[00:00] unfortunately, it's also taken a really long time to fix, and there are still lurking issues [00:01] awe_: well, I've seen lots of discussion; what I haven't seen on the list is "the bug is fixed in devel, if you still have problems after upgrade here's how you you recover without having to reflash" or so [00:02] but the bug has been mentioned in the ML, and the bug does have explicit instructions [00:05] awe_: as far as I'm concerned, if there aren't explicit instructions on how to recover in the bug description, this doesn't count ;) [00:13] slangasek, workaround added to the bug description [00:27] awe_: cheers [03:34] === trainguard: IMAGE 135 DONE (finished: 20140717 03:35) === [03:35] === changelog: http://people.canonical.com/~ogra/touch-image-stats/135.changes === [05:55] stgraber, ogra_ : uh, according to imgbot, image 135 took 9.5 hours to build... [07:40] sil2100, there were new kernels, new hybris and a new android build ... we should roll an image ASAP to cover these [07:41] sil2100, didn't someone try to address the filemanager failures already? [07:41] * ogra_ doubts that ... this is a fresh release [07:41] (of filemanager) [07:44] ogra_, what do you mean? it's had failures since 132 [07:50] brendand, the same ones ? [07:51] * ogra_ sees a new failure in 135 [07:52] ogra_, yes there is an extra one in 135, but the other two have been happening since 132 [07:52] right, but the new one seems to actually be caused by the new code [07:54] brendand: popey's fixes should have helped, they didn't? [07:54] ogra_: yeah, let's kick a new image indeed [07:55] sil2100, which fixes? were they meant to fix the Places tests? [07:55] sil2100, try out your new powers ;) [08:01] in [08:02] brendand: popey mentioned something about a fix in the emulator [08:02] So not sure [08:02] ogra_: oooh ;) [08:02] * sil2100 tries [08:05] sil2100: brendand: do you want me to rerun uitk tests? [08:05] that has a qmlscene crash produced during the test [08:05] psivaa, yeah [08:06] brendand, psivaa: yeah, I guess this is the flakyness as before [08:06] psivaa, and system-settings [08:11] there was a new file manager in 135 which was supposed to fix the tests [08:13] well, it didnt [08:16] Are the same tests failing? [08:16] yeah [08:16] sil2100, the two same ones and an extra one [08:16] and a new one [08:16] Mirv, good morning :) [08:17] sil2100, ping [08:17] tvoss: pong :) [08:18] sil2100, for line 38: I'm aware of the conflict with silo 2, would like to fast-track that mp, though [08:18] I'll have to ask balloons to help me with filemanager when he wakes, it was his fix I pushed through [08:18] I don't know why it's failing. [08:18] unless someone else is an autopilot expert? [08:19] popey, this was meant to be it? https://bugs.launchpad.net/ubuntu-filemanager-app/+bug/1342336 [08:19] Ubuntu bug 1342336 in Ubuntu File Manager App "File Manager test fails. " [Undecided,Fix committed] [08:19] tvoss: ok, so let me give you an ignore-conflicts, but please rebuild silo 002 when this lands [08:19] sil2100, sure [08:21] tvoss: hello! [08:21] brendand: yes === sil2100 changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: cihelp | CI Train Status: #133 promoted | Need CI Train help? Ping: mvo_, trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [08:21] Mirv, hey, cancel my ping [08:21] tvoss: and problem solved, I see [08:21] Mirv, yup :) [08:32] grrrr, hangout problem [08:36] sil2100, got 5 to help with a packaging question? [08:36] tvoss: in a meeting right now, but sure [08:49] popey, i reset https://bugs.launchpad.net/ubuntu-filemanager-app/+bug/1342336 to New [08:49] Ubuntu bug 1342336 in Ubuntu File Manager App "File Manager test fails. " [Undecided,New] [08:53] sil2100: brendand: ogra_: the uitk has 21 failures on the re-run too :/ and a qmlscene crash [08:53] sil2100, the extra failure in filemanager is not reproducible. it's probably down to the crash [08:54] psivaa, well that would do the trick [08:54] definitely something wrong there [09:19] robru: It was started, failed to build, and then way later it was retried [09:19] And the bot (a) is predicting the next image number (b) doesn't notice failures [09:23] Mirv, sil2100 silo 12 could do with some publishing [09:23] * tvoss being annoying [09:25] tvoss, who is not annoying when trying to land a silo? hehe [09:25] mandel, :) [09:31] tvoss: will do [09:31] mvo_, thanks :) [09:31] sorry for the delay [09:33] thanks mvo [09:42] Damn, it's so hot here... [09:43] sil2100, there's supposed to be a heatwave coming here, but i've not seen it yet [09:43] ogra_: do you know if it is possible to boot an emulator without the intro settings? I need to deploy an emulator straight to the shell. [09:43] well, it's going to reach 26 by the afternoon. that's what we call a heatwave here [09:45] heatwaves are relative :) feels pretty unpleasant here [09:49] sil2100: I'm off for lunch now and will resume duty in ~45min or so [09:49] cjwatson: most britons i know say anything above 20 degrees is "horrible weather" =))))) [09:49] mvo_: ok [09:49] === trainguard: IMAGE 136 DONE (finished: 20140717 09:50) === [09:50] === changelog: http://people.canonical.com/~ogra/touch-image-stats/136.changes === [09:53] who is working on qtubuntu to have a look at https://bugs.launchpad.net/ubuntu-app-launch/+bug/1329141 ? [09:53] Ubuntu bug 1329141 in Ubuntu Application Launcher "qmlscene crashed while running test_can_launch_multiple_applications" [Critical,Confirmed] [09:53] xnox: aye [09:54] it is basically blocking all merges to UITK [09:54] tvoss: ^ any ideas? [09:57] new camera in the store with hdr support... [09:57] t1mp, haven't seen the bug before, don't know if someone is working on it [09:58] tvoss: do you know who I can ask about it? it looks like nobody is working on it [09:59] t1mp, hmmm, perhaps loicm has got an idea === psivaa changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: psivaa | CI Train Status: #133 promoted | Need CI Train help? Ping: mvo_, trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [10:00] tvoss: ok, thanks I'll try to catch him [10:00] t1mp, yw === sil2100 changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: psivaa | CI Train Status: #133 promoted | Need CI Train help? Ping: trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [10:08] hmm, system settings crash on 135 [10:34] popey, we need to be more regular in landing new versions of click apps [10:34] popey, it would be way easier to debug if we didn't have 12 revisions to consider [10:37] popey, it's probably this rev that did it: http://bazaar.launchpad.net/~ubuntu-filemanager-dev/ubuntu-filemanager-app/trunk/revision/213 [10:37] popey, it's the same one that landed by accident before and broke filemanager in the same way [10:38] brendand: on a hangout, will reply to this in a bit [10:38] note that kalikiana's u1db-qt's changelog was overblown probably by some missing tag, but it only contains one change (the first one) [11:02] brendand: I have had conversations with dpm and fginther about auto landing clicks in the store, which takes the human (me) out of the loop [11:02] brendand: but that work hasn't been done yet, in the meantime we look at the clicks generated and push to the store fairly regularly, and as you can see from the spreadsheet we set alarms (yellow/red) when we dont update apps often enough [11:03] brendand: https://docs.google.com/a/canonical.com/spreadsheet/ccc?key=0AnZdnhOl8MU5dDJseW1vT1N5RkJvLUJHZTdhalRVd1E&usp=drive_web#gid=1 [11:03] popey, the delta should be revs not days [11:03] brendand: its both [11:03] there are more columns [11:04] popey, i see it now. days is not really relevant though [11:04] maybe not to you, but it is to me [11:05] popey, well i meant from a quality point of view. from a development point of view you might want to see if progress is being made, so yeah [11:05] exactly [11:06] I would like us to have fully automated clicks landing in the store. But we're not there yet. [11:06] popey, ok, good to hear that it's planned though === MacSlow is now known as MacSlow|lunch [11:27] om26er: hi! [11:28] om26er: so, could you dogfood an image for us promotion-wise? [11:28] om26er: smoketesting for #136 are still running but I suppose it might be a candidate [11:28] om26er: could you give it a spin? [11:28] * sil2100 goes for lunch [11:28] o/ [11:31] sil2100, ok [11:32] tvoss, are you actually abusing your silo for test builds and dvelopment ? [11:32] * ogra_ wonders if tvoss is aware of the 100s of mails he sends aroudn