=== dlan_ is now known as dlan === chihchun_afk is now known as chihchun [03:16] http://news.softpedia.com/news/Ubuntu-Touch-RTM-Officially-Released-Screenshot-Tour-459170.shtml [03:16] here some news === chihchun is now known as chihchun_afk [03:56] can i update devel version to rtm? [03:57] or will devel version get all stability as rtm? [04:01] lotuspsychje: If you have the option then devel will continue to get more enhancements for longer (as part of the Ubuntu 14.10 cycle) [04:02] duflu: i have the devel version on n7 right now [04:02] duflu: would i stay on devel or change to rtm? [04:03] lotuspsychje: We're seeing visible improvements regularly so if you want to continue with that then stay on devel. "rtm" will of course not change as much. That doesn't mean it's any more stable, but could become so. [04:04] duflu: well im reading rtm version has been improved on touch interaction [04:04] my devel version still little laggy [04:05] so i was curious about if rtm updates also come fro devel verion? [04:05] and will rtm have its own updates seperate? [04:06] lotuspsychje: For minimum lag, stay on devel. Otherwise you won't get any of the latest improvements. [04:06] okay tnx for info [04:06] you guys doing a great job! [04:06] lotuspsychje: But if stability concerns you, then certainly choose a more stable channel [04:06] lotuspsychje: Which N7? flo? (2013 model) [04:07] i wanna get updates [04:07] yes nexus7 wifi 2013 [04:07] lotuspsychje: OK, that should still get the latest code. The original N7 model will not. [04:08] duflu: you know what happens with the new 'set phone back to factory settings'? [04:08] will that reset ubuntu devel to original state? [04:08] lotuspsychje: Sorry, I'm not familiar with that option [04:08] ok [04:09] maybe ill try to clean out oldest stuff [04:25] duflu: just reset my nexus to ubuntu devel default [04:25] looking just nice === chihchun_afk is now known as chihchun [06:12] when i try to flash rtm/14.09 channel image to my flo, i get the error 'Failed to locate latest image information'. can anyone cofirm if it's expected? thx! === chihchun is now known as chihchun_afk [07:21] good morning === chihchun_afk is now known as chihchun [08:48] Good morning all; happy Monday, and happy Elephant Appreciation Day! :-D [08:55] cjwatson, mvo: when a click package gets removed, does ~/.config// get removed? [08:56] oops, I meant s/config/cache/, of course [08:57] mardy: there is a open bug about this, hold on a sec [08:58] mardy: https://bugs.launchpad.net/ubuntu/+source/click/+bug/1358294 [08:58] Launchpad bug 1358294 in click (Ubuntu) "App .config not removed when app uninstalled" [High,Triaged] [08:59] mardy: and lp:~mvo/click/lp1358294-config-removal that implements it [09:00] mardy: why do you ask? [09:01] mvo: the webapp-container is currently storing its data (such as browser cookies) under ~/.local/share/, which IMHO is wrong [09:02] mvo: I think it should use ~/.cache for cookies and temporary data [09:02] mardy: right [09:02] mvo: if click clears ~/.cache when a webapp gets uninstalled, that would one very strong reason to use this dir [09:03] oSoMoN: hi! Do you agree? (from ~10 lines above ^) [09:06] mvo: anyway, I understand that there is consensus at least to delete ~/.cache and ~/.config, so maybe you should land your branch? :-) [09:07] mardy, that sounds reasonable to me [09:08] oSoMoN: do you what me to change how "dataLocation" is exported by the Ubuntu.Web module, or only change it in webapp-container? [09:11] mardy: we probably need to export an additional "cacheLocation" property, as dataLocation is used in other places [09:12] oSoMoN: OK, so I'll add cacheLocation in Ubuntu.Web, and then use that in the webapp container [09:13] mardy, sounds like a plan, thanks! [09:18] oSoMoN: or actually, I should probably change the dataPath in UbuntuWebContext to be cacheLocation, I guess [09:20] mardy, probably, indeed [09:23] on a nexus 4 top in cyanogenmod shows 4 cpus but with ubuntu it only shows 2 cpus [09:23] actually it only shows 1 [09:24] cat /proc/cpuinfo # shows 4 on both [09:43] hi [10:11] hi [10:14] hello [10:33] oh, that's off, top sometimes shows 1 cpu, sometimes 2 cpus and soemtimes 4, i guess this is a feature not a bug :-) [10:33] s/off/odd/ === MacSlow is now known as MacSlow|lunch [11:39] ogra_: hey; we are missing a package in the seed [11:39] ogra_: signon-apparmor-extension [11:39] i guess it takes an MR against ubuntu-seeds, right? [11:51] dbarth_, yup [11:58] jdstrand: i guess starting on click-system-hooks still isn't early enough :( [12:04] ogra_: ok [12:04] mzanetti: do you remember how setting metadata when taking a picture was done ? [12:05] nerochiaro: uh... no... I don't think I've ever done that === _salem is now known as salem_ [12:24] cwayne: what are you talking about? === salem_ is now known as _salem [12:24] jdstrand: cache's still aren't being properly copied over from /custom on first boot [12:27] cwayne: that seem odd. did krillin get the new kernel? [12:27] jdstrand: not that i know of [12:27] bfiller: do you know who's responsible for the android camera code ? no matter how I set the correct datetime metadata in qtubuntu before capture, it always gets overwritten with the UTC version when capturing [12:28] cwayne: is this on the rtm branch or something else? [12:28] bfiller: so i need to ask someone who is more familiar with that code to see what i am missing [12:29] jdstrand: rtm [12:29] cwayne: this is an upgrade or a first boot? [12:30] jdstrand: first boot [12:32] cwayne: and you're sure the .features files, apparmor and appparmor-easyprof-ubuntu all match? [12:32] dbarth_, alex-abreu: can you help shuduo? in #ubuntu-app-devel he just asked the following: [12:32] dholbach: hi, i have a web app has published to app store and be updated with using webapp-container instead of webbrowser-app as your comment. but i just found the server side detects the webapp-container is a mobile version browser but it serves desktop version only. do you have any advice how to deal with it? I can't see how to set UA to webapp-container. thanks [12:32] file* [12:32] jdstrand: as far as I know yes [12:33] dholbach: joining [12:34] cwayne: can copy your current one aside as a backup and regenerate a new one, verifying apparmor is 2.8.96~2652-0ubuntu3, apparmor-easyprof-ubuntu is 1.2.22 and the features file matches? === MacSlow|lunch is now known as MacSlow [12:40] cwayne: so after verifying that, I think I may know what is happening === boiko_ is now known as boiko [12:43] zbenjamin_: if you check out gallery-app from trunk, can you build it out of the box in qtcreator now ? [12:43] zbenjamin_: i'm trying but it seems to fail on the po files [12:43] jdstrand: verifying it locally then need to do the same on jenkins [12:43] maybe the jenkins utopic pbuilder is out of date [12:44] jdstrand: how do i tell that the features file matches? === _salem is now known as salem_ [12:46] cwayne: look at /etc/apparmor.d/cache/.features on a booted system [12:47] jdstrand: ah ok, so those match [12:47] cwayne: so everything matches? [12:47] jdstrand: still running a local build to be sure, then need to run on jenkins [12:48] but the features file at least does [12:48] jdstrand: i've got 2.8.96~2652-0ubuntu4 of apparmor installed [12:48] and 1.2.25 of apparmor-easyprof-ubuntu [12:49] yeah, that's wrong [12:49] the rtm branch as different versions [12:49] zbenjamin_: nevermind, for some reason intltool wasn't installed in the chroot and cmake wasn't telling me about it [12:50] jdstrand: ah, shit... im doing all this from within a pbuilder, i wonder if we have rtm pbuilders... [12:53] System Settings is taking forever (about two minutes so far) to launch. Is there any way I can collect useful data to report a bug? [12:53] MacSlow: now that saviq is on holiday [12:54] Wellark, hey there [12:54] we need to schedule the sessions for Unity Dialogs and Unity Notifications [12:54] will I do it or do you want the honor? [12:54] these are the "burn snap decisions with fire" [13:01] jdstrand: theoretically I could just force a pbuilder to install those versions by giving it a sources.list i