=== chriadam|away is now known as chriadam [00:19] can not set system time zone after appying the .writable_image flag, can anyone help? [00:26] In 2013 What Touch Screen Laptop Would You Buy? | http://askubuntu.com/q/352074 [01:02] How do I disable screen blanking in Ubuntu Touch | http://askubuntu.com/q/352082 === _salem is now known as salem_ === salem_ is now known as _salem === jhodapp is now known as jhodapp|afk [02:29] is there a way to backup touch to u1 ? === Namidairo`bnc is now known as Namidairo === chihchun_afk is now known as chihchun [03:14] anyone else have a broken camera on mako on image 71/72? [03:19] mfisch, just flash 71 .. now updating to 72 ... will keep posted once done .. [03:21] ejat: do you remember how to run an app from the command line and set the display? [03:22] which app ? nope .. sorry .. [03:24] previously when i update from 70 to 71 timezone working .. n now flash 71 and update 72 timezone not working :( [03:25] ejat: what do you mean? [03:26] time zone .. [03:27] camera not working [03:27] white screen :( [03:27] yeah and it's the same segfault i have with the infographics [03:28] but previously when i was in image 70 upgrade to 71 .. the camera working .. [03:28] and the timezone setting [03:29] * ejat but now ... camera not working .. timezone need to use phablet-config [03:30] ejat: okay I'll let cwayne know that it's broken [03:30] tx [03:31] ejat: I'll file that bug can you confirm it when I do? [03:32] ejat: https://bugs.launchpad.net/sevilerow/+bug/1233489 [03:32] Ubuntu bug 1233489 in The Sevilerow project "setting timezone doesn't work anymore" [Undecided,New] [03:33] ejat: please add the last time it worked for you [03:34] mfisch, ok .. [03:35] i think .. this bug 1195398 also sevilerow need to take a look for oem [03:35] bug 1195398 in ofono (Ubuntu) "[dialer-app] Is not possible to execute costumer service numbers like *144#" [High,Confirmed] https://launchpad.net/bugs/1195398 [03:36] ejat: That will get fixed in time [03:36] probably wont be on our priority list [03:38] :( so i can't check my credit balance and reload my data plan without that bug get fixed :( ... need to wait [03:38] no wifi? [03:39] wifi working .. [03:40] but cant interact with customer services .. that start with *number# [03:41] oh I thought you could do it on a website [03:41] I have unlimited data here but it's expensive [03:42] i have the unlimited data but with other mobile phones .. [03:42] its not only for data .. it also need when to reload / topup credit [03:43] needed* [04:02] cwayne: pong [04:06] Hi, could someone possibly help me with this problem: I'm trying to return my Nexus 7 to Android as I had updated to 4.3 and then tried to installed Ubuntu ontop which didn't work and then I read I need Android 4.2.2 on instead. So I'm trying to install 4.2.2 and I get: "./flash-all.sh: 17: ./flash-all.sh: fastboot: not found" with numbers 17 through 26. Anyone know how to fix that? [04:11] Don't worry worked it out and fixed it myself [04:11] Thanks anyway [04:35] Mirv: hey, would you know anything about the music scope not showing cover art thumbnails for local music? [04:37] cwayne: hi. no, I dodn't know about music scope unfortunately. [04:37] Mirv: ah, sorry. i'd seen your name on the landing pipeline for something about unity thumbnails, and thought that might be relevant :) [04:37] Mirv: any idea whom I could ask? [04:56] cwayne: ah, right, I'm just trying to make sure there are no regressions before releasing, but I don't know the internals [04:56] hmm [04:57] cwayne: this is just a wild guess, but maybe James who has been contributing to the mediascanner lately? https://code.launchpad.net/~unity-team/unity-scope-mediascanner/trunk [04:57] Mirv: seems reasonable, thanks! [05:05] I bought a nexus 7 2013 gen 2, is there anything i can do to help port it faster? Without harming the tablet.....im not afraid of flashing but if i install multi rom which is in kickstarter and they made thier mark.....would that be a safer method? [05:17] I see there isnt even a stable cm yet === widespread_kerne is now known as ignitiongtowngma [05:34] cwayne, my music scope showing only a few thumbnail only .. using local music [05:37] IS there any reason why Ubuntu Touch on my Nexus 7 thinks it is a phone. Or is that just how Ubuntu Touch is built? [05:38] Also can I mess with the security of the Ubuntu Touch? As I'm new to Ubuntu Touch and not quite sure how to. [05:38] Without just editing in bash. === ignitiongtowngma is now known as ignitiongtown [05:41] Powermaniac os that a first or second gen? [05:42] ignitiongtown: First gen [05:43] mfisch, camera app still not working in build 73 [05:43] ignitiongtown: I used the grouper images from here: http://cdimage.ubuntu.com/ubuntu-touch/daily-preinstalled/current/ [05:43] opss forget .. the bugs haven't fixed yet .. my bad [05:43] ignitiongtown: Also are the back, home and settings buttons supported in Ubuntu Touch? [05:45] ejat: I found that it's the same fault that's killing my infogrpahics [05:45] but I should be in bed, so I'm eod [05:45] As for whatever reason when I'm in the settings menu I can't go back [05:45] mfisch, u better go to bed 1st :) .... [05:48] * ejat looking at the changelog .. [05:56] Two more questions: What is the default root password after installing Ubuntu touch. Also is there a way to turn off the battery saving settings that change the screen brightness constantly? [06:00] Will I be able to install Ubuntu Touch on a generic Android phone in future? | http://askubuntu.com/q/352150 [06:14] Powermanaic i was asking because i have a second gen, i wish o [06:14] I wish i could help but im a noob [06:14] ignitiongtown: Oh okay =\ [06:24] QML SQLite code example needed | http://askubuntu.com/q/352157 [06:38] Problem in Call Through Modem | http://askubuntu.com/q/352166 [07:05] good morning [07:22] mzanetti, Saviq: hey there... [07:23] mzanetti, Saviq: is there something wrong with autopilot (on jenkins) atm? My branch that used to pass just fine, all of a sudden fails all over the place. Locally testing any of the failing ap-tests works just fine. === noob is now known as Guest61285 [07:28] pitti, [ 21.974853] type=1400 audit(1380573331.014:55): apparmor="DENIED" operation="open" parent=863 profile="/sbin/dhclient" name="/etc/writable/localtime" pid=1369 comm="dhclient" requested_mask="r" denied_mask="r" fsuid=0 ouid=0 [07:28] pitti, i guess you need to change some apparmor profiles too :) [07:28] Hello [07:29] anyone know anything about porting [07:29] hi all, it looks like the directory /var/lib/usermetrics/ has become owned by root on some images instead of by the usermetrics user [07:29] does anyone know why this could be? [07:32] ogra_: hm, that worked on my workstation; we have additional profiles on the phone apparently.. [07:33] ogra_: do we have a new image with all that stuff landed? [07:33] pitti, several [07:33] pitti, image 71 and above [07:33] ogra_: cool; reflashing, and fixing === vrruiz_ is now known as rvr [07:42] ogra_: added apparmor task for this to bug 1227520 [07:42] bug 1227520 in apparmor (Ubuntu) "Timezone changes are not working due to ro /etc and bind mounts" [High,In progress] https://launchpad.net/bugs/1227520 [07:42] great, thx === zsombi is now known as zsombi|afk [07:55] oSoMoN: hey! [07:56] ogra_: I still can't set my timezone with image 73 [07:56] went to system settings [07:57] time and date [07:57] didrocks, see above [07:57] apparmor="DENIED" operation="open" parent=863 profile="/sbin/dhclient" name="/etc/writable/localtime" pid=1369 comm="dhclient" requested_mask="r" denied_mask="r" fsuid=0 ouid=0 [07:57] ah, additional profile [07:57] missing apparmor fix [07:57] a pity this wasn't tested on phone as requested :/ [07:57] well, i tested the initrd changes far more than the UI parts [07:57] sorry for that [07:57] ogra_: please readd a landing ask for that one [07:57] no worry [07:58] I'll just try to communicate that nicely ;) [07:58] its no regression at least [07:58] tsdgeos: ping [07:58] hi ho === psivaa-afk is now known as psivaa [07:59] tsdgeos: morning! [08:00] tsdgeos: did you have time to look at category reordering problems? [08:00] nope :-/ [08:00] is this something needed for *now* ? [08:02] tsdgeos: not now, but definately this week... [08:03] ok [08:03] will try to squeeze it in [08:03] Saviq: I'll have to put some time on ↑↑↑ [08:04] pitti, can you make sure the settings app works after your fix before uploading ? [08:04] tsdgeos, k [08:04] tsdgeos: thanks [08:05] MacSlow, yeah, we need to get the gsettings schemas upgraded on jenkins [08:05] pstolowski: thank me when we fix the problem :-) [08:05] Saviq, ah ok [08:05] didrocks, can you access the otto runners for upstream merger? [08:06] Saviq, I've also updated the unity-notifications MR, which addresses most of the issues.... apart from the formatting grief [08:06] MacSlow, k [08:06] didrocks: hey [08:06] tsdgeos: I trust you and consider it done already ;) [08:06] * tsdgeos feels the presure on its neck === soul is now known as Guest92241 [08:07] Saviq: no, it doesn't even seem to be otto as "otto" ;) [08:08] oSoMoN: hey! how are you? [08:08] pstolowski, actually... how can this be... "Settings schema 'com.canonical.Unity.Lenses' does not contain a key named 'hidden-scopes'" [08:08] didrocks: good, and you? [08:08] pstolowski, ah, I know how can this be... we're upgrading libunity-core, but not libunity... [08:09] pstolowski, shouldn't libunity-core require a version of libunity that has that schema change? [08:09] oSoMoN: I'm fine thanks ;) [08:09] oSoMoN: did you get my email about the camera-app crash? [08:09] didrocks, do you know who can, in our timezone? [08:09] Saviq: maybe vila === fmasi_afk is now known as fmasi [08:10] Saviq: does unitycore need this key now? [08:10] didrocks: just saw it, reading [08:10] pstolowski, apparently [08:10] Saviq: let me check [08:10] Saviq: I would say it shouldn't, but... [08:10] pstolowski, what I can tell you is that unity8 crashes with that error [08:10] didrocks: what ? [08:11] didrocks: looks like a priority indeed [08:11] oSoMoN: yeah, it's frustrating we can't reproduce it [08:11] ogra_: hm, I re-flahsed and got 70; I need --channel=? [08:11] didrocks: I’ll try it on my side [08:11] ogra_: yes, I'll verify the settings app [08:12] * pitti tries devel-proposed [08:12] vila, actually nothing for now, fginther worked around this last night [08:12] pitti, --channel devel [08:12] Saviq: ha ok, is there an IRC log I can read about that ? [08:13] ogra_: danke [08:13] pitti, oh, wait, proposed is better actually [08:13] so follow your first intention ;) [08:13] ogra_: thanks [08:13] vila, I can tell you, libunity9 introduced a new gsettings key and now unity8, built against libunity-core fails, 'cause it doesn't depend on this new version of libunity9, but seems to require that key [08:13] oSoMoN: ^ [08:14] vila, the reason being the upstream merger mediumtests machines have unity7 installed, but it doesn't get upgraded on each run (unless package dependencies make it so) [08:14] vila, fginther moved us back to the VMs temporarily, where unity7 is not installed, hence the newer version of libunity9 will get installed [08:15] Saviq: and what version of libunity do you have? pre-Sep 6th? === fmasi is now known as fmasi_afk [08:16] pstolowski, whatever's installed on the machines, can't tell, even [08:16] Saviq: pfew, hackathon :-/ So the bug is libunity-core not updating its deps ? libunity9 not bumping its version ? ci-engine not... guessing which package versions it should install ? [08:16] pstolowski, 'cause on the autopilot machines unity7 is installed - and nothing's upgraded unless package deps tell it to [08:17] vila, I'd say -core should be requiring new libunity9 [08:17] pstolowski, what do you say ↑? [08:18] Saviq: ok, do you have a jenkins url to a job that failed because of that ? [08:18] vila, plenty! ;) [08:18] Saviq: hehe, most recent one then ;) [08:18] vila, http://s-jenkins:8080/job/unity8-autolanding/513/console [08:19] vila, with http://10.97.2.10:8080/job/autopilot-testrunner-otto-saucy/544/ being the actual failed job [08:19] Saviq: funnily enough, I was looking at http://10.97.2.10:8080/job/autopilot-testrunner-otto-saucy/550/ when I was pinged ;) [08:20] vila, yeah, that's from our -ci probably, I grabbed the latest -autolanding [08:20] vila, same issue, yeah [08:21] Saviq: and the symptom is the ProcessSearchError: Process exited with exit code: -5 [08:21] ? [08:21] vila, yeah, and, more importantly, " GLib-GIO-ERROR **: Settings schema 'com.canonical.Unity.Lenses' does not contain a key named 'hidden-scopes'" at the top [08:21] Saviq: great ! I was wondering if the -5 one was too generic, indeed. [08:22] Saviq: I see no trace of dependency on this key in libunity-core, still looking [08:22] pstolowski, :/ === fmasi_afk is now known as fmasi [08:26] pitti, [08:26] root@ubuntu-phablet:/# ls /etc/writable/ [08:26] root@ubuntu-phablet:/# [08:26] root@ubuntu-phablet:/# cat /etc/timezone [08:26] cat: /etc/timezone: No such file or directory [08:26] :( [08:26] something