with each of these failing builds [11:33] ogra_, well, kinda :) sorry for the spam [11:33] use the canonical-arm-dev PPA ;) [11:33] ogra_, not sure that I have access there :) [11:34] now you do :) [11:34] (for the next time at least ) === psivaa changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: cihelp | CI Train Status: #133 promoted | Need CI Train help? Ping: trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [12:25] why the double post? === MacSlow|lunch is now known as MacSlow [12:46] om26er: any luck with dogfooding? :) [12:47] sil2100, ingprogress [12:47] so far so good. [12:48] sil2100, Mir 0.5 tested well, and will probably turn "testing done" green soon... Any blockers to land silo 18 on your end? [12:49] camako: I think we should be good to land it normally today - no regressions in sight? [12:49] sil2100, last round of testing about to be done... No regressions. [12:49] * camako crosses his fingers [12:50] o/ === josepht changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: josepht | CI Train Status: #133 promoted | Need CI Train help? Ping: trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [13:09] ogra_: thanks for building 136 [13:09] thanks for the kernels and andriod :) [13:10] :-) [13:12] rsalveti: we just hope you didn't break anything with those! As we're testing that image for promotion [13:12] ;) [13:12] oh, :-) [13:13] hm, reboot/shutdown dialog should be part of the next image [13:13] nice [13:15] sil2100, Hi [13:15] sil2100, so videos are not launching from preview in the latest image [13:16] ouch [13:16] om26er: that was tested on the latest promoted one, yes? [13:16] sil2100, yes, it worked fine in that version [13:17] om26er: ok, no promotion then I guess, since I suppose its a promotion blocking regression [13:18] A visible one so to say [13:19] om26er: could you fill in a bug? I wonder what could have caused this actually... [13:23] ok, I see some media-hub and similar landings in 135 [13:24] om26er could you switch to 135 and confirm its broken there as well? [13:31] The interesting thing is that the media-hub/mediascanner uploads were only no change rebuilds for the new dbus-cpp [13:31] And the mediaplayer-app upload was only a small change IIRC [13:33] self-publishing ^ [13:33] because even that kind of polishing is certainly not August's thing [13:33] Mirv: ok! === Ursinha is now known as Ursinha-afk [13:36] sil2100, where to find diff between images [13:36] http://people.canonical.com/~ogra/touch-image-stats/ [13:38] sil2100: tvoss asked you about this and I tried your incantation (which was what mterry suggested originally). It doesn't work. https://code.launchpad.net/~mir-team/mir/libmircommon/+merge/226704 [13:41] if there becomes a situation where someone would like to test whether Qt 5.3.1 would have a fix for the problem, see the wiki page https://wiki.ubuntu.com/Touch/QtTesting [13:41] note that I haven't found any reported bugs that would be fixed (we've cherry-picked many patches already), and it's not going in before RTM anyway. but just in case it might bring extra information to test on it. [13:48] sil2100, finished testing mir 0.5. All good. [13:51] sil2100, is there a way to upgrade from 133 to 135 ? [13:53] om26er, it should ooffer 135 to you if you have 133 installed [13:53] (it always offers to upgrade to the latest) [13:53] ogra_, 136 is the latest [13:53] oh [13:53] ogra_, I want to upgrade to 135 [13:53] well, then use ubuntu-device-flash [13:53] it will save me a lot of time [13:53] hmm, complete download then. [13:53] ubuntu-device-flash --channel=devel-proposed --revision=135 [13:54] ogra_, system-image-cli doesn't have anything to force a build [13:54] ? [13:54] i dont think you can enfocre a revision to go to ... only one you "come from" [13:55] sil2100: silo-018 needs ack for a new binary package (libmirserver23) - the debdiff looks just fine, its just a packagename bump, what the procedure? it says I need to talk to a archive admin [13:57] om26er, hmm, probably the --filter option might help [13:57] if you force "delta only" it probably offers you the smallest delta, which would be 135 [13:57] but not sure [13:58] mvo_: I'll look [13:59] thanks cjwatson [13:59] mvo_: you've acked the packaging changes? [13:59] cjwatson: not yet [14:00] should i? [14:00] mvo_: Well, a core-dev will need to [14:00] mvo_: Anyway, the new binary package is fine, AA ack [14:01] ok, done [14:01] thanks [14:07] Damn, I feel terrible today, must be because of the heat === Ursinha-afk is now known as Ursinha [14:16] om26er: hm, since I might have missed it, but did you fill in a bug for the b0rken video-playback from scopes? [14:17] sil2100, I will once I I try it on 135 [14:17] its still being downloaded [14:17] Ok, makes sense [14:18] tvoss: ping! Did you have a moment to take a look at bug LP: #1338610 ? [14:18] Launchpad bug 1338610 in platform-api "indicator-location crashing during default, sdk and click_image_tests tests on smoketesting" [Undecided,New] https://launchpad.net/bugs/1338610 [14:19] tvoss: I see charles pointed at you for more information [14:21] popey: btw. since I'm completely not a click person, do you think the change you made to mediaplayer-app to hide the icon could have affected it not launching videos from the scopes? But I guess you only added a NoDisplay=true, right? [14:21] sil2100, shelved, busy getting a silo landed right now :) [14:21] sil2100, but will get to it asap [14:21] sil2100: it shouldnt [14:22] tvoss: thanks :) It's not a big issue, just wanted to make sure it's queued up in your TODO list somewhere [14:22] sil2100: jhodapp tested it [14:23] sil2100, with an emphasize on "somewhere" :) [14:23] sil2100: no, there's no way it would do that [14:23] ;) [14:24] sil2100: does the mediaplayer-app come up after launching from the video scope, or not even coming up? [14:24] om26er: ^ [14:26] it doesnt launch at all from video scope [14:27] * popey reboots and tries again [14:28] hmmm [14:29] Actually, the offending landing might have been in 134 even from what I see [14:29] As libunity-scopes2 landed [14:29] mhr3: hi! What did the recent unity-scopes-api landing carry? Could it have been responsible for the regression we're seeing with video playback not launching properly [14:29] ? [14:30] sil2100, no [14:30] sil2100, ah, wasnt that the one with the extremely wordy changelog ? [14:30] :P [14:31] tedg, did you land the url-dispatcher change? === Ursinha changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: Ursinha | CI Train Status: #133 promoted | Need CI Train help? Ping: trainguards | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [14:31] mhr3, Which one? === alan_g is now known as alan_g|tea [14:32] tedg, file -> video? [14:32] mhr3, No, I marked it as land in a silo with the scopes. [14:32] mhr3, I did it, and charles reviewed, but it's waiting. [14:32] sil2100: jhodapp http://paste.ubuntu.com/7809300/ thats what I get when I launch a video from the video scope [14:33] tedg, k, does look like ual though^ [14:33] jhodapp, the app doesn't come up [14:34] also http://paste.ubuntu.com/7809306/ in my unity log [14:34] jhodapp, the app launching animating appears half onscreen and then goes away [14:34] ogra_: exactly ;) [14:34] om26er: yeah I see that myself on the latest image [14:34] sil2100, sounds like a desktop file gained NoDisplay=true, and something doesn't like that [14:34] where is robru...i need to sing praises over this dashboard...man this thing is great! [14:35] mhr3: should it be something else? [14:35] the simple test would be to edit the desktop file on the device and restart and try again [14:36] popey: yep [14:36] ogra_: curious, when do you guys plan on an image spin ? (had 2 landings happen i'm curious about) [14:36] jhodapp, tedg will probably know why he doesn't like NoDisplay :) [14:36] tedg: ^ [14:36] kgunn, we had one in the european morning ... i think sil2100 wanted one immediately after Mir landed [14:37] jhodapp, mhr3, if it's NoDisplay, it's not an application. [14:38] kgunn, so i guess in a few hours, once Mir has migrated to the archive [14:38] cool [14:38] tedg: so what's the proper way to not display the icon in the app scope but still have it launch? [14:38] jhodapp, Uhm, not. Why would you want that? [14:38] (or faster if it moves quicker) [14:39] tedg: mediaplayer-app can't do anything except when launched from the video scope [14:39] tedg: so it makes no sense to launch it straight [14:39] ogra_: I think Mir landed, so let me double-check