guess, right? [13:08] nerochiaro: rsalveti or jhodapp would be best to ask [13:09] nerochiaro: maybe that is intentional behavior to always store the timestamp in UTC [13:09] nerochiaro: but seems like it's UTC + 2 which would not be correct [13:10] nerochiaro: maybe it's the gallery-app that needs to change to convert the UTC timestamp to local time before adding to the event timeline [13:12] bfiller: i'm on UTC+2 and the timestamps recorded by camera are two hours behind, so that seems correct from my perspective [13:13] nerochiaro: try taking a photo on Android or another camera and see if it stores with UTC timestamp [13:14] bfiller: good idea [13:14] nerochiaro: it probably does make sense, otherwises apps would need to know what TZ the encoding was done in and convert to local time [13:15] bfiller: which would not be possible since the EXIF standard doesn't have any official provision for storing timezone tags or timezones in the timestamps === barry` is now known as barry_ [13:16] nerochiaro: right ok [13:16] nerochiaro: so instead lets fix gallery to convert timestamp from UTC to local time === barry_ is now known as barry [13:18] bfiller: but i just tried and android saves the picture with the local time in the metadata [13:18] nerochiaro: hmnn [13:19] nerochiaro: is there any indication in the metadata what tz it used? === dandrader is now known as dandrader|afk [13:20] bfiller: no, because as i said EXIF have no notion of time zones, it's not part of the standard tags. [13:20] nerochiaro: no custom tags used? [13:21] bfiller: a non standard tag for storing timezone information exists, but we are not using it, and it is now widely used in general [13:22] er, not widely used [13:23] …30 minutes later, System Settings is still trying to launch [13:23] bfiller: i guess if I manage to write that I might as well fix the date to be in local time [13:24] nerochiaro: I would say lets leave the TZ in UTC and in gallery-app do the conversion. It will at least be correct for pictures taken with our camera [13:24] nerochiaro: don't want to get into non-standard tags, etc [13:25] cwayne: I'm not a pbuilder expert (I use sbuild). you should be able adjust it fro ubuntu-rtm. there may be a facility to allow installing applications in it. you may want to talk to dholbach-- I think he is quite adept with pbuilder [13:25] bfiller: what about fixing it to be in local time, so we don't have to do any conversion and all apps can use it, and if I can't figure out how to do that by asking fix gallery ? [13:25] jdstrand: i think i've got it (at least locally) I was able to force installation of the correct versions of apparmor/apparmor-easyprof-ubuntu [13:26] jdstrand: so im going to build it on jenkins, then try an install from there and keep my fingers crossed :) [13:26] jdstrand, thanks for your confidence, but in the last 4 years I haven't had to use anything but pbuilder-dist {create,update,build} :) [13:27] Regarding adb access to my device: Am I correct in understanding that adb push pull is now restricted to /home? [13:27] nerochiaro: let me check what iphone does [13:27] heh [13:30] nerochiaro: iphone does the same thing, stores the local time. So yes I guess that is the correct thing to do then to fix it such that local tz being stored in the metadata [13:33] bfiller: it won't be storing the tz in the metadata. it will write in the metadata the local time without any tz information [13:33] bfiller: which is what android and i suppose ios do as well [13:33] nerochiaro: yes [13:33] nerochiaro: lets do that [13:34] bfiller: ok === gnu is now known as Guest64867 [13:36] rsalveti: jhodapp: does any of you know how to tell android what to put in the DateTimeOriginal and DateTimeDigitized EXIF tags when taking a picture ? I'm trying using AalMetaDataWriterControl::writeMetadata in qtubuntu-camera but it doesn't seem to respect what I set there [13:39] nerochiaro, let me check the other side of that function call to see what it's doing === datkinso1 is now known as datkinson [13:44] jhodapp: there is no other side, as far as i can see it just stores