwpacks though) to have overlays for images on the download server [16:23] which wuld put all hw specific bits in as an overlay over a generic rootfs [16:23] i think thats different from seeds [16:23] Yes, the idea was to have some kind of compressed file with modified MLO & u-boot + any other bits specific to a platform. [16:24] anyway, 1 month is your ETA for it ? [16:24] And a tool that would extract and update an image with these bits. [16:24] Yes, sounds good. [16:24] k, lets move on [16:24] arm-o-faster-preinstall-filesystem [16:24] janimo, ! [16:25] you will speed up our lives [16:25] how long will that take you ? [16:25] the SD alighn should be a week with testing [16:25] yeah, agreed, but thats not all of it, is it ? [16:25] the generic boot speed improvements depending on what the scope is [16:26] I did not research ionto the jasper part of the spec [16:26] oh, right, the arm-o-faster-preinstall-filesystem one was GrueMaster [16:26] i didnt match the line [16:27] I plan on taking a natty image and converting the rootfs to various different filesystems to run benchmarks. [16:27] janimo, i think we can leave out jasper here ... and focus on bootspeed of the installed system [16:27] GrueMaster, just take into account resizeability [16:27] with your selection of filesystems [16:27] GrueMaster, any rough ETA ? [16:28] Of course. My idea is to continue to use the existing format and have jasper convert after resize. [16:28] so there is one spec for SD speed and the generic iterative boot improvements (which are general ubunut-foundation work) [16:28] 1 month? [16:28] GrueMaster, k [16:29] GrueMaster, dont forget that convertion has its own danger ... it would be intresting to have user interaction though if you add a conversion anyway, so he can select what to conver to ;) [16:29] janimo, yeah, any time estimate for them ? [16:29] My thought was to convert to an optimal fs. [16:30] If you want user interaction, that is a whole nother ball game. [16:30] GrueMaster, and why not use that optimal one from the beginning ? [16:30] so SD align 1 week, ongoing boot speed at most a month - depending on how much we want to squeeze [16:30] Because resize tools may not exist. [16:30] user interaction is trivial now that we have plymouth support in jasper [16:30] do a general bootchart/perf evaluation and pick low hanging fruits if there are [16:30] For example, btrfs doesn't support resize, but there is a conversion tool. [16:30] we can even squeeze the boot time from u-boot if needed [16:30] can we all simply add the time estimates to the whiteboards for now ? [16:31] rsalveti, we should definitely cut down the autoboot timeout :) [16:31] yeah, 10 sec for omap 3 [16:31] I'd think we should wait with all timing optimizations after the SD changes land, so we make sure we don't optimize something that won't be there with a better layout [16:31] this includes boot speed and firrefox startup speed [16:32] so add the right dependencies to your specs then [16:32] you can make one spec depend on another [16:32] right [16:32] ogra, good point [16:33] hmpf, i got arm-o-firefox-startup ? [16:33] * ogra cries [16:33] ogra, I am happy to take that [16:33] I did not know it exists [16:33] * NCommander raises hands [16:33] I'M FREE, FREE FROM EVIL! [16:33] well my ETA would have been 6 years and counting for that one :) [16:33] heh [16:33] janimo: I warn, the FF codebase will twist and distort your mind [16:34] I know upstream ff does a lot of work on startupperformance [16:34] anyone who would like it, i'm happy to give it away [16:34] but it is still slow (2-3 seconds on beefey desktop computers) [16:34] much of it due to sick extensions [16:34] NCommander, I am pretty sure when I get to the bottom of the uglies C++ file in FF I'll see a "NCommander wuz here" comment [16:34] on my ac100 its currently at 30sec or so [16:35] vs 10 in maverick [16:35] it got a lot worse in natty for me [16:35] so who takes it? janimo or NCommander [16:35] * NCommander still thinks we should ship Chromium but YMMV [16:35] me [16:35] * NCommander runs away [16:35] k [16:36] * ogra moves it over [16:36] I also think we should ship