and maybe kick a new image [14:39] popey: yep, works again after removing NoDisplay=true...so odd since it was working fine for me, no idea how [14:39] tedg, imo that's misinterpreting NoDisplay [14:39] sil2100, far from landed ... it only entered proposed [14:39] (about 20min ago) [14:39] ogra_: ok, then nvm! [14:39] oh dear [14:40] tedg, you probably want new X-UbuntuNoRun [14:40] tedg: so we definitely need a way of doing that [14:40] jhodapp, It makes no sense to me to have something that shows up as an app but then not have it in the app scope. Why not show the list of local videos when started? [14:41] tedg: because that's not what design designed for the mediaplayer-app [14:41] tedg: mediaplayer apps work this way on iOS and I'm assuming on Android as well [14:42] jhodapp, How do you select content independent of an app on those? [14:42] tedg: you launch it from somewhere else [14:42] tedg: like gallery, browser, etc [14:44] jhodapp, So yes, you can launch the video player that way, but it also shows up in applications. And then shows the recently played videos. (on Android) [14:44] tedg: ok, but that's besides the point...for Ubuntu Touch we don't want to display the mediaplayer-app icon in the app scope [14:44] mhr3, I'd argue that NoDisplay misinterprets the point of a Desktop file :-) [14:45] tedg: is there a way to accomplish this and still have it able to launch? === greyback is now known as greyback|lunch [14:46] tedg, maybe it is hacky, but it's just "dont display in menu" and not "don't dare to run" [14:46] jhodapp, As an application no, but the UX is seeming more like a trusted prompt session to me at this point. [14:47] tvoss, thoughts? [14:47] tedg: can you explain that some more? [14:48] jhodapp, Basically it's a set of Mir surfaces that are tied to an application. So I'd be saying "this is the gallery" even if on top of it there's a video playing in another app. Unity would present it as "Gallery" not an independent app. [14:49] Which makes sense to me. If it's not an application, it should be presented as part of something that is. [14:49] tedg: sounds reasonable, but what about the case where you launch it from the video scope? [14:49] tedg: what would it be apart of then? [14:49] jhodapp, the dash [14:50] jhodapp, That's what we're doing for purchases from the Click scope for instance. [14:50] tedg: implementation wise, how does it change? [14:50] tedg: from mediaplayer-app's perspective [14:50] tedg, so we plan to keep the payui icon in the click scope ? [14:50] * ogra_ finds that extremely confusing ... why isnt that part of the store [14:50] ogra_, No, once trusted prompt session support lands it won't be an application anymore. [14:50] same here [14:51] jhodapp, You need a trusted helper to setup the prompt session. Then you tell media player to use it. [14:51] tedg: you set that up from within mediaplayer-app's C++ code itself/ [14:51] ? [14:51] we have the media-hub server [14:52] No, more like the media-hub would set it up. [14:52] jhodapp, Here's how it works in Pay: https://wiki.ubuntu.com/Pay/Architecture [14:52] pay-service does the setup, and pay-ui is, well, the UI. [14:52] tedg: when is the trusted helper stuff landing, any idea? [14:53] jhodapp, First pieces just landed. We need a couple more to complete it. [14:53] jhodapp, I was *just* harassing people about that :-) [14:53] tedg: ok, I doubt I could get to that before RTM though...is there a way we can "hack" remove the mediaplayer-app icon for now? [14:53] harass more ! [14:54] jhodapp, It's all software, there are always ways, but I'm -1 on them currently. [14:55] tedg: even if we had to blacklist mediaplayer-app in the click scope for now [14:55] Yup, -1 on that :-) [14:55] tedg: meaning you can't do it or don't want to do it? [14:56] he doesnt like to ... [14:56] :) [14:56] Don't want to. It's software, we can do anything. [14:56] tedg: but we have to do something like that and we have very limited time [14:57] tedg: it's very confusing to users to launch mediaplayer-app directly [14:57] jhodapp, I'd argue it's very confusing to users to see an app they can't launch in their panel. [14:57] tedg: but that's not up to us [14:59] I think this is one of those things we need to get right. So, I'd argue we