the metadata in an hash [13:45] blargh, apparmor : Depends: libapparmor-perl but it is not going to be installed [13:45] jhodapp: and i was assuming that android would pick that up when asked to capture [13:45] nerochiaro, you looked at the Android side? [13:45] jhodapp: no, i've never looked into hybris. the function i thought would pick up the metadata is android_camera_take_snapshot [13:46] nerochiaro, that's what I mean by the other side, let's look to see what Android does with the metadata === ricmm_ is now known as ricmm [13:46] MacSlow: so, how do you want to do this? [13:46] I have like 20 sessions to file [13:47] Wellark, ? [13:47] I don't mind filing the notifications and dialogs as well [13:47] Wellark, some context please :) [13:47] MacSlow: it's in the backlog! :) [13:47] MacSlow: from my "ping" [13:48] MacSlow: you replied 15:54 < MacSlow> Wellark, hey there [13:48] Wellark, saw it... bugs got the better of me again in between :) [13:49] Wellark, I can only do a "why snap-decisions must burn in hell" [13:49] MacSlow: I can enter them both [13:50] MacSlow: if that's easier [13:50] so it's a session of Unity Dialogs and Notifications [13:50] where we look into how we implement proper dialogs and purify notifications [13:50] Wellark, I'll hardly find time to work up a proposal for "unity dialogs" considering all the stuff I still have to fix/land [13:50] MacSlow: np. we have it covered [13:51] and the plan is to discuss them in washington [13:51] so no preparations needed [13:51] Wellark, if you've a session in mind for that I gladly help out (and/or do the talk with you) [13:51] MacSlow: ok. great! [13:51] Wellark, just put it anywhere and I'll be there [13:52] MacSlow: ack. :) [13:52] Wellark, so far I'm only there to work :) [13:52] MacSlow: lucky you [13:53] MacSlow: I'm pretty sure I'm just leading a meeting after meeting [13:53] as I said I have a list of around 20 right now [13:53] to fix all the "lessons learned" from the past two years [13:53] snap decisions being just one of them :) [13:53] or let's say [13:54] "to fix all the lessons learned that have made my life miserable" :) [13:54] Wellark, add "and MacSlow's" :) [13:57] cwayne: there was a perl transition after we branched from rtm [13:57] ie, utopic has perl 5.20, rtm does not [13:57] yay [13:57] consistency [13:59] nerochiaro, I don't see the function AalMetaDataWriterControl::writeMetadata in the qtubuntu-camera source === awafaa__ is now known as awafaa [14:00] nerochiaro, do you mean setMetaData()? [14:01] jdstrand: ah, just curious why it worked on my local pbuilder, but not on jenkins.. trying now to explicitly install libapparmor-perl, will see how that goes... === dandrader|afk is now known as dandrader [14:03] jhodapp: yes, sorry, setMetadata [14:04] nerochiaro, yeah, that only sets orientation for our implementation if you look at the source [14:04] nerochiaro, you'd need to implement setting timestamp if you want it to be changeable [14:07] jhodapp: well, ok, in the metadata class i can add a member like there is for the orientation, but who will pick that up ? [14:07] jhodapp: hybris ? [14:07] nerochiaro, what do you want to do exactly with the timestamp? [14:08] jhodapp: setting them to the current local time. android saves UTC time by default [14:08] nerochiaro, what is the timestamp used for though? [14:10] jhodapp: ordering the pictures in gallery [14:10] nerochiaro, so this is for the EXIF data? [14:10] jhodapp: yes [14:11] nerochiaro, ok, I would guess that there is a way to set this on the Android side...I'm not sure if we have the hybris code in place or not but my guess is we do have it by using the same code as setting the orientation [14:12] nerochiaro, I think you'd just need to find the right key name for the timestamp metadata === VargaD_ is now known as VargaD [14:12] jhodapp: ok. what is the branch of hybris that i should use ? i tried lp:libhybris but it does not seem to be the same as what is used by qtubuntu-camera trunk [14:17] nerochiaro, it's in our git repo...let me get you a link [14:18] nerochiaro, take a