=== ChanServ changed the topic of #ubuntu-uds-core-1 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1311/core-1/ - http://irclogs.ubuntu.com/2013/11/21/%23ubuntu-uds-core-1.html [03:00] hola de que se trata esto? === tvoss_ is now known as tvoss|test === tvoss|test is now known as tvoss [07:13] Hello, does the meeting start at 18:05 UTC ? === lol is now known as Guest50536 [14:54] good [morning|afternoon|evening|night] [14:55] cjohnston, do you know who is running this track? [14:55] fginther: hi, https://plus.google.com/hangouts/_/76cpirvjdcq5ofrvqvkmkqslps?authuser=0&hl=en === ChanServ changed the topic of #ubuntu-uds-core-1 to: Track: Core | Upstream Merger 2.0 | Url: http://summit.ubuntu.com/uds-1311/meeting/21983/core-1311-upstream-merger-20/ [14:55] ogasawara, thanks! [14:58] cjohnston, Ursinha, want to join the hangout? [14:58] fginther, sure [14:58] not sure I need to [14:58] https://plus.google.com/hangouts/_/76cpirvjdcq5ofrvqvkmkqslps?authuser=0&hl=en [14:59] cjohnston, worst case you won't say a thing :) [15:00] Saviq, want to join the hangout? https://plus.google.com/hangouts/_/76cpirvjdcq5ofrvqvkmkqslps?authuser=0&hl=en [15:03] * tedg puts on his asbestos suit, let's rock! :-) [15:03] tedg: heh [15:03] tedg, dobey, either of you want to join the hangout? [15:03] https://plus.google.com/hangouts/_/76cpirvjdcq5ofrvqvkmkqslps?authuser=0&hl=en [15:03] anyone else ^ [15:03] i can i guess [15:03] fginther, Sure [15:04] * tedg does his hair [15:04] haha [15:04] https://plus.google.com/hangouts/_/76cpirvjdcq5ofrvqvkmkqslps?authuser=0&hl=en [15:07] lower thirds please :) [15:09] tedg, we did have ddebs for most of the manhattan stuff [15:09] sergiusens, Yeah, I know, I just want to ensure it's a feature into the future. [15:11] what is the timeframe for these changes? [15:14] awesome, thanks [15:16] no lower third with no camera :) [15:16] josepht: ^^ just saw your comment [15:18] I have lots of questions, but they don't seem in scope [15:18] how is the emulator going to solve the ci problems? [15:18] sergiusens, ask anyway :-) [15:18] is click testing tied to the emulator? [15:19] there is still going to be *some* actual hw testing right? [15:19] or will that just be sergiusens's papi testing? [15:20] Saviq, we actually don't do that for clicks either [15:20] Saviq, I've always viewed QML as a cry for help, nice to hear you say so too! ;-) [15:20] cwayne, lol, I cannot not laugh when reading papi testing [15:20] heh [15:20] sergiusens, lol i know, i love it [15:21] :D [15:21] but is that testing going to be the only hw testing now? [15:22] i mean, we do need some level of hw testing obviously, but i see where it would make more sense for apps to be tested on emulators [15:22] there really should be no reason to test on actual hardware, for things that don't talk directly to hardware; for mir it makes sense, for a unity scope, not so much [15:23] Saviq: well, you could probably hook into the JS engine and check read/write access of public properties [15:23] tedg: in qml _every_ line is "executed" at startup [15:23] mzanetti, Okay, I've got some recommendations for speeding up startup then :-) [15:24] tedg: you need to read it to know when to creat it [15:24] so you need to read the file at least once. and that's what happening - at the startup [15:24] dobey, right, i'm +1 ont hat, but just making sure we will still have some baremetal testing, even if it is just for mir for example [15:25] mzanetti, You need to read the file, but you don't need to create objects based on it or parse deeper than "oh, I'll need this someday" [15:25] Sarvatt: but that you can count [15:25] Saviq: ^ - but sure it's really not trivial [15:26] mzanetti, yeah, for bindings, we'd have to see whether all the branches were executed [15:26] tedg: well, depending on how you write it, they are not created. but still interpreted [15:26] real