have time for it. [15:00] tedg: that's in an ideal world...I know I don't have the time for it [15:01] tedg: you are welcome to take a stab at it if you have the time [15:03] tedg: what do you think? [15:05] jhodapp, I think spending the time on that is better than spending the time on adding work around and hacks. [15:05] jhodapp, Personally, I can't imagine getting to it. [15:05] Well, before the end of August. [15:08] tedg: right, so we need a very quick workaround and then we can file a bug to track actually doing it the proper way for an update to the RTM release [15:15] jhodapp, why can't media-player app do anything unless when run from the video scope? That sound super wrong [15:15] tvoss: it doesn't have a way of displaying a list of media to play from within itself [15:16] tvoss: I'm not sure, that's how design wanted it [15:17] bfiller, ^ [15:17] jhodapp, as I understand it, nothing prevents the media app from accessing content on the device, it is unconfined [15:17] tvoss: right, but there's no design for choosing content from within the mediaplayer-app [15:18] bfiller, can you help out here? jhodapp , couldn't we replicate the music-app design? [15:18] seems redundant to show content when we have the video scope [15:18] jhodapp: what's the question? [15:18] tvoss, volunteering to implement that ? [15:19] bfiller, we cant hide the mediplayer icon from the click scope it seems [15:19] bfiller: design never provided a way to play video content from within the mediaplayer-app if you launch the app directly and not from a scope [15:19] hiding it makes ubuntu-app-launch not consider it for execution [15:19] ogra_, jhodapp : that's right [15:19] bfiller: and so yes, we need to be able to hide the mediaplayer-app icon from the click scope [15:19] bfiller: NoDisplay=true in the desktop file makes the app not launchable [15:19] jhodapp: no we don't, the app displays an appropriate message when launched from the scope [15:20] this has been discussed before with deisgn [15:20] should be fine the way it is today === gatox is now known as gatox_lunch [15:20] uh, really ? [15:20] yup [15:20] with an icon that does nothing? [15:20] thats rather ugly ... [15:20] bfiller: I still get a lot of messages from people confused about mediaplayer-app when they launch it directly [15:20] popey, it does something [15:20] tap it :P [15:21] two options: hide it from scope or live with it :) [15:21] its just pointless to have an error message generator icon by default [15:21] bfiller: right, I'd like to hide it but there doesn't seem to be a way currently [15:21] I think we have a lot bigger problems then this guys [15:22] bfiller: agreed...that's why I wanted a simple way to not display the icon in the click scope, file a bug to do it a proper way later, and be done with it [15:22] * sil2100 just reads in [15:23] So in the end the NoDisplay=true change caused the regression? I would say let's revert it and just leave the icon as it is [15:23] sil2100: yes, please revert it [15:23] * sil2100 checks if he can do it [15:23] sil2100: we can get you another MR that undoes it if necessary [15:23] Hah, I seem to have teh powa, let me try [15:24] jhodapp: apparently it's not that simple, tedg knows about it. It's what we wanted from the beginning but apparently there are issues with hiding it [15:24] bfiller: ok, so I guess we'll just revert it and live with the icon for now [15:25] jhodapp: but yes please file a bug about it if you don't mind [15:25] jhodapp: ok, so I could revert it manually (which would be faster probably) but since this change will be in for a while and is not jsut a 'quick temporary workaround', let me do it through CI Train so that it's all merged in and nice [15:25] bfiller: sure [15:25] sil2100: ok [15:28] hi all, I need some info on the procedure to build images... [15:28] the click scope needs to run a binary when the image is being built, to create a database of departments for the preinstalled apps. [15:28] We have the binary ready, and we'd like to know how to get it run in the image build process, and who to ping about this. [15:28] probably want to add a hook to livecd-rootfs [15:29] but actually [15:29] yeah [15:29] * cjwatson thinks [15:29] live-build/ubuntu-touch/hooks/60-install-click.chroot