look online here: https://code-review.phablet.ubuntu.com/gitweb?p=aosp/ubuntu/libhybris.git;a=tree;h=refs/heads/master;hb=refs/heads/master [14:18] jhodapp: can i build that code on the device ? [14:18] nerochiaro, you can build the code that's in the hybris directory on the device [14:19] nerochiaro, the other code is part of the Android tree, so it must be built locally [14:19] nerochiaro, you can get the entire tree by doing "phablet-dev-bootstrap" [14:21] jhodapp: ok, i'm giving it a shot in a few minutes. i'll poke you if i have problems [14:21] nerochiaro, ok awesome, good luck [14:21] jhodapp: i assume once i rebuild the lib i can just overwrite the system one with it and it will work [14:22] nerochiaro, yes indeed === chihchun is now known as chihchun_afk [14:35] jamesh_: thanks for the comments on https://bugs.launchpad.net/gallery-app/+bug/1369652 [14:35] Launchpad bug 1369652 in gallery-app "Can't open photos from My Pictures scope" [Critical,Confirmed] [14:35] jamesh_: do you know who is working on integrating the My Photos scope with the mediascanner? [14:36] bfiller: it's kyleN [14:36] bfiller: The scope already uses mediascanner: it just isn't being maintained as part of the same source tree as the other mediascanner-using scopes === jamesh_ is now known as jamesh [14:37] cwayne: thansk [14:37] bfiller, I wrote it and it is an open question as to whether it will be integrated with mediascanner === chihchun_afk is now known as chihchun [14:38] kyleN: trying to figure out what is needed to make it use photo:// urls that we adding into the gallery-app [14:38] kyleN: seems like it should work like all the other scopes [14:38] I mean in terms of integrating with mediascanner [14:39] yes, it should use the same url (photos://... or whatever) for certain. currently is uses a file:// url [14:39] kyleN: huh? since when is that an open question [14:39] bfiller: It's really just the scope that needs to make sure its action button uses a photo:// URI [14:39] photo:// that is [14:40] bfiller: mediascanner itself only deals with files and doesn't know about these URI schemes [14:40] jamesh: got it, ok [14:40] jamesh: so I should probably remove that bug task for mediascanner then? === ev__ is now known as ev === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [15:02] charles: ping [15:04] nik90: pong [15:04] :) [15:07] pete-woods: hey, so usermetrics hasn't been working for a while. I am guessing it is because the /var/lib/usermetrics directory and its contents are owned by 'messagebus' [15:07] pete-woods: this may be related to bug #1363129 [15:07] bug 1363129 in rsyslog (Ubuntu) "No more syslog update after the 21st of August" [High,Confirmed] https://launchpad.net/bugs/1363129 [15:08] pete-woods: in that rsyslog stopped working around the same time because /var/log/syslog was owned by usermetrics === MacSlow is now known as MacSlow|afk [15:11] ogra_: fyi, it's possible bug 1359022 and bug 1372502 are related [15:11] bug 1359022 in unity8 (Ubuntu) "Welcome screen on image #200 always states that there are no data sources available" [Undecided,Incomplete] https://launchpad.net/bugs/1359022 [15:11] bug 1372502 in libusermetrics (Ubuntu) "usermetrics stopped working some time ago due to wrong group for files" [Undecided,New] https://launchpad.net/bugs/1372502 [15:12] ogra_: meh [15:12] o is so close to p :) [15:12] ogra_: sorry, it is possible bug 1363129 and bug 1372502 are related [15:12] bug 1363129 in rsyslog (Ubuntu) "No more syslog update after the 21st of August" [High,Confirmed] https://launchpad.net/bugs/1363129 [15:12] jdstrand, thats a duplicate [15:13] ogra_: I am not being clear [15:13] but i know what you want to tell :) [15:13] the UIDs change underneath [15:13] ogra_: the rsyslog bug is because /var/log/syslog was owned by usermetrics. usermetrics not working is because /var/lib/usermetrics/* is owned by messagebus [15:13] yes [15:18] ogra_: in http://people.canonical.com/~ogra/touch-image-stats/rtm/ I notice that 52 is the latest rtm image. But on mako, I have image 49 from rtm/devel-proposed which seems