h/w testing will continue. we just want to find a better balance between that and virtualized to increase scalability and reliability. [15:26] +1 on that [15:26] i'd +more-than-1 if i could :) [15:26] mzanetti, Makes sense, so code coverage would be when objects are created, not interpreted. [15:27] fginther, click packages require ubuntu touch [15:27] tedg: I'm thinking more about "used" instead of "created". but yeah, something along those lines [15:27] sergiusens, wouldn't that work on a touch emulator? [15:27] sergiusens, join the hangout? :) [15:27] fginther, I suggest you try to run the emulator on the cloud now that it's there ;-) [15:33] sergiusens: can you add the lower third? [15:34] dobey, sure, should be automatic :-) [15:34] ogasawara, you can close the hangout now [15:35] thanks, will do [15:36] thanks fginther [15:36] thanks all! [15:37] thanks guys! === ChanServ changed the topic of #ubuntu-uds-core-1 to: Track: Core | Low battery handling during boot | Url: http://summit.ubuntu.com/uds-1311/meeting/22108/core-1311-low-battery-boot-mode/ [16:00] https://plus.google.com/hangouts/_/7ecpi1hhd2cgo0j1h4c242ik7k?authuser=0&hl=en [16:02] lool, want to be in the fishbowl ? [16:02] apw, lool, rsalveti: ^^ fyi [16:05] joining in [16:06] https://plus.google.com/hangouts/_/7ecpi1hhd2cgo0j1h4c242ik7k?authuser=0&hl=en [16:06] sforshee, intrested in being in the fishbowl ? [16:07] im interested [16:07] to not talk [16:07] heh [16:07] ogra_: I'll jump in if needed [16:08] ogra_: I'm a bit under the weather, so if I don't have anything to contribute then there's no reason to make everyone watch me blowing my nose ;-) [16:09] heya [16:09] which is that [16:09] https://plus.google.com/hangouts/_/7ecpi1hhd2cgo0j1h4c242ik7k?authuser=0&hl=en [16:12] rsalveti: you don't want powerd, if anything upower [16:12] rsalveti: but the battery information is in /sys/class/power_supply [16:13] ogra_: the kernel provides the abstraction: /sys/class/power_supply [16:13] though some logic is needed to figure out the right place to get the battery status [16:14] let me join [16:14] the reliability of the data can be platform specific though [16:16] ChickenCutlass, https://plus.google.com/hangouts/_/7ecpi1hhd2cgo0j1h4c242ik7k?authuser=0&hl=en [16:16] * cking notes that a 1% charge could be a long time on some large batteries [16:18] some devices, 1% may be a very short time, so it may be device specific, e.g. 5 mins? 10 mins? 20 seconds? ;-) [16:18] http://pad.ubuntu.com/uds-1311-core-1311-low-battery-boot-mode [16:25] ogra_: /etc/init/friendly-recovery.conf [16:26] ogra_: we just need to change /usr/share/initramfs-tools/init to pass the custom startup event see last line of that file. [16:26] ogra_: last 2 lines :) [16:31] ogra_: subvert the boot if battery is low (as above). When battery level > some threshold, "initctl emit startup" to resume normal boot. [16:32] yeah [16:36] np! :) [16:37] ogasawara, you can stop the recording [16:37] and thanks all, for coming !" [16:37] ogra_: thanks, done. === ChanServ changed the topic of #ubuntu-uds-core-1 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1311/core-1/ - http://irclogs.ubuntu.com/2013/11/21/%23ubuntu-uds-core-1.html === ChanServ changed the topic of #ubuntu-uds-core-1 to: Track: Core | Ubuntu Touch Emulator | Url: http://summit.ubuntu.com/uds-1311/meeting/22113/core-1311-touch-emulator/ [17:59] * rsalveti waves [18:00] https://plus.google.com/hangouts/_/76cpiorbncb4s4j02ib1noi6is?authuser=0&hl=en it seems [18:01] http://pad.ubuntu.com/uds-1311-core-1311-touch-emulator [18:01] pmcgowan: ^ [18:02] is zoltan around? [18:02] rsalveti, he is but cannot joing the HO [18:02] hola! [18:02] we need someone from CI [18:03] plars: ^? [18:03] is there already some preview version of the emulator? [18:03] dkessel: yup [18:03] dkessel: https://lists.launchpad.net/ubuntu-phone/msg05195.html [18:05] rsalveti, ah, a normal package... nice :) [18:05] rsalveti: I do not join to the hangout because I am still mute [18:08] bzoltan, ping me if I need to hop on the session [18:08] pmcgowan: OK [18:12] Yes [18:12] rsalveti: Yes, [18:13] rsalveti: Yes, you start the emlator and it works with the QtC out of box [18:13] bzoltan, oh, it does already ? [18:13] thats cool [18:13] rsalveti: The emulator is just an adb device [18:13] (and fast ... for two days) [18:14] ALL [18:14] :) [18:14] rsalveti: we backport to PQRS [18:14] packaging is cool [18:14] ++ [18:15] packaging yes [18:15] it's cool unless the target is fast moving :) [18:15] at the SDK team we move fast ... [18:16] rsalveti: we need to fix the adb port forwarding and small things [18:17] +1 [18:18] yes [18:18] it is a single line [18:18] whats wrong with ssh ? [18:18] ogra_: QtC uses adb [18:19] No [18:19] I need to fix the QtC scripts [18:19] Yes [18:19] It is QtC side [18:19] :D [18:19] there is a delay [18:19] sorry [18:20] are we ? i thought we grab the tarball from scimge [18:20] *cdimage [18:21] rsalveti: may I have a minimal rootsrap to use as a builder? We have click chroot as main target... but a fully emulated native builder would be cool as fallback [18:21] bzoltan, that will be slow as hell [18:21] rsalveti: or an R&D enabled image [18:21] ogra_: it is not that slow ... [18:22] bzoltan, qemu only emulates a 400MHz single core [18:22] ogra_: I just have built the UITK in the emulator [18:22] bzoltan, did you try building on the emulator? [18:22] and thats faster that a qemu chroot? [18:22] sergiusens: I did not _TRY_ [18:22] sergiusens: I did :) [18:22] bzoltan, if you want a qemu-system based emulator to build in there are way better options [18:23] ogra_: I know ... [18:23] (we have a qemu highbank option afaik ... ) [18:23] ogra_: I need to talk to you about it [18:23] sure [18:23] :) [18:23] plars, has emulator on a VM (cloud) been discussed? [18:23] fginther: not yet [18:24] fginther: should be possible I would think, but it may be awfully slow [18:24] plars, sergiusens expressed some concerns other then slowness in a prior session [18:24] fginther: I was going to ask about it, since I know it's something we want to do :) [18:24] thx [18:29] Ctrl+F12 to rotate [18:29] ChickenCutlass, rsalveti, sergiusens ↑ [18:29] Saviq, nice [18:29] ok [18:29] wow [18:30] that works really nicely [18:30] ogra_, didn't rotate the browser for me, though [18:31] well, still booting here [18:31] but i can rotate while it boots :) [18:31] lol [18:31] and comes up sideways [18:31] ogra_, see! it emulates real life ;) [18:31] ogra_, Saviq: true real-world case :) [18:31] heh [18:32] Saviq: awesome, need to give it a try [18:32] rsalveti, so it just looks like the sensor does not give up the rotation change [18:32] rsalveti, so just more hw enablement [18:32] Saviq: right [18:32] yeah, browser is sideways here [18:32] should be easy to fix :-) [18:32] i guess we need some sensor hook-up [18:33] QUESTION: can we look into a snapshot option? it doesn't have to be complicated-- use qcow2 disk and add --snapshot and --revert options to run-emulator. this will allow developers to easily revert to a pristine snapshot, which is highly useful for testing [18:33] poer button doesnt work either etc ... i guess thats similar [18:33] jdstrand, only if that ancient qemu supports that ... and we dont have qcow images atm [18:34] ancient qemu should support snaphost, but if not you could also use backing store [18:34] s/snaphost/snapshot/ [18:34] so feels if we do real driver testing we need a maas cloud [18:34] FYI: more keyboard shortcuts http://developer.android.com/tools/help/emulator.html#KeyMapping [18:34] so we can provision on bare metal etc. [18:34] or can we share drivers across multiple VMs? [18:35] what do you mean by drivers exactly ? [18:35] technically, it