is wrong here [08:32] ogra_: http://cdimage.ubuntu.com/ubuntu-touch/daily-preinstalled/pending/saucy-preinstalled-touch-armhf.tar.gz has etc/timezone etc/writable with files [08:32] ogra_: hm, I flashed with "devel" and still got 70 [08:32] ogra_: might be a different channel than "devel" after all? [08:32] Saviq: I'd say it won't hurt to require newer libunity, the version it currently depends on is a bit old; but again, I can't find any use of this key in libunity-core; is it possible a newer libunity has been installed but the schema wasn't compiled for some reason? [08:32] don't know why they aren't symlinks in tar.gz [08:32] ogra_: danke [08:32] pitti, oh, wait, proposed is better actually [08:32] so follow your first intention ;) [08:32] * pitti tries devel-proposed, not knowing what that is [08:33] ogra_: ah, missed that [08:33] pstolowski, in theory, yes, but I'm not seeing it being installed on the machine [08:33] ogra_: well, s/intention/the only thing that --help says/ :) [08:33] pstolowski, http://10.97.2.10:8080/job/autopilot-testrunner-otto-saucy/550/consoleFull [08:33] pitti, i also think the linking code in livecd-rootfs is wrong ... you dont cd into /etc before creating the relative link [08:34] laney@iota> ls -l timezone ~/temp/unpacked/etc [08:34] lrwxrwxrwx 1 laney laney 17 Oct 1 07:22 timezone -> writable/timezone [08:34] it's right [08:34] pstolowski, here's the packages removed / upgraded / installed: http://pastebin.ubuntu.com/6178774/ [08:34] I'm guessing it is the 'synced' stuff that is breaking it [08:34] Laney, hmm, then it must be proken on boot [08:34] *broken === fmasi is now known as fmasi_afk [08:35] the system image stuff is only re-compressing afaik [08:35] * Laney flashes with d-proposed too [08:35] do you get logs from the writable-paths migrations? [08:36] ogra_: why would I need to cd to /etc? [08:37] echo "I: Linking /etc/$f to /etc/writable/" [08:37] ln -s writable/$f /etc/$f [08:37] ln -s doesn't care in which directory you run it (other than where to place a relative target, of course) [08:37] you are in / [08:37] I already showed you it's right [08:37] you specify the verbatim target of the link in ln -s, it doesn't translate it [08:37] don't understand why you're trying to claim it isn't [08:37] Laney, i see it is, i still wanted to answer pitti [08:38] I see [08:38] ah, so "synced" doesn't copy the initial files from the origanl image to /userdata/ ? [08:38] I thought it did / was supposed to [08:38] * Laney checks changelog [08:38] Saviq: ok, this looks weird... libunity9 should be upgraded together with libunity-protocol-private0 [08:41] Saviq: can we get versions of all the important packages on that machine? [08:42] what is the changes in build 74 ? [08:42] ogra_: oh, hang on -- might it be that restoring the backup might have broken this? [08:42] Saviq, mostly of protocol and libunity9 [08:42] ogra_: this isn't upgradeable, due to the broken change that landed in the previous livecd-rootfs (adding /etc/timezone to writable-files) [08:43] vila, can you get that info for mhr3_ and pstolowski ↑ ? [08:43] vila, just the versions installed in a "clean state" would be enough [08:43] ogra_, Laney: I just flashed "phablet-flash ubuntu-system --channel devel-proposed" which gives me image 74, and it's alright [08:43] /etc/timezone -> writable/timezone [08:43] pstolowski, mhr3_, yeah I did see the -protocol-private thing there and thought it could be the culprit [08:43] -rw-r--r-- 1 root root 8 Oct 1 06:08 /etc/writable/timezone [08:43] root@ubuntu-phablet:/# mount|grep writable [08:43] pitti: I'm trying it now too [08:43] /dev/mmcblk0p12 on /etc/writable type ext4 (rw,relatime,barrier=1,data=ordered) [08:43] i guess an empty /etc/writable is mounted on top of the one having the files [08:44] ogra_: yep, that would be the old one from the bad livecd-rootfs [08:44] Waiting for install to finish on device... [08:44] we don't seem to have upgrade quirks [08:44] ogra; so perhaps you need to add a --no-backup once? [08:45] pitti, u mean timezone r writetable in 74 ? [08:45] MacSlow, also, the fact that there is no longer the Hints enum is API change [08:45] ejat-: /etc/writable/{timezone,localtime} are writable, yes [08:46] pitti, --no-backup wipes all data [08:46] MacSlow, there's still Type::ExtSnapDecision, shouldn't all of that go away [08:46] ? [08:46] Morning alls! [08:46] ogra_: well, whatever is necessary to get rid of the broken /userdata/etc/timezone and friends [08:46] MacSlow, in NotificationModel.cpp [08:47] pitti, but still cant change the timezone from setting right ? [08:47] ejat-: I'm trying right now [08:47] ive tried [08:47] ejat-, yes, thats still broken [08:47] still not working .. [08:47] same goes to camera n infographic after build 71 [08:47] Ogra will probably be busy October 17th =) [08:48] hm, settings is broken; typing "augs" it shows me "Augsburg, DE", but I can't click on it [08:48] same for other cities [08:48] pitti, yes, try to go back now, go forward again and it should be selected (oramnge) [08:48] pitti, yups .. same goes to me .. find but cant click .. [08:48] the UI is behaving very weird [08:49] clock is gone again [08:49] did it change the timezone? [08:49] ogra_: back and forth makes it orange, but still not selected [08:49] and yes, the clock is gone from the indicators [08:49] the same now sometimes happens on desktop [08:49] orange but cant select [08:49] pitti, clock gone from indicators? [08:50] yes [08:50] Saviq, fixed [08:50] it should select if the timezone changes [08:50] pitti, the UI is confusing, what should happen is that the selected location should be an orange line [08:50] MacSlow, sorry for switching channels ;) [08:50] nothing else [08:50] Saviq, I can cope :) [08:50] pitti, you can try on your desktop [08:50] seb128: well, "Augsburg" is now in orange, but I can't select it nor is there an "OK" button or so [08:50] Is there a site we can go do to see apps for UT? [08:50] did it change? :( [08:50] pitti, if it's in orange, it's selected [08:51] pitti, the call happens when you click [08:51] it selects in response to changes, so if that fails to happen then it probably won't update [08:51] seb128: and if I go back, it still says "Etc/UTC UTC+0" [08:51] look at /etc/timezone [08:51] so the dbus call failed I guess [08:51] Saviq, works ok here ... https://code.launchpad.net/~saviq/unity-notifications/fix-themed-icons/+merge/188483 approved it [08:51] oh, wow [08:51] hang on [08:51] $ timedatectl [08:51] Timezone: Europe/Berlin (UTC, +0000) [08:51] MacSlow, cheers [08:51] ah [08:51] ive clicked the orange also its not select [08:51] so, it did try to select it, but failed [08:51] pitti, https://bugs.launchpad.net/ubuntu/+source/ubuntu-system-settings/+bug/1232114 [08:51] Ubuntu bug 1232114 in ubuntu-system-settings (Ubuntu) "Adding timezone behviour is odd" [Low,New] [08:51] $ cat /etc/timezone [08:51] pitti, i even have that in the UI [08:51] for the confusing UI [08:51] ejat-: we heard, please stop re-confirming [08:51] Etc/UTC [08:52] Saviq, only jenkins doesn't want to play along... so I didn't top-approve yet [08:52] * ejat- silent .. [08:52] MacSlow, yeah, let me have a quick look [08:54] ogra_, Laney: oh, I know [08:54] I wrote the systemd patch to check for symlinks to /etc/writable/ [08:54] but cjwatson suggested to make the symlinks relative [08:54] so they now start with "writable/" [08:54] * pitti adjusts [08:54] heh [08:54] MacSlow, right, it's running unity8 AP tests which are currently broken due to libunity/libunity-core version mismatch [08:55] MacSlow, we're trying to get to the bottom of it on #ubuntu-unity [08:55] ok [08:55] ogra_: btw, I don't have the AppArmor violation (not sure why, I should have it) [08:56] pitti: I admit that I assumed that nobody was using the symlinks wrongly ;-) [08:56] cjwatson: not your fault :) [08:56] (and it was a good suggestion) [08:56] oSoMoN: mind joining a hangout? === ejat- is now known as ejat [08:58] didrocks: sure [08:58] oSoMoN: https://plus.google.com/hangouts/_/2677f6ccd372954e7dfcef79ec21a8422315cc06 [08:59] it's a french hangout :p [09:01] pitti, https://jenkins.qa.ubuntu.com/job/saucy-touch_ro-maguro-smoke-camera-app-autopilot/111/artifact/clientlogs/dmesg.log/*view*/ we have it a few times on the test devices during test runs [09:01] ogra_: yep; I uploaded the apparmor fix already [09:02] i dont see it locally either here [09:02] anyway, what do we do with /etc/writable now [09:03] ogra_: give me some minutes, working on the systemd fix :) [09:03] k [09:03] :) [09:06] I recompiled powerd one day an told it to look at /sys/class/leds for brightness settings.. but it did not give me access to brightness changes in the UI... Is there another package I could look at? [09:32] pitti, i think i know whats wrong with /etc/writable ... [09:32] * ogra_ tries something === fmasi_afk is now known as fmasi [09:33] ogra_: I bet you still have it in /userdata/system-data/etc/ [09:33] pitti, it bootstrapped a new install, cant be [09:34] pitti, and even then it wouldnt do any harm [09:34] nothing would linka against it [09:35] root@ubuntu-phablet:/# ls /etc/writable/ [09:35] localtime timezone [09:35] yay [09:38] i still cant change iot though, but adding the right stuff to /etc7system-image/writable-paths works at least [09:38] (needs the full filenames [09:38] ) [09:39] ogra_: what did you have in writable-paths? should be [09:39] /etc/writable auto synced none none [09:39] thats not enough [09:39] root@ubuntu-phablet:/# mount|grep writable [09:39] /dev/mmcblk0p12 on /etc/writable type ext4 (rw,relatime,barrier=1,data=ordered) [09:39] /dev/mmcblk0p12 on /etc/writable/timezone type ext4 (rw,relatime,barrier=1,data=ordered) [09:39] /dev/mmcblk0p12 on /etc/writable/localtime type ext4 (rw,relatime,barrier=1,data=ordered) [09:39] that works :) [09:39] erk [09:39] why would you want to do that? [09:39] the code doesnt copy the files [09:40] pete-woods: ping [09:40] it does if you specify them [09:40] (i know thats a crude workaround) [09:40] ogra_: it's not a workaround [09:40] pitti, i can edit the files [09:40] it breaks the very reason why we introduced /etc/writable/ in the first place [09:40] well [09:40] pitti, then the whole design needs to change [09:41] root@ubuntu-phablet:/# echo foo > /etc/writable/timezone.new [09:41] root@ubuntu-phablet:/# mv /etc/writable/timezone.new /etc/writable/timezone [09:41] oSoMoN: hi [09:41] that works just fine in image 75 [09:41] err, 74 [09:41] so I don't see what's broken [09:41] pitti, if you copy anything to /etc/writable during image build thats not the 7etc7writable you have on a booted device [09:42] ogra_: but that's what "synced" was supposed to do, and it seems to work fine [09:42] and stephanes code doesnt copy any contents into the empty loop image he creates [09:42] pitti, /etc/writable is empty by default [09:42] ogra_: not here [09:42] pete-woods: hey, we’re seeing applications crash (when running smoke tests), and it seems the crash is in libusermetrics [09:42] I got the files in it here [09:42] ogra_: please don't upload anything which adds files to writable-files [09:42] oSoMoN: it's related to the writable problem that the other guys here are discussing [09:43] pitti, then you flashed something wrongly ... did you use --no-backup ? else you have all old data around [09:43] that is just plain wrong [09:43] agreed [09:43] oSoMoN: there's an update to the QML bindings that apps are using (not released yet) that will stop them crashing [09:43] ogra_: I used --no-backup the first time when I got 70 again (as I didn't specify --channel) [09:43] pitti, well, i thought you want localtime and timezone in there [09:43] ogra_: I didn't use --no-backup the second time [09:43] oSoMoN: but the underlying issue is the non writable /var/lib after update [09:43] ogra_: yes, and I do [09:43] pete-woods: ah, I didn’t quite follow the discussion, I’ll read the backlog [09:44] pete-woods: is there a bug report to track the issue? [09:44] pitti, right, flash again with --no-backup. /etc/writable is empty (and the code doesnt copy a single thing when mounting the whole dir into a loop image) [09:44] oSoMoN: https://bugs.launchpad.net/touch-preview-images/+bug/1233396 was logged against me [09:44] Ubuntu bug 1233396 in The Sevilerow project "infographics and camera-app broken in build 71/72 - core dumped" [Critical,Confirmed] [09:44] so i donbt see how you end up having anything in there [09:44] ogra_: ok, I'll try that; then it seems something in the new "synced" mode is buggy [09:45] right [09:45] * ogra_ re-flashes his maguro again to be sure [09:45] * pitti runs phablet-flash [09:45] $ phablet-flash ubuntu-system --channel devel-proposed --no-backup [09:45] right? [09:45] phablet-flash ubuntu-system --channel saucy-proposed --no-backup -d maguro [09:46] thats what i use here [09:46] so yeah, its the same [09:46] yeah, mako here but that should hardly matter [09:46] right [09:46] and the channel name is aliased [09:46] points to the same pool [09:47] * ogra_ wishes the flashing would take years on maguro [09:48] "not"? :) [09:48] heh, yeah [09:48] unpacking the tarball takes ~15min [09:49] didrocks: https://bugs.launchpad.net/touch-preview-images/+bug/1233396 [09:49] Ubuntu bug 1233396 in The Sevilerow project "infographics and camera-app broken in build 71/72 - core dumped" [Critical,Confirmed] [09:49] xz and ARM arent such good friends [09:49] ogra_: camera seems broken in mako [09:49] popey, in maguro too [09:49] k [09:49] didrocks, ^^^^ [09:49] oSoMoN: oh, nice catch! [09:49] popey: hum, it works here and for others as well [09:49] popey, didrocks claimed his worked though [09:49] so so… what do we do differently? [09:50] http://popey.com/~alan/device-2013-10-01-105001.png [09:50] phablet-flash ubuntu-system --channel saucy-proposed --no-backup -d maguro [09:50] thats what I get [09:50] I guess it's all linked to the AP crash [09:50] alan@deep-thought:~$ phablet-flash ubuntu-system --channel saucy-proposed --no-backup -d mako [09:50] open camera, get a white screen [09:50] thats what i did [09:50] i didnt do anything more [09:50] ogra_: I didn't try --no-backup [09:50] didrocks, ugh [09:50] do you think that's important? [09:50] didrocks, you always should to get untaintedt results [09:50] ogra_: this is clearing /home/phablet, right? [09:51] didrocks, yes, else your old data is preserved ... that might include settings etc [09:51] yes, and a few other bits [09:51] ok, let's try [09:51] its essentially a clean bootstrap [09:51] maybe that's the difference we see [09:51] definitely [09:51] oSoMoN: FYI ^ [09:52] "Usermetrics is not running. It's upset that the DB file is missing: [09:52] " [09:52] pete-woods: hey, around? [09:53] didrocks: hi [09:53] ogra_: yes, confirmed [09:53] great [09:53] pete-woods: so, it seems a lot of us are suffering from https://bugs.launchpad.net/touch-preview-images/+bug/1233396 [09:53] Ubuntu bug 1233396 in The Sevilerow project "infographics and camera-app broken in build 71/72 - core dumped" [Critical,Confirmed] [09:53] so how do we fix this [09:54] pete-woods: and it's what we are seeing in the AP tests [09:54] didrocks: the problem is that something has screwed up /var/lib/usermetrics and made it owned by root [09:54] (camera-app and media-app crashing) [09:54] blindly copying content seems evil, the content could grow over time and really harm boot speed at some point [09:54] didrocks: I have a fix for the QML bindings, which will stop apps crashing (obviously bad) [09:54] pete-woods: hum, ok, so libusermetrics trunk should be released? [09:55] didrocks: I need to fix something else, but yes, it should be shortly [09:55] pete-woods: ok, then, please try libusermetrics trunk on latest image [09:55] and run AP tests for camera-app and gallery-app [09:55] to ensure we don't regress those compared to latest promoted image [09:56] didrocks: will do [09:56] ogra_: I don't see what could have changed /var/lib/usermetrics perm from the diff on image #71 [09:56] udev shouldn't touch those [09:56] didrocks, are you sure you are barking at the right tree ? [09:57] if you didnt use --no-backup i would realyl re-do the tests [09:57] ogra_: it's installing the new one [09:57] just rebooting, let's see [09:57] yeah :) [09:59] ogra_: I guess we should wait for stgraber to fix the "synced" mode in initramfs-tools-ubuntu-touch [09:59] pitti, hmm... k ... [10:00] (note that we hold all image production until this is fixed) [10:00] ogra_: it's not such a big blocker as a nonexistant /etc/timezone will just count as "UTC" anyway [10:00] and as soon as you change it, the files ought to be created [10:00] but it shoudl be fixed anyway of course [10:00] ogra_: I have a number of people saying that the usermetrics service has broken, and from investigation it seems like /var/lib/usermetrics has become owned by root, instead of the usermetrics user [10:00] pitti, well its kind of a desaster since this was supposed to be tested enough to not break if it lands [10:01] ogra_: popey: screwed camera-app confirmed \o/ [10:01] pitti, and per the new rules we need to either roll back everything or have a fix really soon [10:01] \o/ [10:01] * didrocks likes when we all agree ;) [10:01] great ! [10:01] Ditto. [10:01] ogra_: well, we never had a release where time zones worked, so where would you roll back to? [10:02] pitti, no idea, its the rules :P [10:02] and we didn't roll back the broken livecd-rootfs either, because we can't realy [10:02] dont tell me [10:02] The rules are that we roll back if there's a regression, but this isn't a regression since it has never worked. [10:02] So that doesn't apply. [10:02] i only obey to the rules, i dont make them [10:02] Take responsibility for common sense! [10:03] * popey feels that should be a poster [10:03] cjwatson, right, i dont plan to roll back to less functional, the point is that the testing in this landing seems to have completely failed though [10:03] It still doesn't mean a rollback is necessary. [10:04] Was broken, is still broken, nothing lost. [10:05] i do agree ... and am happyy asac is on vac this week :) [10:05] (don't tell it top loudly ;)) [10:05] (he would insist to roll back) [10:05] I would argue the same thing to asac directly [10:05] so I confirm: drwxr-x--- 2 root root 4096 Oct 1 09:57 /var/lib/usermetrics [10:05] well, there is nothing to roll back *to* [10:05] i know [10:06] pitti, he would probably call it "back out", not roll back [10:06] and i still didnt manage to teach him that the only way back is forward :) [10:06] oSoMoN: I think you are fine, no need for you to continue on the mediaplayer/camera-app side [10:07] Our technical culture shouldn't include "do stupid things because the rules say so" [10:07] cjwatson++ [10:07] So let's not enforce that when common sense says otherwise :) [10:07] (And also let's not claim that somebody else is telling us to do so when they aren't here) [10:07] i was just mentioning what technically would have to happen, i wasnt saying i'd do it [10:07] didrocks: ok [10:08] ogra_: do we already have something that will unpack a rootfs tarball on first boot on the userdata partition, which I can either run from initramfs, or by booting into alternative initramfs/recovery first, and later rebooting again "normally" ? [10:08] xnox, not on touch, no ... we have that in the n7 tarball´-installer package [10:08] *tarball-installer [10:08] ogra_: ideally i'd like to simply drop the tarball and unpack it on first boot, with all permissions preserved & et.al. as at the moment mkyaffs2 image looses all permissions on all files (+x in particular) [10:09] ogra_: let me see, if I can refactor / lift some code of there. [10:09] xnox, how about you create a separate initrd script for this we can source from the touch script ? [10:09] ogra_: yeah, only conditionally when booting emulator though. [10:09] so that we keep the code a bit distinct between normal booting and emulator processing [10:10] * xnox nods [10:10] pete-woods: ogra_: popey: lool: confirming, giving write access to others on /var/lib/usermetrics works and I can now launch camera-app [10:10] didrocks: whew! :) [10:10] so what sets the permissions there ? [10:11] (or unsets them) [10:11] yeah, that's the question [10:11] pete-woods: you do use that folder for a long time, right? [10:11] pete-woods: it's not recent? [10:12] didrocks: months now [10:12] ok, not really newish then [10:12] didrocks: it gets created by the debian maintainer scripts for libusermetrics [10:12] oh oh [10:12] so maybe on the image generation side? [10:12] ogra_: as we don't run postinst, do you know of anything changing? [10:13] didrocks: could be, I have no idea how that magic works [10:13] didrocks, ? [10:13] that would explain why we can't trace that to any changes that happened to the image [10:13] why wouzld we not run postinst ? [10:13] We run the postinst, just not on the device [10:13] ogra_: that's what asac told me, there was magic to run successfully postinst in the image with fixing permissions [10:13] ok, so it's the regular image creation? [10:13] * didrocks will stop trusting asac when he uses the world "magic" :p [10:13] word [10:14] didrocks, it is the same as iso creation, we have a few sripts that run additionally, but nothing touches usermetrics [10:14] or anything that could cause such an issue [10:14] debian/usermetricsservice.preinst: chmod 0750 /var/lib/usermetrics [10:14] adduser --system --ingroup usermetrics --home /var/lib/usermetrics usermetrics [10:14] chown -R usermetrics:usermetrics /var/lib/usermetrics [10:15] hum… [10:15] root@ubuntu-phablet:/# grep usermet /etc/passwd [10:15] usermetrics:x:102:104:User Metrics:/var/lib/usermetrics:/bin/false [10:15] all fine [10:15] root@ubuntu-phablet:/# mount|grep usermetrics [10:15] /dev/mmcblk0p12 on /var/lib/usermetrics type ext4 (rw,relatime,barrier=1,data=ordered) [10:15] i guess the issue lies somewhere in the mounting [10:15] But /var/lib/usermetrics is in writable-paths, so something needs to apply that change [10:15] right [10:15] There's a scheme for post-image-upgrade hooks now [10:15] yep [10:16] but the bind mount should just inherit the permissions [10:16] you need additional steps for paths that are in writable paths? [10:16] you shouldnt [10:16] i guess we need to wait for stgraber [10:16] didrocks: Yes [10:17] I think [10:17] you really shouldnt, it should just inherit the target settings ... at least thats what stgraber explained to me once [10:17] A persistent mount has its backing store in userdata, and inherits permissions from there, not from the version in the image [10:17] Surely [10:17] That's kind of the point of being persistent [10:18] hrm [10:18] This is why we have to be careful about adding persistent mounts [10:18] well, it worked three images ago [10:18] with the same mount [10:19] New persistent paths get copied to userdata when they're first seen; after that, something needs to sync changes manually if needed [10:19] (read: yesterday, we had a lot of image builds over night) [10:20] this mount exists since weeks and used to work before [10:21] Morning all [10:22] ogra_: fixed systemd uploaded, tested with relative symlinks now [10:23] ogra_: cjwatson: so, we have to wait for stgraber to know the detail on what can possibly have changed? [10:24] * ogra_ looks at the change to the initrd script at http://launchpadlibrarian.net/151908224/initramfs-tools-ubuntu-touch_0.49_0.50.diff.gz ... but i dont see how that would have changed the carryover of permissions [10:24] which was added in http://launchpadlibrarian.net/149606706/initramfs-tools-ubuntu-touch_0.48_0.49.diff.gz [10:26] didrocks, i fear we have to, yeah [10:26] ogra_: I looked at lxc-android-config, doesn't seem as well related… [10:26] i see that he added code to carry over permissions and ownership on sep. 7th [10:26] and apparently the last changes broke it, but i dont see how by looking at the code [10:27] pete-woods: if you can have the non crash code, at least, we have a way to go forward until stgraber is back (and you are more robust anyway) [10:27] didrocks, lxc-android-config only carries the config [10:27] didrocks: I'm going to ask for that to be landed today [10:27] yeah, I was looking if something was dropped [10:27] initramfs-tools-ubuntu-touch creates the acrual mounts based on the config [10:27] pete-woods: yeah, just tell us when it's ready, we'll process it quickly [10:28] didrocks: I guess the tests will still fail (given the RO /var/lib/...) but at least the apps won't crash [10:30] pete-woods: yeah, it's already a start for people dogfooding ;) [10:30] * didrocks still stays positive and is somewhat happy we know what are making all those tests regressions === alan_g is now known as alan_g|afk === hikiko is now known as hikiko|lunch === oreneeshy_ is now known as oreneeshy [10:36] ogra_: mtp has a couple of issues now. 1 lots of boxes if you reboot the device, 2. if you adb shell before the popup nautilus window opens for mtp it seems to kick you out of the adb session, nice :) === alan_g|afk is now known as alan_g [10:38] davmor2, hmm, i dont have that here [10:39] oh [10:39] in fact [10:39] oot@ubuntu-phablet:/# ps ax|grep mtp [10:39] 67 ? S< 0:00 [f_mtp] [10:39] 2090 pts/3 S+ 0:00 grep --color=auto mtp [10:40] ogra_: although todays image seems to of lost the lots of boxes by the look of it so it was 70 played up 74 seems okay I'll try the getting kicked out of adb again in a second [10:40] well, the above is on 74 [10:40] mtp doesnt seem to start at all [10:42] ogra_: it starts here [10:42] didrocks, i'm on maguro [10:42] phablet 2991 0.1 0.0 3300 1180 ? Ss 10:42 0:00 /usr/bin/mtp-server [10:42] davmor2 too [10:42] interesting [10:42] cyphermox: I guess this will be for you ^ [10:43] ogra_: mtp is showing up here [10:43] didrocks, i dont think it will [10:43] didrocks, i guess thats another one for stgraber [10:43] ogra_: do you think it's a configuration issue? [10:43] (weird that mako isn't impacted by that one though) [10:43] he changed the lxc-android-config job ... which emits the android event ... which is the trigger event for mtp [10:43] ogra_: as in the nautilus window is opening