to be the latest. What's going on here? [15:19] * nik90 hopes there is a different image number based on the device it is produced for [15:19] isn't* [15:24] hi [15:26] jdstrand, sorry, took me a while to find it, duplicated properly now [15:26] nik90, stop judging images by the image number :) they are no expected to be in sync anymore [15:27] ogra_: how do I then find the image changelogs at your website if not with the image number? [15:27] ogra_: I am trying to find where a regression started appearing (alarms not working anymore) [15:27] nik90, http://ci.ubuntu.com/smokeng/utopic/touch ... take a look at the "build number" column [15:28] ok [15:28] you see three bits separated by columnns ... [15:28] the second is the rootfs [15:28] the changelogs for non rtm are all based on mako builds [15:28] the rtm ones are based on krillin builds [15:29] hmm the link is not opening for me [15:29] the only safe way to compare two images is by the rootfs [15:29] ok [15:29] well, a version string looks like: 48:20140921:20140919.1 [15:30] this would be rtm image 48 for mako ... it uses the 20140921 rootfs from http://cdimage.ubuntu.com/ubuntu-touch/ubuntu-rtm/14.09/daily-preinstalled/ and the device tarball from the 20140919.1 build === dholbach_ is now known as dholbach [15:31] nik90, and you are right, looks like the testing dashboard is down ... [15:40] jdstrand: that sounds plausible, I hadn't realised that libusermetrics had stopped working, though [15:42] pete-woods, bug 1332538 [15:42] bug 1332538 in system-image (Ubuntu) "No UID checks on rootfs updates" [High,Confirmed] https://launchpad.net/bugs/1332538 [15:43] stgraber, whats the status on that one btw ? we need it before thu. [15:44] ogra_: I'm working fulltime on non-phone stuff [15:44] slangasek: ^ [15:44] hmm, thats pretty bad [15:44] it seems to bite us more often recently [15:46] let's see if I can come up with a prototype in a couple of minutes [15:46] note, this isn't a new bug for me, just newly reported. I've lived with it for a little while [15:46] stgraber: this is a pretty important fix; if you can take care of it quickly that would be appreciated [15:47] slangasek: I'm not very familiar with live-build, so trying to figure out how to get a hook executed right after debootstrap and before anything else [15:48] hmm, I can't say I know either [15:49] stgraber, why do you have to do it at that early point ? [15:49] touch has hoos for "after build, before tarball is produced" [15:49] shipped in livecd-rootfs [15:49] *hooks as well [15:50] wouldnt that be sufficient ? [15:50] no [15:50] the idea is to create all the users and groups in the right order before the packages are installed === MacSlow|afk is now known as MacSlow [15:51] that way we guarantee their uids/gids. We also compare that list of initial users and groups with what we have at the end, if that differs, we fail the build. [15:54] nik90, Looks like clock icon broke: https://errors.ubuntu.com/problem/bf60f656677183c4c5f4b017d94a3cb690c0ea7b [15:55] tedg: I do not have permission to see the data [15:56] nik90, Hmm, okay. It's just a UAL error saying it can't find the icon in the click package. [15:56] nik90, clock 3.1.109 [15:56] tedg: hmm I did see that warning in the click-reviewers tool, but when I installed the click the icon appeared fine === chihchun is now known as chihchun_afk === alecu is now known as alecu_afk [16:00] mpt, I've got a couple "bad url" errors because the app is not installed that they're trying to switch to. [16:01] mpt, I think as we discuss dialogs for url-dispatcher on conflicts, perhaps we should add one to install apps as well. Or go to the click scope with a search. [16:01] kenvandine, I now have a precommit hook that runs the pep/flake test. Sorry that it took me months to get here. :p [16:04] jgdx, no worries [16:05] jdstrand: so I made sure to have the right versions of apparmor and apparmor-easyprof-ubuntu, but it still seems to be compiling on first boot :/ [16:06] tedg: mzanetti do you know if we have a bug filed / tracked for the fact that if an app (for example