in livecd-rootfs should be the right place [15:30] is it definitely just preinstalled apps? does the database get incrementally upgraded with user-installed apps? how about carrier apps in /custom? if the database also includes user-installed apps, what happens on image upgrade? [15:30] yeah, probably an additional boot hook is needed [15:31] cjwatson: this database gets filled with new apps installed by the user, when they are installed by the click scope. [15:31] jhodapp: can I get an approval from you? :) https://code.launchpad.net/~sil2100/mediaplayer-app/revert_239/+merge/227208 [15:31] sil2100: sure [15:32] cjwatson: about carrier apps... I guess that the binary should be run after those are added by the carrier, yes. [15:32] sil2100: approved [15:33] cjwatson: regarding image upgrades... we need to think more about that point. [15:34] we have a boot hook architecture for that bit [15:34] (image updates always require reboots) [15:34] jhodapp: thanks! [15:35] ogra_: sounds good [15:35] ogra_: I would say, once we revert mediaplayer-app and mir migrates completely, let's kick a new image :) [15:35] alecu, whatever extracts the customization tarball should also call your binary though ... since that might not require a reboot ... talk to cwayne for that [15:35] alecu: this needs more thought before you put anything in livecd-rootfs [15:35] alecu: I would recommend that you have multiple databases in different bits of the filesystem [15:35] alecu: if that's possible [15:36] cjwatson: right. I'll take a look at that. [15:36] sil2100, yeah, i wonder what happens first :) [15:36] I'm auto-publishing this one once it builds, so I hope it will only take a few moments [15:36] alecu: if it has to be a single database, then you should make sure it's in the writeable area of the filesystem, and you should generate the whole thing at boot time using a system-level click hook, rather than putting anything in livecd-rootfs [15:37] depends how slow it is :) [15:37] alecu: though, thinking about it, if it's multiple databases, then you need to replicate all of click's logic for which of different versions of apps take precedence [15:37] alecu: so if it can be made fast enough to run at boot from a hook, then that would be preferable, IMO [15:37] alecu: and then you can just have a single database [15:37] cjwatson: the binary needs network access to generate this db, so that's why I'm proposing to generate it at image build time, or at carrier customization time. [15:38] oh [15:38] alecu: then the merging is distinctly problematic [15:38] and you MUST have multiple databases in that case [15:38] it just won't work otherwise [15:38] there'll need to be one for preinstalled, one for custom, one for user-installed - and the scope will need to replicate all of click's logic [15:38] the network bit at image build time might be problematic ... depending on waht you want to access [15:38] that too [15:39] is there a mail thread about this design? I think it needs discussion [15:39] but it will also be problematic on first boot ... before you even set up a SIM or WLAN [15:40] you could generate it once you have network access, perhaps? or do you need it before? [15:40] again ... depending how slow/resource hungry it is ... [15:40] really shouldn't need to be that slow [15:40] you dont want your phone go down to its kneew after enabling the wlan for the first time [15:41] *knees [15:41] cjwatson: I can start a mail thread with the thoughts on this so far. [15:41] the benefit of using click hooks for anything that needs to iterate over all installed packages is that click gives you a directory full of symlinks pointing to each of the installed apps; this honours things like users or carriers installing a newer version of a preinstalled app, users or carriers hiding a preinstalled app, etc. [15:41] also consider multi-user support [15:41] cjwatson: yes: we plan to have a local db for each user, with the apps they have installed. [15:41] if it's something visible per-user, then you might want to consider it being a user-level hook rather than system-level; individual users can hide individual preinstalled apps [15:42] alecu: is there a reason you can't just fish this out of the existing click database? [15:42]