say Galaxy Nexus [10:44] davmor2, right, doesnt happen on my install here [10:44] and i flashed twice already [10:44] oh right [10:45] * ogra_ switches on some debugging [10:46] davmor2@boromir:~$ adb shell ps aux | grep mtp [10:46] root 67 0.0 0.0 0 0 ? S< 10:38 0:00 [f_mtp] [10:46] phablet 1676 0.0 0.1 3304 1360 ? Ss 10:39 0:00 /usr/bin/mtp-se [10:46] davmor2, flashed with --no-backup ? [10:46] ogra_: always [10:46] k [10:47] root@ubuntu-phablet:/# getprop sys.usb.config [10:47] mtp,adb [10:47] root@ubuntu-phablet:/# ps ax|grep mtp-server|grep -v grep [10:47] root@ubuntu-phablet:/# [10:47] :( [10:48] root@ubuntu-phablet:/# cat /var/log/upstart/lxc-android-config.log [10:48] 1277 blocks [10:48] cp: cannot stat '/var/lib/lxc/android/overrides/*': No such file or directory [10:48] aha [10:50] ah that line has an || true [10:50] hmm [10:51] root@ubuntu-phablet:/# cat /var/log/upstart/mtp-server-bootup.log [10:51] could not set property [10:51] could not set property [10:51] ogra_: Critical bug in 74 no sound notification for an incoming call [10:51] aha [10:52] davmor2, 74 is dead beef already .... and i think thats fallout of the usermetrics thing too [10:52] hi all, I got my phone in a state OTA updates did not install, so I thought I'd redo a phablet-flash ubuntu-system. That failed mid-through and I was left in the recovery screen. Does anyone know what the easiest way to reflash with ubuntu-system is? The instructions at https://wiki.ubuntu.com/Touch/Install#Manual_Download_.26_Installation seem to apply for the RW images only? [10:52] (dead beef as in completely broken) [10:52] ogra_: do you know when 75 will land then? [10:52] davmor2, after we have fixes for all the mess [10:55] aha, manually setting the property makes it work [10:55] lets see if that persists [10:55] ogra_: Man you must of known for more than a minute and it isn't fixed yet? :D [10:56] davmor2, well i still dont have an idea why it broke [10:56] and in fact it doesnt persist over a reboot if i set the property [10:56] ogra_: so what else is the usermetrics likely to of effected badly then? [10:56] camera [10:56] probably other stuff [10:57] ogra_: yeap camera is dead, I'm assuming sms [10:57] aha [10:57] root@ubuntu-phablet:/# getprop sys.usb.config [10:57] mtp,adb [10:57] root@ubuntu-phablet:/# setprop sys.usb.config mtp,adb [10:57] ogra@anubis:~/Devel/packages$ adb shell [10:57] its the prder ! [10:57] *order [10:58] why the heck is that backwards here and not for you then ? [10:58] hmm, no [10:58] davmor2@boromir:~$ adb shell getprop sys.usb.config [10:58] mtp,adb [10:59] yeah, red herring [10:59] ogra_: ^ [11:00] ogra_: however that might explain why adb gets kicked once mtp finishes loading in nautilus maybe? [11:00] something isnt right, yeah [11:03] ogra_: so for me I get the mtp icon in the launcher, then welcome screen on the Phone, I do adb shell as soon as the welcome screen appears, then wait for the Nautilus window then this happens in the terminal where adb is running [11:03] davmor2@boromir:~$ adb shell [11:03] root@ubuntu-phablet:/# davmor2@boromir:~$ [11:03] yeah, as i said, something is wrong here [11:04] ogra_: I'll write up a bug for that then :) [11:04] assign to me [11:04] ogra_: I need to figure out what to file it against yet :) [11:05] either mtp-server/mtp or android-tools-adbd/android [11:05] pick one, they fight over the property it seems [11:05] or even better file against one, add the other [11:06] oh, that would be android-tools-adbd/android-tools [11:06] not just android [11:07] my Listview objects is overlaping with the previous item while scrolling , anyidea ?? [11:07] ogra_: thanks, will do === MacSlow is now known as MacSlow|lunch [11:18] sil2100, ping [11:18] gatox: pong [11:19] sil2100, hi... i've been told that the click-update-manager app still has no icon, i've added one, and configure the .desktop file, but it seems it needs to be added somewhere in the cmake file.... do you know sometihng about that? [11:20] gatox: hm, I have a branch ready for click-update-manager related to AP tests packaging, I'll check what's up and try to fix that in that branch, ok? I just got my internet back so I need to backtrack things [11:21] ogra_: https://bugs.launchpad.net/mtp/+bug/1233613 mtp won't let me assign you apparently LP tells me you are no such item, but worked fine with adb [11:21] Ubuntu bug 1233613 in android-tools (Ubuntu) "Adb and mtp priority issue" [Undecided,New] [11:22] ok [11:22] didrocks, i thinnk i found the usermetrics issue but will need stgraber to confirm === chihchun is now known as chihchun_afk [11:23] ogra_: oh? [11:23] /var/lib/usermetrics auto persistent transition none [11:23] in /etc/system-image/writable-paths [11:24] that will carry over the permissions if the target file doesnt exist [11:24] (i.e. on first boot) [11:24] ogra_: I guess the question is what changed? It didn't though? [11:24] (for usermetrics) [11:25] didrocks, right, must be caused by the initrd script changes that it doesnt get the owner and permissions carried over [11:25] yeah, possible [11:25] sil2100, ok, thanks, let me know [11:25] anyway, i have a working camera again [11:26] didrocks, ogra_: I blame the internet, somewhere along the lines it's bound to be the internets fault [11:26] heh ;) [11:26] after changing the file and removing /userdata/system-data/var/lib/usermetrics [11:26] ogra_: ah nice! let's hope stgraber will be around soon [11:26] oot@ubuntu-phablet:/# mount|grep usermetrics [11:26] /dev/mmcblk0p12 on /var/lib/usermetrics type ext4 (rw,relatime,barrier=1,data=ordered) [11:26] root@ubuntu-phablet:/# ls -ld /var/lib/usermetrics/ [11:26] drwxr-x--- 2 usermetrics usermetrics 4096 Oct 1 11:24 /var/lib/usermetrics/ [11:27] didrocks: just keep pinging stgraber till he is :D [11:27] hah [11:27] * ogra_ goes afk for a moment [11:29] sergiusens, ping [11:30] mandel, pong [11:30] sergiusens, hola! one small question, is the installlation via qt creator of build-deps working on the nexus 7? [11:31] sergiusens, I'm getting that /data/ubuntu does not exists.. and I'm tempted to just create it === hikiko|lunch is now known as hikiko [11:32] you have a working rom and you don't see /data/ubuntu? [11:32] OrokuSaki, yes, is working and I get the following: [13:29:35] chroot: cannot change root directory to /data/ubuntu: No such file or directory [11:33] strange [11:33] mandel, hm, /data/ubuntu shouldn't be used anymore [11:33] OrokuSaki, I ssh in the device and doing a sudo ls /data I can confirm that I have no ubuntu dir [11:33] sergiusens, that is what qt creator tools complain about :-/ [11:33] @sergiusens... ahh man.. guess I gotta update again.. was this recent in the last 7days or so? [11:34] mandel, it's a Qtreator bug, it should detect it's running a system image [11:34] OrokuSaki, depends on what you use, cdimage type images do use /data/ubuntu, system images don't [11:34] sergiusens, bummer... and sudo apt-get build-dep is returning => E: The package lists or status file could not be parsed or opened. [11:35] mandel, touch /userdata/.writable_image ? [11:37] sergiusens, http://paste.ubuntu.com/6179218/ [11:38] sergiusens, is an image from the devel channel [11:40] mandel: did you reboot after adding the setting [11:41] mandel: for both /userdata/.writable_image and /home/phablet/.display-mir you need to reboot the device to make the change take effect [11:42] was my next question, did you reboot? [11:42] davmor2, ah! had no idea [11:44] sergiusens, davmor2 that was the issue, thx! [11:44] mandel: no worries dude :) [11:45] can i install "ubuntu for tablets" on my windows 8 tablet "acer iconia w511"? [11:45] and is "ubuntu for tablets" released? === fmasi is now known as fmasi_afk [11:48] patchyy: Probably not as the uefi on windows 8 tablets is locked on. [11:48] patchyy: As for the Ubuntu Touch release that happens in October [11:49] can i run it in virtualbox x86 then, maybe? [11:50] patchyy: no easily, You can install ubuntu 13.10 and then install unity8 to get an idea on how is will feel. [11:51] aha..hmm [11:51] ubuntu should make it easy to install on windows tablets, as the tablets have a pretty good hardware, and is sold very cheaply since nobody wants windows tablets :P heheh [11:53] somebody has started with porting to acer iconia a700 -> https://wiki.ubuntu.com/Touch/Devices/a700 [11:55] patchyy: Yes, that is android based though [11:56] ahh, ok [11:56] and then its not using uefi? [12:01] ogra_: ext4 on MTD is not happy, so I will be using sdcard device for userdata / ubuntu rootfs. [12:01] sil2100, it seems there is a branch proposed for the icon thing already: https://code.launchpad.net/~mfisch/click-update-manager/fix-icon/+merge/188496 [12:01] stgraber: heya [12:02] stgraber: we have an owner/group issue with var/lib/usermetrics; it's 102:104 in the cdimage tarball, but it's root:root under /userdata/system-data on a phablet-flashed image [12:02] stgraber: that's https://bugs.launchpad.net/touch-preview-images/+bug/1233396 [12:02] Ubuntu bug 1233396 in The Sevilerow project "infographics and camera-app broken in build 71/72 - core dumped" [Critical,Confirmed] [12:02] pete-woods: hey [12:03] pete-woods: did you see that libusermetrics was FTBFSing again despite the valgrind suppressions? [12:03] pete-woods: oh actually it seems you fixed this already, nevermind :-) [12:03] lool: hi, where can I see these errors first hand? I only know where to get them from the bug reports people file [12:04] pete-woods: it was from a build-record in the ~ubuntu-unity/daily-build PPA [12:05] pete-woods: https://launchpad.net/~ubuntu-unity/+archive/daily-build/+packages?field.name_filter=libusermetrics&field.status_filter=superseded&field.series_filter=saucy [12:05] pete-woods: second package listed there [12:05] pete-woods: i386 build log: https://launchpadlibrarian.net/151943506/buildlog_ubuntu-saucy-i386.libusermetrics_1.1.1%2B13.10.20130930.1-0ubuntu1_FAILEDTOBUILD.txt.gz [12:06] pete-woods: problem is that the snapshot was taken before your fix landed in bzr [12:06] AFAICT [12:06] lool: thanks for that info, been meaning to find that URL for ages [12:07] pete-woods: but you should get the same issue if building for i386 surely? [12:07] or I guess if doing a full build with build-indep step [12:07] lool: for some reason it seems like my jenkins jobs aren't configured to build i386 [12:09] pete-woods: are these the upstream merger ones? [12:09] lool: yes [12:10] pete-woods: TBH I don't know what configs are available there, you could chat with Francis to find out whether he could build the build-indep step, and/or build on i386 [12:10] pete-woods: actually, it's not clear to me why this is i386 specific; the build-deps and rules are identical; no build-indep step, no build-depends-indep [12:11] pete-woods: so maybe there's truly an i386 failure here (in Qt I guess) [12:13] is ubuntu-touch a distinct entity to "ubuntu for phones" or is there enough overlap that the two terms can be used interchangeably? I imagine ubuntu-touch pertains to the interface specifically [12:13] robjh: we prefer Ubuntu for phones [12:13] albeit we still have "touch" in a bunch of places [12:13] lool; hence the channel name? [12:14] lool: the Qt symbol is just different in i386 vs x86_64 _nwm vs nwj [12:14] * pete-woods doesn't have enough kung-fu to know hy [12:16] robjh: hence the channel name :-) [12:17] robjh: but the helpfully named ubuntu-phone@ list! === jhodapp|afk is now known as jhodapp [12:17] pete-woods: echo that-symbol | c++filt, most likely it's the same just 32 vs 64 size difference. [12:17] pete-woods: _Znwm -> operator new(unsigned long) [12:17] pete-woods: _Znwj -> operator new(unsigned int) [12:17] xnox: yeah exactly [12:18] look: okay, thank you :) [12:18] * lool was fishing the actual names from the suppression file [12:18] lool: well we had #ubuntu-mobile way back when, it at the moment redirects here. [12:18] xnox: these were the days! [12:18] lool: i like c++filt a lot, it demangles on the fly any output one pipes into it, without corrupting anything else. E.g. including like build-logs. [12:19] xnox: do you know whether valgrind suppression files would work with real symbol names? [12:19] sorry I mean with human readable ones [12:20] lool: i don't know, but e.g. dpkg-symbols support filters thus one can have human names in the .symbols file with the filter prefix, and it then mangles them first before comparing. [12:20] you don't need to echo it separately btw [12:20] oddly enough there does not seem to be a mangling tool [12:21] xnox: .symbols do support the demangled names I believe [12:21] lool: i guess you could maintain human readable and then use c++filt to mangle them.... or maybe it demangles & compares. [12:21] yeah [12:22] Laney: i default to pipe syntax instead of args whenever I can, cause i often build longer pipes or chop them up. And i hate have to go back and move stuff around ;-) [12:22] hello ubuntu touch world! Could someone tell me if its possible to use the ubuntu modules on Qt5.1.1. This is for my own desktop apps only, I just want to use the ubutnu touch styling for