browser) is suspended by lifecycle then you can't restart it from another app (for example clicking a link in an ebook reader). [16:07] cwayne: when the system boots, does the custom tarball rely on the click system hooks to be run or are they already in place? [16:07] popey, That should work, url-dispatcher should resume it. [16:07] popey: not that I'm aware of [16:07] * mzanetti tries [16:07] popey, Well, it sends a focus event to unity8 [16:08] popey: have an easy test case handy? [16:08] jdstrand: well when it first boots, the caches are in /custom, and then we copy them to /var/cache/apparmor before system hooks are run [16:08] its easy to reproduce, just open browser, then open something else with links in it [16:08] jdstrand: do we need system hooks for anything else? [16:08] twitter, beru (ebook with urls in it) [16:08] * mzanetti generates a qr code with url in it [16:08] then scroll through twitter for a while, wait for browser to get killed/suspended, and then click a link in twitter [16:09] Hmm, if that's broken I'm going to have a hard time with this URL Dispatcher landing, that's part of it's acceptance tests :-) [16:09] oh, also, is there a bug for the low-quality screenshots you see when an app is suspended? [16:09] hah [16:09] soz [16:09] cwayne: are you making sure to preserve the time stamps, otherwise copy order is important [16:09] ah... hmm.. then its not suspended, but stopped [16:09] tedg: ^ [16:09] popey: I don't think so. It was a deliberate choice, but design may disapprove [16:09] the compiled profile must be newer than the profile and all its includes [16:09] greyback_: looks pretty awful, blurry. [16:10] cwayne: the apparmor hook will look at the symlinks in /var/lib/apparmor/clicks. it will then see if there is a profile for it in /var/lib/apparmor/profiles. if not, it will generate one [16:10] jdstrand: and will then recompile because the profile is newer? [16:10] popey: we did it to save memory, and see if anyone really noticed. You're the first to comment ;) [16:10] ʘ‿ಠ [16:11] cwayne: that is probably what is happening-- the click apparmor hook is generating the profile in /var/lib/apparmor/profiles new, and then apparmor_parser is called and sees the timestamp is newer [16:11] that was a detail I forgot to consider initially [16:12] so, you'll need to ship both the cache file and the profile [16:13] cwayne: does that make sense? [16:14] let me check something [16:14] ogra_: what are those system and radio users and why are they regular users with a login shell and a >1000 uid/gid? [16:15] stgraber, they can be /bin/false (i never noticed they have a shell) [16:15] jdstrand: yeah, so i'd have to alter the upstart job to cp -nu the profiles into /var/lib/apparmor/profiles [16:15] stgraber, they are driver or daemon users that the binary bits need to exist to work right [16:15] cwayne: actually, that is likely not enough. there are 3 pieces [16:16] cwayne: there is the click security manifest in /var/lib/apparmor/clicks, there is the profile in /var/lib/apparmor/profiles and there is the cache file in /var/cache/apparmor [16:16] ogra_: ok, and do the uid/gid of those matter? because based on what you're seeing, they really ought to be system users with a < 1000 uid/gid [16:16] if the click manifest is newer than the apparmor profile, the profile will be regenerated and so will the cache [16:17] if the profile is newer thatn the manifest, but also newer than the cache, the cache will be recompiled [16:18] jdstrand: so i should copy manifest, then profile, then cache? [16:18] popey: yeah... confirmed... doesn't work when the browser is stopped [16:18] so, you need to have the mtime of the file in /var/lib/apparmor/clicks < the mtime of the file in /var/lib/apparmor/profiles < the mtime of the file in /var/cache/apparmor [16:18] stgraber, yes, UID and GID are as essential as the name [16:19] jdstrand: man, this has ended up being a lot more complicated than just copying one dir from /custom :) [16:19] cwayne: yes that is the idea, however, you need to make sure the security manifest file gets