chromium FWIW [16:36] not sure about the security team implications here :) [16:36] we had that discussion at several UDSes already [16:36] whoa! [16:37] no way, not without a lot of discussion [16:37] upstream introduces regressions and then doesn't fix them for weeks [16:37] plus there is no advance warning on security issues, even though we have tried [16:37] heh [16:37] we are completely at the mercy of upstream [16:37] everyone is lurking today :) [16:38] * NCommander wonders if jdstrand has 'security' has a highlight [16:38] I think there may be a bp on it. micahg is the one to talk to if interested [16:38] * jdstrand will never tell [16:38] :) [16:38] TBH on natty chromium on the ac100 isnt much faster than FF [16:38] they are both slow like wet sponges [16:39] ogra: We really can't use an unsupported system as a benchmark. [16:39] i didnt test on panda but wouldnt expect much difference [16:39] GrueMaster, come on [16:39] We should ship lynx [16:39] +1 [16:39] we used to ship w3m [16:39] -1 [16:39] why would lynx be better ? :P [16:39] Its not emacs === doko_ is now known as doko [16:39] * janimo writes up Add GLES2 backend to Lynx BP and assigns to NCommander [16:40] ++ [16:40] i also want GLES animations in u-boot ! [16:40] * jdstrand hands w3m-el to NCommander [16:40] add that to him too ... he said he has no evil yet :) [16:40] how the heck are you supposed to accelerate a CLI application?! [16:40] plus, w3m-img is funky [16:40] * jdstrand stops being distracting now [16:41] On my panda, ff comes up in 15 seconds first time, 5 seconds second time. [16:41] Base image. [16:41] k, my MMC in the ac100 is probably a tad slower [16:41] GrueMaster,the logical solution is to skip the first start everytime [16:41] haha [16:41] well, lets move on ... there are more specs [16:41] arm-o-netinstall [16:42] thats mine [16:42] a week or less [16:42] arm-o-partman-chs ... NCommander [16:42] (that should really not have partman in the name) [16:42] Need to research what it will take to get partman to write a partition table OMAP will like [16:42] er, parted [16:42] well, the question is, do we want to use it at all [16:42] (parted appartantly grew CHS paritioning when no one was looking, but damned if I can find documentation) [16:43] ogra: to be discussed? [16:43] you could add such stuff to the new flash-kernel-installer architecture as a hook [16:43] or some such [16:43] yeah, that will need a lot of discussion [16:43] ogra: uh, that won't work. [16:43] and wont be easy to implement [16:43] Lets discuss at UDS [16:44] a typical NCommander spec then ;) [16:44] * NCommander shouts family-unfriendly words [16:44] mind wrapping and hard to do ... :) [16:44] NCommander, make sure to invite cjwatson or ev to the session [16:45] ogra: is TI going to be at UDS? [16:45] NCommander, any eta ? 4 months ? [16:45] not clear yet [16:45] probably not [16:45] I'd like to know specifically what there bootrom needs [16:45] ugh [16:45] the TI guys from the linaro landing team will probably be around though [16:45] they should be able to help [16:45] NCommander: partman/alignment=cylinder, probably [16:46] and i thinnk rsalveti knows some internals about the rom code too [16:46] (and if you aren't using partman, why not ...) [16:46] cjwatson: you've been implementing useful features again :-P [16:46] cjwatson, rom code fo omap expects a certain CHS setup [16:46] yup, and the trm can help here, usually they put the restrictions there [16:47] right, but lets discuss that at UDS :) [16:47] sure :-) [16:47] arm-o-port-lightspark [16:47] janimo, !!! [16:47] flash for everyone !! [16:48] * NCommander suspects ligthspark has some sorta JIT in it [16:48] well, would be nice to have a foss flahs p[layer on arm [16:48] does gnash not work on arm? [16:48] lightspark is promising if still not too mature(ponly suport youtube) [16:48] * micahg knows it FTBFS last time around [16:48] micahg, works fine ... exactly one time for me [16:48] micah yes it fdoes [16:48] and then never again until next boot [16:48] but most flash content is moving AIUI towards actionscript3 /Adobe Virtual MAchine 2 which gnash does not implement [16:49] we could indeed just add an auto-reboot to its code ;) [16:49] and just release note it *g* [16:49] ogra: so we'd turn it into Microsoft WIndows? [16:49] lightspark needs some more testing and optimizations for arm [16:49] maybe 3 weeks [16:49] NCommander, win has such a cool feature ? [16:50] * NCommander thinks emulating Microsoft's design decisions w.r.t. to error handling (auto-rebooting) is a bad idea [16:50] * ogra was surely not serious :) [16:50] anyway, moving [16:50] crash-only software is not that bad if it boots up fast enough :) [16:50] which gets us to the next spec :) [16:50] arm-o-preinstall-sd-optimisation [16:50] i think we covered that above [16:51] arm-o-uboot-tftp [16:51] another famous janimo spec [16:51] NCommander, the JIT is LLVM based (and quite experimental) so should not be an issue, as that is the point of using LLVM [16:51] only the SD one, uboot is not my cup of coffee :) [16:51] that's rsalveti right? [16:51] well, its assigned to you atm [16:51] hmm, weird [16:51] tsk [16:51] no, you are right [16:52] i cant read today or the CSS on LP tricks me [16:52] i always catch the wrong line [16:52] Maybe your laptop. [16:52] rsalveti, so tftp in uboot :) [16:52] uboot == testing booting == testing hw == cables all over the desk == chaos [16:52] GrueMaster, i'm working on a 24" screen atm [16:52] lol [16:53] should be easy to work with, if the patches are really in a good shape and working :-) [16:53] and if not ? [16:53] take the worst case for the time estimate [16:54] then we clearly need some work, probably around 1 month [16:54] k, put that in the whiteboard [16:54] NCommander, ! [16:54] sure [16:54] foundations-o-image-build-docume... [16:54] long title :) [16:54] LP chopped it [16:54] NCommander, i think thats just "take notes while helping ScottK to implement mx5 images" [16:55] so we have a wikipage for later community images [16:55] I thought it was me watching. [16:55] I was actually looking at documenting the stack [16:55] ScottK, well, you being the customer, we help and document :) [16:55] A large problem of ANYONE trying to do anything with our infrastructure is there is a big "?" when you go look for documentation [16:56] NCommander, seems a bit over the top, but your spec, as you like :) [16:56] NCommander, ETA ? [16:56] Two weeks, plus however long it takes my test vict^W^W ScottK to run through it [16:56] hehe [16:57] make it 4 weeks then, to be on the safe side [16:57] k [16:57] we will have finalized times after UDS anyway [16:57] i doubt they will stick [16:57] kernel-o-flash-kernel-postinst ... thats me [16:58] and largely implementing the postinst logic of grub into our kernel images for flash-kernel ... i will pester cjwatson and apw for that at UDS [16:58] two weeks i'd say [16:58] i think thats all specs ... [16:58] NCommander, take over the meeting again (and sorry for hoggin it ) [16:59] (you have 1min for AOB and closing ;) ) [16:59] [topic] AOB [16:59] New Topic: AOB [17:00] nothing here [17:00] are these all the specs? [17:00] janimo: feel free to donate more [17:00] janimo, yeah [17:00] do we get more (possibly from others?) [17:00] we will [17:00] ANyway [17:00] ok [17:00] if nothing else [17:00] closing in 3 [17:00] 2 [17:00] 2 [17:00] 1 [17:00] \1.5 [17:00] 2 [17:00] 3 [17:00] 0.5 [17:00] #endmeeting [17:00] Meeting finished at 11:00. [17:00] 1.725 [17:01] bah [17:01] cheers [17:01] thanks all ... enjoy the post release relaxation ! [17:03] * ScottK wonders what that is (just did two toolchain uploads ...) === jj-afk is now known as jjohansen === noy_ is now known as noy === Narwhal is now known as IdleOne === noy_ is now known as noy [18:48] i cant blv i waited soo long for 11.04 and finally ended up with having a ridiculous unsolved backlight issue! cant see anything in it! safer being with 10.10 === Andre_Gondim is now known as Lucas_Laudeci === Lucas_Laudeci is now known as Andre_Gondim === cody-somerville_ is now known as cody-somerville === cody-somerville_ is now known as cody-somerville [20:47] sorry, wrong channel === UndiFineD is now known as hajour1 === hajour1 is now known as UndiFineD === SWAT_ is now known as SWAT