some uni projects I am working on [12:28] lool, switching the usermetrics entry in /etc/system-image/writable-paths to "transition" and deleting /userdata/system-data/var/lib/usermetrics gets it working for me on next boot [12:29] so the ubuntu phone is coming or what? [12:30] thief_and_a_liar: our software will be released on 17th October [12:31] ok, but is it interesting for anyone, so there will be phones out of this? [12:32] thief_and_a_liar: maybe [12:35] ogra_: it does very much seem like the correct fix [12:35] ogra_: however let's give 30mn for stgraber to show up and confirm [12:35] yeah [12:36] thanks for digging into this, keps being context switched to something else while tracing this :-) [12:36] sadly we can only really confirm the fix with a fresh image build [12:36] as the first boot is essential here [12:37] lool, the strange point is that there was no change [12:37] it worked before on image 70 [12:37] (to that specific line i mean... and i dont get why it worked without having "transition" set) [12:40] ogra_: that's also why I find all this weird [12:40] we also have a weird issue where url-dispatcher gets connection refused on dbus [12:40] which I find scary [12:40] yeah [12:40] I'm just realizing [12:40] could this be apparmor [12:40] for dbus ? [12:41] yeah [12:41] didrocks: ^ [12:41] you should see a lot of spam in the logs if thats the case [12:41] jarope, yes it works, use the packages from the canonical-qt5-edgers PPA === chihchun_afk is now known as chihchun [12:42] if you think it is apparmor, do 'grep DEB /var/log/syslog' [12:42] err [12:42] 'gerp DEN /var/log/syslog' [12:42] heh [12:42] man [12:42] ogra_: I wonder whether some app tries to open an URL with liburl-dispatcher, but doesn't have apparmor permission to do so and gets caught doing that [12:42] I can't type [12:43] 'grep DEN /var/log/syslog' [12:43] pmcgowan: many thanks will do that [12:43] jdstrand, i always type apt-shell :) [12:43] just add repo and updaqte /upgrade? [12:43] heh [12:43] jdstrand: don't have syslog sadly [12:43] lool, right, but you should see that relatively clear in the logs ... bad if you dont have them indeed [12:44] well, url-dispatcher is out of process. it should all happen underneatch via dbus. app tries to open url -> dbus -> url dispatcher [12:44] and we allow calls to url dispatcher in the ubuntu* templates [12:44] do we ? [12:44] note that polkit is now active [12:45] lool, might be a polkit thing [12:45] jdstrand: oh actualy we do have dmesg [12:45] and there's an unrelated DENIED there [12:45] [ 21.966674] type=1400 audit(1380573832.836:55): apparmor="DENIED" operation="open" parent=896 profile="/sbin/dhclient" name="/etc/writable/localtime" pid=1400 comm="dhclient" requested_mask="r" denied_mask="r" fsuid=0 ouid=0 [12:45] yeah. pitti added a fix for that already [12:45] yes, that's in saucy, just needs a new image [12:45] plus the apps are unconfined, so apparmor is probably a red herring [12:45] it is highly likely that polkit is getting in the way [12:45] lool, i'd liik at polkit [12:46] *look [12:46] jarope, add-apt-repository; apt-get update; then install the ubuntu-sdk [12:46] note, that apps will see AppArmor in the dbus denial too. there are other security denials for dbus-- like polkit [12:47] ogra_: Hmm I see no dbus policy for it === chihchun is now known as chihchun_afk [12:47] lool, well, the question is "do we need one" :) [12:47] pmcgowan: arr ok so should I purge the sdk first to remove current version? [12:47] i guess [12:48] ogra_: but then everything talking it to it would fail [12:48] also, it's actually url-dispatcher failing to connect to dbus, not some app failing to connect to url-dispatcher [12:48] usually you should be able to if you have a seat assigned by logind [12:48] but that doesnt work for i.e. NM [12:48] it still needs a pkla file [12:48] jarope, I think it will just upgrade [12:49] ok thanks [12:49] is there a trick to disable the screen timeout? [12:49] it's unnerving to have to unlock the screen every minute for running autopilot tests [12:50] lool, for a test, copy /etc/polkit-1/localauthority/50-local.d/org.freedesktop.NetworkManager.pkla and create a pkla for your specific dbus call, see if it starts working [12:51] ogra_: I don't know how to trigger the issue though === fmasi_afk is now known as fmasi [12:51] pmcgowan: on launch pad it says that ppa is obsolete? ? ? ? / [12:51] pitti, powerd-cli display on [12:52] pitti, you will waste a shell for this though :) [12:52] jarope, there are I think 5 ppas there, I think its beta2, Mirv would know [12:52] Mirv, which ppa has latest 5.1.1 work? [12:52] ogra_: ah, thanks; & FTW :) [12:52] thanks pmcgowan [12:53] ogra_: we so much need to install byobu or at least tmux on that thing during development :) [12:53] heh [12:53] pitti, convince pmcgowan ... he whines because of wasted space all the time :) [12:53] j/k [12:54] (and yes, we install a ton of stuff there which we don't need in production) [12:54] ogra_, I do not whine [12:54] pmcgowan: do -- lean == good! [12:54] hah ;) [12:54] :) [12:56] +1 for screen / tmux / byobu! [12:57] the terminal app should just ship that in the click package ;) [12:59] pmcgowan: it does indeed look like 2 has the current 5.1.1 modules [12:59] many thanks for your help [13:00] jarope, ah good [13:01] now to make all this years uni assignments touchified! [13:01] barry: joining -meeting? :-) [13:01] lool: morning [13:01] stgraber: \o/ [13:01] on that note, I still don't understand why we ship the whole python and autopilot stack [13:01] stgraber, ! [13:01] stgraber: want to backlog or prefer a short summary? [13:01] one pretty much always needs to install additional packages anyway in order to run autopilot tests [13:01] pitti, I just noticed that yesterday myself [13:02] didrocks: I read the highlights but a summary would spare me the rest of the backlog [13:02] not sure why [13:02] at least teh ubuntu ui toolkit stuff [13:02] ask QA [13:02] so we might just as well install everything [13:02] stgraber: ok, so we have a lot of AP regressions [13:02] the toolkit runtime stuff is all installed [13:02] stgraber: the consequence is because on a fresh install (with --no-backup), some apps don't start [13:02] stgraber: libusermetrics is segfaultin [13:02] because /var/lib/usermetrics is 700, owned by root [13:03] instead beeing own by the usermetrics user [13:03] pitti, ogra_ yes and no [13:03] the postinst of the libusermetrics package is correct [13:03] stgraber, seems this should be "transition" in /etc/system-image/writable-paths [13:03] larsu, please don't hate me. https://wiki.ubuntu.com/StatusBar?action=diff&rev2=11&rev1=10 [13:03] but it seems something changed in the RO image [13:03] stgraber, which it isnt ... [13:03] in the way the mount is done [13:03] pitti, ogra_ yes today, no when we don't trigger rw anymore [13:04] stgraber, talking about the /var/lib/usermetrics entry in there [13:05] so if we can understand what changed and what is the correct fix, it will rock :) [13:05] and apply it :) [13:05] ogra_: I was about missing that one! thanks :p [13:06] :) [13:06] ogra_: you don't need transition for that, userdata is an empty path, there's nothing to transition and the code does a stat + chown/chmod specifically to avoid that problem [13:06] so something else is wrong [13:06] stgraber, well, it doesnt since image 71 [13:06] mpt: I could never hate you! [13:06] awwww [13:06] mpt: WHAT HAVE YOU DONE!!!!eleven!! [13:06] stgraber: http://people.canonical.com/~ogra/touch-image-stats/20130930.changes [13:07] do you see anything potentially creating this there? [13:07] mpt: (just joking, of course) [13:07] initramfs-tools-ubuntu-touch (which i stared at dfor ages now and cant find an issue) [13:08] yeah, I re-read the diff a dozen time myself just now and can't find anything obviously wrong. I was also running with the new scripts/touch for a couple of days last week before it was uploaded so I'm surprised I didn't see anything then... [13:09] so - fully appreciating that people are busy - is there anyone about who would be willing to start being a mentor for some one who is keen but still green? I would like to have a go at building a "proper" touch app - a touch irc client based. I used Qt to build an android client last year so dont see why it couldnt be tweaked to work [13:09] so it's going to be easier if I have a broken device, let me update mine real quick [13:10] stgraber, better re-flash [13:10] with --no-backup [13:10] its definitely related to first boot [13:10] root@ubuntu-phablet:/# ls -ld /var/lib/usermetrics/ [13:10] drwxr-x--- 2 usermetrics usermetrics 4096 Oct 1 13:10 /var/lib/usermetrics [13:11] stgraber, so unmounting and deleting the file in /userdata gets me a properly created one after reboot [13:11] * ogra_ tries the same on a feshly flashed install [13:11] ogra_: ok, reproduced (there are much faster ways to reset a device than to reflash ;)) [13:12] stgraber, well, but they dont get you the test env we use [13:12] we should all be on the same page when testing stuff [13:12] ogra_: removing everything in /userdata but system.img certainly will [13:12] (so please promote your faster way so we all can ues it ;)) [13:13] ok, so I've got a few ideas of what may be happening but I need to add some debugging to my initrd to figure it out [13:13] we should have a script for that while we lack a UI tool for reset [13:13] ogra_: that + turn on the write mode === _salem is now known as salem_ [13:14] didrocks, touch /userdata/.writable_image is to much ? [13:15] ogra_: I always have to find the name of the file [13:15] (note with the final release you will even have toi enable adb from the terminal first [13:15] ) [13:15] doesn't map my mental model [13:15] heh [13:15] ogra_: not sure the management thinks the same, but let's see ;) [13:15] I'm sure it's the plan for stgraber to get up his blog activity up ;) [13:15] ? [13:15] mgmt thinks the same about what ? [13:16] ogra_: about adb off by default being a priority [13:16] i have an order to care for it [13:16] I guess we'll try first to land what we try (the 4 big items) [13:16] then, we'll see :) [13:16] it has to be off in what we release as final image [13:16] ogra_: ok, I think I found the problem and it's what I suspected, just trying to figure out how that happened [13:17] (which is worst case a trivial .override file) [13:17] ogra_: I propose that we implement the konami code to enable developer mode [13:17] ogra_: sorry, really no idea how that happened... [13:17] capcom* [13:17] where was my mind? :) [13:17] ogra_: so the problem is that ./hooks/touch was ignored somehow [13:18] ogra_: if you look inside the initrd binary, we're missing all the binaries it's supposed to copy [13:18] ogra_: including /bin/chown (that was my guess) [13:18] ogra_: but also adbd, ... [13:18] wow [13:19] now the problem is that I have absolutely no clue what could have caused that... [13:19] yeah, teh log agrees https://launchpadlibrarian.net/151908853/buildlog_ubuntu-saucy-armhf.initramfs-tools-ubuntu-touch_0.50_UPLOADING.txt.gz [13:20] /usr/share/initramfs-tools/hooks/touch ignored: not executable [13:21] it certainly is executable in my branch here, so it's a problem on the system of the uploader... [13:22] which would be mine ... [13:22] * ogra_ checks [13:24] ogra@anubis:~/Devel/packages/initramfs-tools-ubuntu-touch-0.49$ head -1 debian/changelog [13:24] initramfs-tools-ubuntu-touch (0.50) saucy; urgency=low [13:24] ogra@anubis:~/Devel/packages/initramfs-tools-ubuntu-touch-0.49$ ls -l hooks/ [13:24] insgesamt 4 [13:24] -rwxrwxr-x 1 ogra ogra 593 Sep 6 18:45 touch [13:24] all executable [13:25] pmcgowan: jarope: ppa:canonical-qt5-edgers/qt5-beta-proper since it has the autorebuilding of unity8 & co. [13:25] * xnox had fixperm "helpfully" strip -x [13:25] ogra_: ^ [13:25] oo ok Mirv thanks === MacSlow|lunch is now known as MacSlow [13:26] will go swap out the beta2 [13:26] xnox, hmm, let me check the binary [13:27] jarope: updated the PPA descriptions a bit too, for those who read them :) [13:28] it happens [13:28] hi all [13:28] and who reads these days anyway [13:28] ogra@anubis:~/Devel/packages/tmp$ dpkg -x initramfs-tools-ubuntu-touch_0.50_armhf.deb . [13:28] ogra@anubis:~/Devel/packages/tmp$ ls -l usr/share/initramfs-tools/hooks/ [13:28] insgesamt 4 [13:28] -rwxr-xr-x 1 ogra ogra 593 Sep 6 18:45 touch [13:28] hmm [13:28] all fine it seems [13:29] i think yoda said it best Do or Do not, there is no read [13:30] we also have:ERROR: could not open directory /tmp/mkinitramfs_yCk0sK/lib/modules/3.2.0-51-highbank: No such file or directory [13:30] FATAL: could not search modules: No such file or directory [13:30] yeah [13:30] thats fine [13:31] would have to compare with a previous buildlog to see exactly what's normal and what isn't... [13:31] tehse are normal moaning from update-initramfs [13:32] i wonder if the new fakeroot might have something to do with it [13:32] https://launchpad.net/ubuntu/saucy/+source/fakeroot/1.20-1 [13:32] (the build uses fakeroot and fakechroot to roll the initrd) [13:33] (i