handled correctly-- it is a symlink [16:19] stgraber, UID and GID are hardcoded in android all over the place (there is a header file that acts as /etc/password) [16:20] so we need to use the right numbers and names [16:22] cwayne: I think you should verify with cjwatson the best method for dealing with the symlinks in /var/lib/apparmor/clicks. I think the click database is actually the filesystem contents, so as long as they end up as symlinks, it should be ok [16:23] cjwatson: ^ [16:23] cjwatson: (you can read backscroll approximately to 6 minutes back for context [16:24] cwayne: note: for the file in /va/lib/apparmor/clicks, I am talking about the mtime of the symlink itself, not the mtime of the file it points at === alecu_afk is now known as alecu === gatox is now known as gatox_lunch [16:29] mzanetti: yay, need a bug filing? [16:29] popey: please [16:30] unity8? [16:30] popey: qtmir I think [16:30] k [16:30] popey: note: suspended == sleeping (ctrl+z) [16:30] popey: stopped == process ended, just fake screenshot in spread [16:30] ok [16:31] mzanetti: lolz, already filed it 2 weeks ago ㋛ https://bugs.launchpad.net/ubuntu/+source/qtmir/+bug/1368101 [16:31] Launchpad bug 1368101 in url-dispatcher (Ubuntu) "Suspended apps don't restart when launched from another app" [Undecided,New] [16:33] popey: interesting... [16:33] I really thought we wouldn't do so bad with at least looking at bugs [16:33] popey: sorry for that [16:34] no problem === dandrader is now known as dandrader|lunch [16:50] ogra_: so we want http://paste.ubuntu.com/8404456/ right after debootstrap and http://paste.ubuntu.com/8404457/ at the end of the build [16:50] that obviously expects proper behaviour from all the preinst/postinst scripts which is that they shouldn't fail if their user and group already exists [16:52] that should ensure ordering and successfuly detect any uid/gid change coming from either the base system or from added/changed packages [16:53] stgraber, what is lxc-andsmasq (and why do we have it on the phone) :) [16:54] uuh [16:54] and why do we have www-daat [16:54] *data [16:55] stgraber, the code looks fine to me [16:56] jgdx: ping https://bugs.launchpad.net/ubuntu/+source/ubuntu-system-settings/+bug/1371810 who can I ask to triage this bug? [16:56] Launchpad bug 1371810 in ubuntu-system-settings (Ubuntu) "system settings hangs with white screen or crashes on check for updates " [Critical,Confirmed] [16:57] robotfuel, gatox would be my best guess [16:59] jgdx, robotfuel I can look at that, kenvandine just rewrote a pile of relevant code [16:59] pmcgowan: thanks [17:01] robotfuel, does that happen with other multi-level panels do you know? and do you have to visit the OS page first? [17:01] there was a similar bug triggere by OS page a while ago that seemed to disappear [17:01] pmcgowan: yes I have to visit the os page first. [17:03] robotfuel, ok so that bug is not gone completely, it used to happen 100% of the time [17:04] pmcgowan: it doesn't happen every time, it's racy. some times it happens the 1st time some times I have to do it 3 times before it hangs or crashes. [17:04] robotfuel, ok [17:12] robotfuel, can you add the build and device info to that bug, unable to reproduce it yet anywhere [17:14] pmcgowan: done, leo was able to reproduce on rtm image #1 [17:15] robotfuel, ok, may be tricky to find [17:15] pmcgowan: I found that late friday, I am in the middle of an update and will try with todays image. [17:17] ogra_: lxc-dnsmasq comes with lxc, that's the user used to run lxc's dnsmasq server when it's used (it's not on the phone) [17:17] sh [17:17] ogra_: www-data is now part of the base groups that you get after a debootstrap [17:17] err [17:17] ah [17:18] oh, really ? [17:18] wow [17:18] crazy decision [17:20] I guess that was simpler than have a dozen web servers all try to create/remove it, possibly with disastreous effects :) === dandrader|lunch is now known as dandrader [17:22] pmcgowan: the crash is still there in today's image version_detail: ubuntu=20140922,device=20140919-1b3e670,custom=1411084873,version=54 [17:22]