could be opt in-- you could use qemu-img to convert to qcow2 after downloading the image [18:35] if we have servers with GPUs :) [18:35] oh, yeah, the server definitely needs the GPU [18:35] we could run them on bare metal and use GL properly... otherwise i assumed we would just run against software mesa [18:35] most likely even running X [18:35] etc. [18:35] in virtual... but might be too slow [18:36] or xvfb ... not sure that provides GL [18:36] right. but our current machines ... namely otto etc. are always busted [18:36] because we do raw GPU testing [18:36] then we need a dedicated machine [18:36] so we need maas to automatically recover rather then alwys do manual maintenance [18:36] image snapshotting doesn't have to be complicated btw. people can get super clever-- only really need snapshot and revert to snapshot [18:36] desktop class machine ... [18:37] desktop is where the test machines always hang [18:37] because of GPU crashes [18:37] -> e.g. that should be maas with automatically reflashing machines with a safe iamge [18:37] asac, right, it wouldnt put any load on the GPU [18:37] but it needs something at the desktop end to hand through the GL [18:37] so you need some X server/Xvfb etc running [18:38] sure. X would run etc. [18:38] if we want to use pass through... still needs to be maased etc. [18:39] IT WILL MESS UP TIMEOUTS :) [18:39] ogra_: what version of qemu is it based on? [18:39] and we should fix those [18:39] hehe [18:39] jdstrand, dunno, but likely older than your grandma ... [18:39] rsalveti knwos i think [18:40] 0.9 or some such ? [18:40] I was doing backing stores with qemu a *long* time ago [18:40] i think the emulator is well usable ... even though pretty slow. the problem i see is just how i can do gestures. can we mayb hook the buttons up to various gestures? [18:40] like the home button gets the swipe to home? [18:40] asac, yes, that would be good. [18:41] rsalveti: ChickenCutlass: ^^ [18:41] do we have a ppa from the emulator or is it just packaged for trusty? [18:41] asac, the buttons should go away [18:41] oh already answered [18:41] I thought we'd remove all buttons :-P [18:41] ++ [18:41] ogra_: we can use different images... doing a swipe is not very easy with mouse [18:41] well, we should keep power and volume [18:41] and especially in slowness for now [18:41] asac, works for me ... you need to be patient :) [18:41] * Saviq just started a unity8 autopilot run [18:41] so having a button to do a full left/right/up/down swipe [18:41] would be good for productivity [18:42] ogra_, asac the performance is good for me. [18:42] yeah, same here [18:42] need to wait until the system settles [18:42] if it is triggered by emulator [18:42] please let's not hook into unity [18:42] it should come from hardware as a swipe at best :) [18:42] or well, at most through autopilot [18:43] UInputError: "/dev/uinput" does not exist or is not a character device file [18:43] plars, ↑ when trying to run autopilot tests [18:43] Saviq, autopilot wont be there for devs (for their brandnew app they test) [18:43] ogra_, ap is seeded [18:43] sure [18:43] but if you are $random_dev you want to get your app done [18:43] yeah maas is a CI engine topic to explore [18:44] e.g. how to provisionm good worker nodes for the emulator [18:44] I'm curious now [18:44] Question: Will there be (command-line) switches to select a specific device to emulate (N4, N10 etc)? [18:44] MacSlow, lol, no [18:44] I think I'll play with something (ridiculously rough) to see if we can do backing stores or snapshots [18:44] MacSlow, it is completely based on the goldfish qemu emulator from google [18:45] ogra_, ok [18:45] MacSlow, to get other device emu, you would have to write a complete qemu machine that emulates that device [18:45] MacSlow, currently we are limited to a single core 400MHz CPU with 512M ram [18:46] ogra_, probably