dont see how reading the changelog) [13:33] not in an obvious way at least... [13:33] right [13:33] the timing would fit though [13:34] ogra_: the tarball uploaded 0.50 has executable bits removed. [13:34] I'm running a rebuild here to see if it can be reproduced at least [13:34] xnox, the binary package has them though [13:35] i could force them from the build script easily, but that doesnt explain why it just started to happen out of nowhere [13:36] well, let's first see if that still happens, if it does then we can try to understand what's causing it [13:37] right, we need to get the images fixed first anyway [13:39] the good news is that we didn't introduce any new non-root path to the image with that update, so people who updated won't have to fix their devices manually. Only those who reflashed from scratch will have seen the problem and that's likely mostly QA who reflash from scratch every time anyway. === dwatkins_ is now known as dwatkins [13:40] well, one hopes that on release day people will do a fresh bootstrap [13:40] we should suggest that in the release announcement :) [13:40] balloons, hey, why did you set https://bugs.launchpad.net/ubuntu/+source/content-hub/+bug/1231368 as confirmed for ubuntu-system-settings? [13:40] that'd be nice but I wouldn't count on that ;) [13:40] Ubuntu bug 1231368 in ubuntu-system-settings (Ubuntu) "Can't select a background anymore since the 2013-09-25 update" [High,Confirmed] [13:41] ricmm, any luck with the unity8 max cpu thing [13:41] seb128, because its all your fault ! [13:41] seb128, sorry about that! launchpad was acting really funny and slow. I looked at it but didn't mean to mark anything. I was confirming other bugs [13:42] balloons, ok, I'm closing it then, the issue was not in the settings and has been fixed with the content-hub and gallery updates [13:42] :-) [13:43] Adding binary /bin/chown [13:43] Adding binary /bin/readlink [13:43] Adding binary /usr/bin/adbd [13:43] ogra_: ^ [13:43] phew === dandrader is now known as dandrader|afk [13:44] so we blame cosmic rays ? [13:44] that's with a rebuild here, so I'd recommend we just no-change rebuild it and see if we get the same on the buildd [13:44] yeah :( [13:44] to bad we need an android rebuild for that [13:44] +1 on cosmic rays [13:44] so we won't even ever know why the initramfs was busted [13:45] didrocks, hey, need a pkg help [13:46] ogra_: uploaded [13:46] thx [13:46] thanks stgraber ;) [13:46] didrocks, need a papertrail or do we just let it in [13:46] now let's hope we'll get the same result or it's going to be a major pain to debug! [13:46] yeah [13:46] ogra_: we already have one [13:46] ah, k [13:46] 71 [13:46] didrocks, we need to make sure that if you install libunity-protocol > 7.1.1, you can't have libunity < 7.1.1, but there's no dep from protocol to libunity, cause libunity deps on protocol [13:47] (because my test board here is as close as you can get to a calxeda node and should be using the same build chroot as the LP buildds, so I sure hope I'll get the same result ;)) [13:47] * ogra_ has spent the last hours staring at logs and code ... kind of missed watching the spreadsheet [13:47] didrocks, is conflicts for protocol enough, or do we need conflicts+replace+??? [13:47] didrocks, ogra_, stgraber: Don't blame cosmic rays, blame comic sans it's far more dangerous [13:47] davmor2, you sense a microsoft conspiracy ? [13:48] mhr3: breaks: is the correct relationship [13:48] didrocks, just breaks? [13:48] so libunity-protocol Breaks: libunity (<< 7.1.1) [13:50] ogra_: :D No I think it's far more damaging than a bit of sunburn :) [13:50] heh [13:50] * ogra_ twiddles thumbs wildly reloading https://launchpad.net/ubuntu/+source/initramfs-tools-ubuntu-touch/0.51/+build/5066577 === Namidairo is now known as Namidairo`bnc [13:51] same here ;) [13:52] I also bumped the i386 and amd64 one to try and get things ready ASAP since we seem to have quite busy buildds today (there was a 4 hours wait for i386 and amd64) [13:52] ugh [13:52] we should drop these obsolete arches [13:53] their buildds just cant cope [13:53] :P [13:53] didrocks, could you peek at https://code.launchpad.net/~mhr3/libunity/same-ver-dep/+merge/188596 ? [13:54] ogra_: not seeing the error this time around [13:54] stgraber: I thought you may be interested in this script that jcollado has put together: http://bazaar.launchpad.net/~javier.collado/+junk/upgrade-tests/view/head:/scripts/full-update.py [13:54] it srolled by to fast for me [13:54] mhr3: are you sure you want a version << 7.1.1? You are at 7.1.2+13.10.20131001-0ubuntu1 [13:54] (just double checking) [13:55] didrocks, yea, good enough [13:55] ogra_: I just saw the beginning of the initramfs call and didn't see anything related to hooks/touch, I didn't see the copy though (went by too fast) [13:55] didrocks, it's cause schema was changed in 7.1.1 [13:55] same here [13:55] * ogra_ waits for the log to publish [13:55] Adding binary /bin/chown [13:55] Adding binary /bin/readlink [13:55] Adding binary /usr/bin/adbd [13:55] \o/ [13:55] yay! [13:56] k, so now waiting for android ... [13:56] ok, so android no change rebuild now (that one is building with -proposed enabled right?) [13:56] not sure [13:56] xnox, created it ... i think he made it so [13:56] ok, let's see [13:56] heh [13:57] note that android is cross built ... will queue up on the x86 builder [13:57] I see $dist-proposed in debian/rules so looks like we don't need to wait [13:57] right [13:57] stgraber: yes, pulls from -proposed. But it needs to be well, published in -proposed. [13:58] on ftpmaster.internal [13:58] xnox: right, which it should be soonish. I'll prepare the upload and push when rmadison tells me it's fine [13:58] mhr3: packaging side approved [14:00] ok so looks like with some fiddling about - i.e. remove the install of Qt5.1.1 I had - ubuntu-sdk is up to 5.1.1 . Question is now have certain QtQuick Modules been removed and replaced by Ubuntu.components? QtQuick.Controls and QtQuick.Layouts are not recognised. Is Ubuntu.Components replacing them two? [14:00] didrocks, thx, just to make sure, now if protocol + libunity 7.0.0 is installed and a package will want to upgrade protocol, libunity will be upgraded as well, right? === dandrader|afk is now known as dandrader [14:01] mhr3: right [14:01] didrocks, awesome, thx [14:03] come on LP, publish faster... [14:04] stgraber: you don't have that cycle to help it? ;) [14:05] ogra_: once mir, unity-mir and the initramfs are in, you will rekick an image? [14:05] :) [14:05] didrocks, with pleasure ! [14:05] i want to see the media stack working [14:05] ... finally [14:05] same here! [14:06] (and timezone configuration too ... ) [14:06] stgraber, do you think this will also fix the /etc/writable issues ? [14:07] ogra_: what's the issue with /etc/writable? [14:07] stgraber, its empty on boot [14:08] i.e. no files are copied when the loop mount appears [14:08] (unmounting it makes the files visible underneath) [14:08] weird, let me check, hopefully it's the same issue [14:09] root@ubuntu-phablet:/# ls /etc/writable/ [14:09] root@ubuntu-phablet:/# umount /etc/writable [14:09] root@ubuntu-phablet:/# ls /etc/writable/ [14:09] localtime timezone [14:09] thats my first login after a fresh flash [14:09] actually, let me just grab the new initrd, flash it and see what happens, will be faster [14:10] heh [14:10] * ogra_ gest fresh coffee [14:10] what is doing icon lookup for apps in the dash in unity8? is it the dash or apps scope? [14:13] seb128: hey. i was told you're the one to ask about how to open the accounts settings panel from within a scope. i've got it opening system-settings, but it's not switching to the accounts panel. i've tried many variations on ${plugin_name} in settings:///system/${plugin_name} to no avail [14:14] stgraber: do we need to rebuild android? [14:14] lool: yes, eventually [14:14] dobey, hey, is the name of the .settings in /usr/share/ubuntu/settings/system/ [14:15] ogra_: seems to be a different problem [14:15] :( [14:15] lool: though I'd rather not rebuild android now and have to do it again in 30min when I figure out what's breaking /etc/writable [14:15] dobey, and the code to handle settings:/// url landed in trunk for ubuntu-system-settings but not in saucy/touch yet [14:15] dobey, so if you try, use trunk [14:15] stgraber, the whole process is stalled atm [14:15] stgraber, i'm fine doing the second android upload if needed [14:16] seb128: i'm running on the daily-build ppa right now [14:16] but we need to get working images back asap === Guest10061 is now known as gbyers [14:16] dobey, "system-settings settings:///system/online-accounts" works here [14:16] ogra_: the publisher isn't done yet anyway [14:16] stgraber: the usermetrics fix is needed to promote a green image; the /etc/writable fix is nice to have for this image, but can wait til next one [14:17] stgraber, gyeah, thought so [14:17] dobey, do you have ubuntu-system-settings-online-accounts installed? [14:17] seb128: yes [14:17] dobey, and the command I gave you doesn't work? [14:18] ogra_: I think I know what's going on [14:18] checking quickly [14:18] seb128: it doesn't work trying to open settings:/// from within a scope [14:18] i think its just not copying on first mount [14:18] dobey, check with ted, he's the one who worked on the url-dispatcher [14:18] yep, looks like it's the case, I'm just confirming it does on the second [14:19] and it does [14:19] right, that's easy, I'll fix that [14:19] :) [14:19] seb128: nor does running that command line [14:19] Plugin settings://system/online-accounts does not exist. [14:19] err, also with settings:/// [14:19] stgraber, lets just hope the build doesnt regress again then :) [14:20] ogra_: http://paste.ubuntu.com/6179745/ [14:20] dobey, hum, can you run system-settings and click on online-accounts in there? does it work? [14:20] * ogra_ is now scared to even upload the initrd stuff [14:20] stgraber, heh, yeah, thats what i thought [14:20] seb128: yes, i can edit the accounts just fine [14:20] dobey, that's what they did for indicators: https://code.launchpad.net/~charlesk/indicator-sound/use-url-dispatcher/+merge/187608 [14:20] ok, uploading that one quickly then [14:21] dobey, but check with ted,he did most of the work on the dispatcher, I only know about the settings side [14:21] seb128: the dispatcher seems to at least halfway work. it is opening system-settings [14:21] dobey, that's what we merged in the settings side, https://code.launchpad.net/~mardy/ubuntu-system-settings/lp1227690/+merge/188284 [14:22] dobey, does "system-settings settings:///system/online-accounts" work or not? [14:22] seb128: but system-settings settings:///system/online-accounts just isn't working [14:22] ogra_: uploaded. I'll bump all the builds and then go finish to wake up while that stuff builds and publish, will be back in time for android. [14:22] seb128: no, it gives the error: Plugin settings:///system/online-accounts does not exist. [14:22] dobey, can you check /usr/share/doc/ubuntu-system-settings/changelog.Debian.gz [14:22] dobey, what is the top entry/the rev listed? [14:22] stgraber, ok, i'll keep an eye on the buildlog to make sure we dont lose anything again [14:23] seb128: 0.1+13.10.20130930-0ubuntu1 is the version i have installed [14:23] * Automatic snapshot from revision 401 [14:23] dobey, that's too old [14:24] mzanetti: ping [14:24] dobey, it has r401, the url support has been merged as r404 [14:24] dobey, https://code.launchpad.net/~mardy/ubuntu-system-settings/lp1227690/+merge/188284 [14:24] ok [14:24] ogra_: accepted and rescored [14:24] great [14:24] enjoy your coffee [14:24] seb128: that's a very deceptive error message for something it supposedly doesn't support at all [14:25] cwayne: pong [14:25] dobey, yep, sorry about that [14:26] mzanetti: hey, do you know where i could find the call to AccountService to add a click app to the launcher? [14:26] rsalveti, so i was taking a long look at the adbd upstart job, and i think we're doing it all wrong there ... i guess we should drop all the excoing into /sys and rely on android to dtrt by using properties [14:26] dobey: hey, im confused as to why this was rejected? https://code.launchpad.net/~mfisch/click-update-manager/fix-icon/+merge/188496 [14:26] *echoing [14:27] cwayne: there isn't any [14:28] ogra_: how we looking for 75? [14:28] * ogra_ waits for "patches accepted" [14:28] jdstrand: ping [14:28] mzanetti: is there any programattic way to add one? what is actually run when the user selects 'pin to launcher'? [14:28] davmor2, waiting for the last initramfs-toools-ubuntu-toouch upload to publish, then an android rebuilds and a new image will be spun [14:28] oSoMoN: hey [14:29] cwayne: rejected? [14:29] ogra_: nice thanks [14:29] jdstrand: hey, the facebook webapp is terminated by apparmor when trying to access the location (bug #1233678), would it be acceptable to update the corresponding profile to allow access to the location service? [14:29] bug 1233678 in webbrowser-app (Ubuntu) "webbrowser crashes when facebook webapp tries to access