depends on what MacSlow meant to select a device [18:46] ogra_, one of the more important things is the resolution, which obviously would be available [18:46] we will enable different screen resolutions [18:46] that was discussed [18:46] but we cant easily emulate another HW [18:46] rsalveti: ChickenCutlass: also we want landscape/portrait mode [18:46] yeah, not sure what else is different between devices [18:46] is that something the emulator need to support? [18:46] asac, yup, already covered that [18:46] ogra_, Saviq: well mostly default orientation and resolution... CPU/GPU not so much atm [18:46] ah ok [18:46] thanks [18:46] asac, Ctrl+F12 [18:46] asac, already does [18:47] MacSlow, that will be supported [18:47] ogra_, ok [18:47] ok. nice on accell and orientation [18:47] rsalveti, that was mostly about resolutions [18:48] is someone from sdk in the sessionm? [18:48] _YES_ [18:48] asac, bzoltan is here [18:48] (all the time) [18:48] :) [18:48] well. you cant see that in hyoutube [18:48] I am listening [18:48] if you dont have a pic :) [18:48] he isnt in the hangout [18:49] would be nice to hear what the sdk vision is [18:49] though [18:49] anyway :) [18:49] yeah ok [18:49] I am here! [18:49] was discussed above [18:49] :) [18:49] read the backlog [18:49] asac: I am temporarily mute [18:49] ok sorry [18:49] there's prior art on QtCreator + emulator [18:49] hard to catch up on a video stream :) [18:49] for MeeGo at least [18:49] and then match the irc backlog [18:49] hehe [18:50] lol [18:50] asac, yeah, the important stuff happened in the IRC channel anyway [18:50] Saviq: I know... I was working on it :) [18:50] ignore the guys on youtube :P [18:50] bzoltan, good! [18:50] they just fill the pad over there [18:50] ogra_: they should join here and stop the crap there :) [18:50] yeah [18:51] good stuff [18:51] o/ [18:51] ++ [18:51] Question: Regarding autopilot... ap-test passing on real hardware, but failing on the emulator will be considered an autopilot-binding bug on the emulator, right?! [18:51] MacSlow: nomen est omen :) [18:51] MacSlow, they will have to be researched [18:52] MacSlow, rather a bug in the test [18:52] emulator will be the first target [18:52] we cant just blame the emulator, we have to look at the reasoning etc [18:52] can't fail in the emulator [18:52] Saviq, really? [18:52] can fail in devices, but still be critical [18:52] MacSlow, if something would fail on the emulator, but pass on devices [18:52] MacSlow, means it wouldn't pass on a high-load device either [18:53] ah [18:53] MacSlow, 'cause that would just be timing-related [18:53] version 0.12.0 [18:53] jdstrand, oh, that new ! [18:53] MacSlow, so you'll need to make the test more robust [18:53] external/qemu/Changelog [18:53] * ogra_ expected older actually [18:53] Saviq, ok... in that particular case (timing) I'd agree... [18:53] o would have put my bets on 0.8 or 0.9 [18:53] so, qemu-img still defaults to compat=0.10 [18:53] MacSlow, any other reason - as ogra_ said, would need to be investigatesd [18:53] -s [18:53] so I think we should be ok with at least backing store (which is usable) [18:54] MacSlow, but that should be really rare [18:54] Saviq, I hope so [18:54] jdstrand: cool [18:54] I'm good [18:54] MacSlow, good thing is - Ubuntu in there doesn't even know it's running on an emulator [18:54] so, let me play with this [18:54] jdstrand: :D [18:54] I can resurrect some old scripts and maybe hand them off [18:54] MacSlow, from our PoV it's just another device - albeit a slow one [18:54] summary time [18:55] Saviq, ~4 times slower than a N10 I'd say [18:56] oh, that was fast. /me found the scripts :) [18:56] Saviq, ogra_: I'll give it a shot tomorrow === ChanServ changed the topic of #ubuntu-uds-core-1 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1311/core-1/ - http://irclogs.ubuntu.com/2013/11/21/%23ubuntu-uds-core-1.html