location" [Medium,Confirmed] https://launchpad.net/bugs/1233678 [14:29] dobey: i suppose by rejected i mean not approved :) [14:30] cwayne: Launcher/LauncherPanel.qml:529 [14:30] mzanetti: cool, i'll take a look, thanks [14:30] stgraber, argh ... again ! [14:31] stgraber, https://launchpadlibrarian.net/152040049/buildlog_ubuntu-saucy-armhf.initramfs-tools-ubuntu-touch_0.52_UPLOADING.txt.gz [14:31] oSoMoN: seems reasonable. note, there is no prompting for access to location service currently [14:31] cwayne: it is wrong [14:31] jdstrand: there is a prompt in the browser itself [14:31] * ogra_ blames infinity and his fakeroot upload [14:32] oSoMoN: ok-- note, there may be two prompts once trust-store is being used with location-- but we can cross that bridge when we get to it [14:32] jdstrand: agreed [14:32] jdstrand: who can I assign the bug to have the profile updated then? [14:32] ogra_: we could, but then we'd need to wait the container to be up first [14:33] ogra_: the good part of that upstart job itself is that it can start adb before the container is fully up [14:33] or even started [14:33] rsalveti, ah, crap, forgot that [14:33] oSoMoN: dbarth_ afaik [14:33] jdstrand: ok, thanks [14:34] cwayne: btw... LauncherModel has a pin() method too. you don't need to call invokeQuicklistAction [14:34] stgraber, 0.53 uploaded as no-change rebuild [14:34] mzanetti: hm, i was just looking at that, but im a little confused as to how i could call that in say, a bash script === fmasi is now known as fmasi_afk [14:35] cwayne: that's not possible [14:35] rsalveti, right, the bad part about it is that mtp and adb wrangle about the device and you get disconnectes etc [14:35] rsalveti, bug 1233613 [14:35] rsalveti, ogra_ wrt to the /sys setup... it done differently for mtp; so at least we need one way of doing it [14:35] bug 1233613 in android-tools (Ubuntu) "Adb and mtp priority issue" [Undecided,New] https://launchpad.net/bugs/1233613 [14:36] sergiusens, well, we cant do the mtp way for adbd since there is no property system [14:36] sergiusens, the mtp way is surely the cleaner one [14:36] ogra_, I know, that's why I mentioned if we wanted the setup for the bus to live on ubuntu [14:36] adbd actually forcefully unsets everything that was set [14:36] pete-woods: ping [14:36] cwayne: hi [14:37] cwayne: trying to land libusermetrics today [14:37] ogra_: that bug is because the it's setting the property again, which causes adb to reconnect [14:37] pete-woods: hi, the infographic stuff seems to have broken since build 71, but it doesn't look like anything changed in libusermetrics in that time.. [14:37] not again, I mean, right after boot [14:37] rsalveti, right, but it has to [14:37] pete-woods: wondering if you had any insight as to what else could have broken it :) [14:37] because originally it's "" [14:37] ogra_: yup [14:37] rsalveti, and when adbd starts its forced to adb [14:37] cwayne: it's something to do with the way the image is being built, it has removed privileges from the usermetrics user from /var/lib/usermetrics [14:38] ogra_: let me check what happens if you boot with the property set as "adb" [14:38] pete-woods: ah, that makes sense. is that being fixed do you know? [14:38] Laney: ping [14:38] I think stgraber is working it [14:38] gi cwayne [14:38] hi [14:38] rsalveti, ogra_ it's always adb [14:39] sergiusens: not by default [14:39] Laney: is timezone supposed to work in today's -proposed from system settings? [14:39] not yet [14:39] rsalveti, yes it is init.rc: setprop persist.sys.usb.config adb [14:39] still working out some final bugs [14:39] sergiusens: hm, did we change that? [14:39] seb128: ok, looks like there is a newer version in the PPA, and after installing it, opening it from the command line works [14:39] sergiusens: oh, right [14:39] root@ubuntu-phablet:/# cat /data/property/persist.sys.usb.config [14:39] adb [14:39] rsalveti, 4 months ago [14:39] my other change was doing that sed as part of lxc-android-config [14:39] Laney: what about from adb shell? should that gdbus call work? [14:40] no [14:40] oh ok [14:40] that is exactly what is broken [14:40] ogra_: cross your fingers more dude! [14:40] (for this rebuild) [14:40] Laney: ah, alright. was just making sure it wasnt just a system-settings app thing :) [14:40] didrocks, they are already wrapping around each other ! [14:40] heh [14:41] didrocks, that still doesnt solve the issue that the build in itself is unreliable [14:41] and i dont get why [14:41] ogra_: sergiusens: well, not sure if there is an easy way to fix that, maybe starting with adb,mtp, but this is actually the expected behavior it seems [14:41] as it's changing the usb mode [14:41] yeah, starting to be frighten about that one [14:41] * ogra_ blames infinity and his fakeroot upload even more [14:41] Do you have any non-circumstantial evidence for that? [14:41] Like traces? [14:41] cjwatson, none [14:41] I suggest gathering that before blaming [14:42] cjwatson, just the timing mtaches the breakage appearing [14:42] pmcgowan: Mirv many thanks guys I now have a video playing with applied shaders linked to sliders "touchStyle" [14:42] *matches [14:42] cjwatson, and thats the odd package that uses fakechroot and fakeroot all over the place for building [14:42] stgraber: not sure whether you saw the bug mail, we still seem to have a problem with "synced" [14:42] rsalveti, fastest solution is check the config on ubuntu and forget about the property system (from an upstart perspective) [14:42] pitti, fixed [14:43] jarope, awesome [14:43] pitti, according to stephane [14:43] seb128: but opening it from the scope via the Unity API doesn't seem to open the panel. soo tedg ? it seems trying to open online-accounts panel from with a scope via the unity actions API isn't working. [14:43] ogra_: So what you're saying is that nobody has tried to investigate directly? [14:43] ogra_: oh, awesome! [14:43] sergiusens: but then what happens once mtp is started? [14:43] Adding binary /bin/chown [14:43] Adding binary /bin/readlink [14:43] Adding binary /usr/bin/adbd [14:43] didrocks, ^^^ [14:43] we're good with this one [14:43] \o/ [14:43] ogra_: so, some release team member (Laney, *hint* *hint*) still needs to review/accept the systemd fix for that [14:43] phew [14:43] rsalveti, we should only start mtp if the switch we control on ubuntu has mtp in it [14:43] the problem is that we don't want adb to reconnect while we change the usb mode [14:44] sergiusens: not sure I follow [14:44] pitti: is the writable stuff fixed? [14:44] sergiusens, rsalveti, we also need to drop adb as a default for the final image [14:44] pmcgowan: it is indeed means I can tackle another year of Uni assignments with out having to use GLUT and try once again to push Qt/QML as a more modern and suitable windowing system [14:44] Laney, in flight, but should be [14:44] cjwatson: could you take a quick look at the systemd diff in unapproved? [14:44] Laney: yes, that's the point of that systemd upload, to get along with the change to relative symlinks [14:44] ogra_: right, then do that first [14:44] ogra_, I had that implemented a month ago, but we decided to go with the property system [14:44] Laney: I didn't look at why /etc/writable/ is empty, but ogra_ just said that stgraber fixed that [14:44] yes [14:44] ogra_: so, from what pitti told, we need systemd for having the system-settings ui date & time to work? [14:44] there were two parts to fix [14:45] sergiusens, yeah i remembered that thats why i saked you about it yesterday :) [14:45] rsalveti, wait, we decided to use the property system for enabling disabling adb anyways [14:45] *asked [14:45] didrocks, right [14:45] didrocks: yes, currently it's trying to change the readonly files [14:45] sergiusens: right, so let's get that done first [14:45] ok, let's wait on that one as well anyway [14:45] ogra_, rsalveti don't get hasty about stuff [14:45] didrocks, we also still need the subsequent android upload/build [14:45] sergiusens, i surely wont :) [14:45] :-) [14:45] ogra_: not for that one? just for the initramfs part? === fmasi_afk is now known as fmasi [14:46] sergiusens, rsalveti i managed to get adbd run as phablet user btw ... [14:46] didrocks: https://launchpad.net/ubuntu/saucy/+queue?queue_state=1&queue_text=systemd this one, plus presumably an initramfs/android rebuild for the "sycned" fix [14:46] sergiusens, rsalveti, funnily it still logs me in as root [14:46] Laney: LGTM if the intent is that result is "/etc/" at the end [14:46] ogra_: hahaha [14:46] well, we'll rebuild the android side for a good reason ;) [14:46] dobey, I'm not sure what you mean by Unity Actions here, URL dispatcher? [14:46] ogra_: right, but we don't need to fix that now [14:46] no [14:46] Laney: Actually [14:46] sudo will break all your effort :-) [14:46] just turning off the default on release day should suffice [14:46] Laney: Won't that break if the symlink is absolute after all? [14:46] rsalveti, sudo is fine [14:46] Laney: This all seems like a lot of very dubious effort to avoid a few string copies [14:47] cjwatson: meet pitti [14:47] :P [14:47] lol [14:47] pitti: ^- [14:47] ogra_: if we can do sudo, we can start adb as root :-) [14:47] tedg: i don't know what UnityActionResponse uses, [14:47] rsalveti, indeed [14:47] pitti: I would recommend reaching for some equivalent of canonicalize_file_name instead of all this sketchy stuff [14:47] return new Unity.ActivationResponse (Unity.HandledType.NOT_HANDLED, [14:47] gah, paste fail [14:47] rsalveti, unless adbd doesnt allow root at all ;) [14:47] pitti: It worries me when code only works with one of relative or absolute symlinks; that's a timebomb for later [14:47] cjwatson: I can't use canonicalize_file_name() [14:48] Ah, yes, only one level [14:48] cjwatson: I can use the dynamic malloc stuff from util.c and copy it back into the static buffer [14:48] tedg: Unity.ActivationResponse (Unity.HandledType.NOT_HANDLED, "settings:///system/online-accounts"); <- this results in system-settings being opened, but not to the accounts page [14:48] But still, is avoiding readlink_and_make_absolute really worthwhile? [14:48] I would say a test for this code should be that it works with either relative or absolute [14:48] ogra_: wow, that's scary... [14:48] ogra_: alright, bumping 0.53 now [14:48] cjwatson: ok, I'll change it then [14:48] thanks [14:48] * pitti rejects from the queue [14:48] stgraber, can you bump 0.53 on the other arches [14:48] heh, snap [14:48] kenvandine: ping [14:49] mhall119, pong [14:49] kenvandine: hey, I have questions about the content-hub [14:49] dobey, Which version of system settings? I don't think that support got added until the 0930 release. [14:49] kenvandine: I'd like to get the docviewer-app working so you can open PDF (and other) files in it from another app (like webbrowser-app), and I was told that this should go through the content-hub [14:49] mhall119, sure, what's up? [14:50] tedg: system-settings does the right thing if i run it on the command line with the url [14:50] well, i have to have it already running on the device, to actually get a window, when i run it from within adb shell [14:50] mhall119, yes indeed [14:50] kenvandine: my test usecase is 1) Open the filemanager-app, 2) click "open" on a pdf file, 3) docviewer-app is called with the file URI as an argument [14:50] mhall119, i need to reboot, i [14:50] 'll ping you in a moment [14:50] k [14:51] dobey, Hmm, I need to upgrade to get that version before I can check. [14:52] dobey, Can you look to see what the /proc/*/cmdline is for that process? [14:52] dobey, See if it got passed the URL. [14:53] tedg: it appears it wasn't passed it: [14:53] system-settings^@--desktop_file_hint=/usr/share/applications/ubuntu-system-settings.desktop^@ [14:53] dobey, That's probably why it's not working :-) [14:53] indeed [14:53] tedg: is this your bug, or libunity? [14:54] dobey, Looks like system-settings... doesn't have a %u in it's Exec line for it's desktop file. [14:54] mhall119, actually that should more like url-dispatcher [14:54] dobey, I'll propose a merge. [14:54] tedg, ^^ [14:54] should have paid attention to apt-get dist-upgrade: http://paste.ubuntu.com/6179859/ [14:54] didrocks: ^^ [14:54] stgraber, ogra_: Hmm how is the +x of things we copy fixed? [14:54] I don't see this change [14:54] slangasek: ^ [14:54] kenvandine: tedg and pmcgowan both agreed it should be the content-hub that pushed the file uri to docviewer-app [14:54] not url-dispatcher [14:54] lool, no change needed for this ... it already happens [14:55] mhall119, humm, which apps provides the file? [14:55] s/apps/app [14:55] kenvandine: potentially any app can provide the source file [14:55] kenvandine, It shouldn't be "a url" it should be "content" [14:55] kenvandine, i.e. the browser downloads a PDF [14:55] ogra_: so what was the missing +x flag? [14:55] tedg: ah. adding that does fix it here! [14:55] mhall119: what does install libunity-core-6.0-8 tells you? [14:55] tedg: thanks [14:55] lool, the hook in the initramfs creation that copies chown/chmod etc in place doesnt get executed during build [14:55] tedg: currently the docviewer-app deals with file uris [14:55] lool, randomly [14:55] tedg, agreed.. but filemanager-app is outside of that [14:55] didrocks: I'mwaiting for dist-upgrade to finish [14:56] lool, so we end up without the needed binaries in the initrd if we dont watch the build closely [14:56] ogra_: but does that hook need to be +c? [14:56] didrocks: then I'll try installing that [14:56] +x [14:56] lool, that hook is +x [14:56] always was [14:56] not in lp:ubuntu AFAICT [14:56] kenvandine, ? I mean, it still needs helpers? Isn't it the "last resort" for content-hub? [14:56] webbrowser-app downloads the pdf, does it download it to a central location or it's own app storage? [14:56] oh it is now [14:56] lool, it is with apt-get source3 and it is also +x in the binary package [14:57] lool, that never changed [14:57] kenvandine, Own app storage [14:57] tedg, in the content-hub view of the world, an app owns the content and provides the UI to export it [14:57] another app wants it, and makes the request [14:57] so webbrowser-app would provide the export for it [14:57] kenvandine, Hmm, last we'd discussed it, we needed a push as well. For, for instance, printing. [14:57] lool, but the randomness of the failure on the buildd is worrying [14:58] tedg, yes... which is providing an export [14:58] s/worrying/extremely scary/ [14:58] kenvandine, ? Isn't printing providing an import? [14:58] tedg, we still don't have a way to initiate it from that side [14:58] kenvandine, So that's the feature that mhall119 needs :-) [14:58] kenvandine, He's a sink not a source [14:58] stgraber, ogra_: So slangasek was pointing out scripts/touch is not +x in lp:ubuntu/... branch [14:58] yeah [14:58] lool, but it is [14:59] we don't have a model for 3 apps, just source and destination [14:59] so webbrowser-app -> docviewer-app [14:59] i have all versions here locally of that package ... well all that i ever uploaded or sponsored [14:59] they all are *x [14:59] +x [14:59] what we need for that use case is webbrowser-app to be able to say i need to export this content to an app that knows what to do with it [14:59] which would be docviewer-app [15:00] docviewer-app would get started and view the content [15:00] kenvandine: and we don't have that currently? [15:00] lool: hooks/touch certainly is executable, scripts/touch isn't but doesn't need to be [15:00] mhall119, 2.0 [15:00] mhall119, no.... sorry [15:00] tedg: so can we hack it onto url-dispatcher until then? [15:00] $ tar tzvf initramfs-tools-ubuntu-touch_0.52.tar.gz | grep scripts/touch [15:00] -rw-r--r-- 0/0 7811 2013-10-01 15:19 initramfs-tools-ubuntu-touch-0.52/scripts/touch [15:00] ogra_: ^- [15:00] bfiller: do you know if there's a way to know the names of the settings plugins ? i am trying to launch settings:///system/online-accounts but it says it's not handleable, so maybe online-accounts is not the right name [15:00] ogra_: Clearly nothing to do with fakeroot [15:01] mhall119, you could trick it with url-dispatcher [15:01] ogra_: The source package is wrong [15:01] cjwatson: hooks/touch not scripts/touch [15:01] cjwatson, well, its clearly +x for me after apt-get source [15:01] have filemanager-app open docviewer-app with a URI [15:01] Oh, you were talking about scripts/touch above [15:01] stgraber: why doesn't it need to be +x? [15:01] Yeah, hooks/touch is +x [15:01] kenvandine: I know it's technically possible, what I don't know is whether tedg will approve the MP to do it :) [15:01] yeah [15:01] lool: ah, scripts/touch is in the same class as scripts/{local,nfs}, so doesn't need to be executable - l 4good now [15:01] when docviewer-app opens, it sees the URI and requests content from the hub [15:01] slangasek: ok [15:01] lool, it is sourced by /init [15:01] alright, all good [15:01] which then could silently (no UI) create the transfer based on the URI [15:02] slangasek, well, nothing is good, since the buioldd randomly ignores it [15:02] right, that's "all good now" - stupid unhelpful cpu lag [15:02] lool: because it gets sourced, not called [15:02] ogra_: randomly ignores what? [15:02] seems every second upload the hook doesnt get executred at all [15:03] interesting hack though, i bet it could do it all without even opening the providing app [15:03] ogra_: please point me to an example [15:03] slangasek, https://launchpad.net/ubuntu/+source/initramfs-tools-ubuntu-touch/0.53/+build/5066796/+files/buildlog_ubuntu-saucy-armhf.initramfs-tools-ubuntu-touch_0.53_UPLOADING.txt.gz working build ... vs [15:03] slangasek, broken build https://launchpad.net/ubuntu/+source/initramfs-tools-ubuntu-touch/0.52/+build/5066754/+files/buildlog_ubuntu-saucy-armhf.initramfs-tools-ubuntu-touch_0.52_UPLOADING.txt.gz [15:03] didrocks: libunity-core-6.0-8: Installed: 7.1.1+13.10.20130927.1-0ubuntu1 [15:03] look at the update-iniramfs output [15:04] so I got it at some point === dandrader is now known as dandrader|lunch [15:04] not sure why the previous warnings [15:04] mhall119: weird… not sure about the -common complain [15:04] slangasek: last time it happened we uploaded a no change rebuild within 10min and that fixed it (so that should clear any chance of change on the builder side between the two builds) [15:05] stgraber: ok, but why was it *wrong*? [15:05] slangasek, thats the myth here [15:05] it seems to just randomly happen [15:05] and only started yesterday [15:06] this is creepy [15:08] stgraber, ogra_: Can we fail the build when we detect this? [15:08] stgraber: hey, its looking like /home/phablet has incorrect permissions in latest build, did you already know about this? [15:09] ogra_: I'm confused. This is a build log for initramfs-tools-ubuntu-touch, which is the package that *contains* /usr/share/initramfs-tools/hooks/touch. Hopefully this package does not have a circular dep on itself? [15:09] ping cyphermox or whomever. do you know the config file needed for a "straight talk" (att reseller) sim in order for data to work? [15:09] slangasek, no, but it creates the initrd during build [15:10] cwayne: yes I did, same problem as everything else, we don't have chown in the initrd [15:10] slangasek, and thats not the point, nothing changed in that package wrt building ... it was working perfectly fine pover the last months exactly as it is === kentb-out is now known as kentb [15:10] slangasek, it stopped working yesterday evening at some point [15:11] stgraber: alright, figured you were aware, but that i should double-check, as it seems bad :) [15:11] ogra_: no, that's very much part of the point. Why is it building an initramfs during build that depends on a hook provided this package to be *already installed* on the filesystem? [15:12] ogra_: I see a debootstrap in the build, I don't see initramfs-tools-ubuntu-touch being unpacked in that chroot - so of course it's not executable if it doesn't exist? [15:12] slangasek, why did it stop working all of a sudden when it worked for months ? [15:12] slangasek, it copies the bits in place during build [15:12] look at the source [15:12] build-initrd.sh specifically [15:13] (which hasnt changed in more than 6 weeks and never caused issues, something else must have changed that causes this behavior) [15:14] cyphermox, awe_ : sims from wallmart (straight talk) need a config file changing do you have details on what and how at all? [15:14] ogra_: ok, I see that here. It would be nice to have more verbosity there so you can see that from the build log... [15:14] slangasek, the fakeroot upload suspicilusly matches the time where it started failing ... but i have no evidence that would point at it [15:14] slangasek, ogra_: I wonder whether the problem is with the cps being out of the fakeroot/fakechroots setup [15:14] lool, but why didnt that cause issues for the last three months ? [15:15] ogra_: maybe it's indeed random because fakechroot has no data on the files? [15:15] or fakeroot [15:15] that script is used since we have flipped containers and didnt change .... it moved over about 6 weeks ago when i merged both packages [15:15] but had no issues even after the megre (and there were plenty uploads of it) [15:15] I suspect that we are invoking fakeroot in invalid ways [15:15] There are at least two things I see wrong [15:16] (a) using -i without a matching -s and expecting anything to work [15:16] Saviq: heya, any update from the Qt guys about the missing scope views? [15:16] (b) manipulating things inside the fakeroot when outside the saved environment [15:16] both of these things are documented as "you cannot rely on this working" [15:16] cjwatson, but again, that worked until yesterday and hasnt changed [15:16] ogra_: that's the nature of doing unreliable things [15:16] perfectly correct changes may break you [15:17] hmpf [15:17] cjwatson: I hope you aren't breaking ogra_ s they are hard to come by [15:17] haha [15:18] similarly, I'd have no sympathy if we were relying on undefined behaviour in C and a gcc/glibc change broke it [15:18] cwayne, no update, we're going for distro-patching [15:18] Saviq: great, any idea when it might land? [15:18] cwayne, we'll try to get it in tomorrow, then however long it takes to get into distro [15:20] Saviq: thanks mfisch^ [15:23] Do you know whether I can use reverse USB tethering with Ubuntu Touch somehow? I would like to use the Ethernet connection of my Ubuntu workstation with my Nexus 7 through USB [15:24] davmor2, you should be able to change the details of the data context using the ofono scripts [15:24] pitti: do you think we'll have a systemd soon or we shouldn't rely on it? [15:25] awe_: I just found the bug and directed the people inquiring to it thanks https://bugs.launchpad.net/ubuntu/+source/ofono/+bug/1221969 [15:25] Ubuntu bug 1221969 in ofono (Ubuntu) "3g data not working with Straightalk (MVNO) SIM" [Undecided,Confirmed] [15:25] davmor2, but actually... probably better to stop NM, stop ofono, and then hand edit the gprs settings file in ofono [15:25] davmor2, thanks I can update the bug if need be [15:25] davmor2, the details of what the settings should be can be found on straighttalk's website [15:26] awe_: great thanks [15:26] stgraber: i'm not getting cellular data today -- do you think that could be related to the initrd as well, or should we chase this down? [15:27] davmor2, comment #3 has detailed instructions [15:27] cwayne: wait for the next image, do a clean reflash and try again, I have no idea what else may be broken due to a dozen binaries being missing in the initrd [15:28] stgraber: righto [15:28] awe_: yeap thanks, I was pretty sure there was a bug I just couldn't find it on my initial search :) [15:29] didrocks, ping [15:29] mandel: pong [15:31] stgraber, how about the android upload ? [15:31] oSoMoN: in system settings, can you manage to log in to a facebook account ? the browser there is pretty broken for me. it's super hard to tap into the text fields and sometime it freezes [15:31] ogra_: I'm trying to fix this build-initrd.sh script up to be more reliable [15:31] * ogra_ hugs cjwatson [15:32] ogra_: not yet, waiting for i386 to publish [15:33] ah [15:33] i see [15:33] should be good after the next publisher (it finally built) [15:33] yeah [15:33] nerochiaro: yeah, I can manage to log in, although the auth page is definitely not rendered as it should [15:34] oSoMoN: ok, maybe my fingers are too fat. it's a total pain [15:34] stgraber: please review this for sanity: https://bugs.launchpad.net/ubuntu-system-image/+bug/1221844/comments/3 [15:34] Ubuntu bug 1221844 in Ubuntu system image "Support channel aliases tracking" [Critical,In progress] [15:35] mhr3: we'd like to add that icon ordering fix to the landing plan, any comments? [15:35] mfisch, should be there already [15:35] mfisch, part of #112 in the plan [15:35] Let's ask this in another way: do you if it's possible to enable USB networking between an Ubuntu workstationa and a Nexus 7 with Ubuntu Touch? [15:36] oSoMoN: can you please file a bug for that bad rendering ? i think you know better what's wrong since you have more exprience with browser [15:36] mfisch, and it's not icon ordering, it's category ordering [15:36] mhr3: yeah, you're right, category ordering [15:36] mfisch, for a moment i was worried you want something else :) [15:36] nerochiaro: it’s not the browser itself, it’s the online accounts app, and mardy is already aware of the problem [15:37] mhr3: #112? The last landing I see is 78 [15:37] oSoMoN: excellent [15:38] ogra_: stgraber: let's forget systemd for now I guess [15:38] mfisch, well... you thought landing things is simple and quick? :) [15:38] I previously had the onValueChanged: working for my sliders but it looks like this isnt implemented with Ubuntu is this correct? and if so what would be the work around [15:39] didrocks, well, the buildds are busy, queues are full etc it seems [15:39] yeah [15:39] didrocks, i fear it has a good chance to land at the same time as android [15:39] (at 10pm or so) [15:39] mfisch, not to mention that usually merging a home-scope branch takes 15minutes, now it's 5 hours since approve and still not merged... [15:39] ogra_: yeah, but systemd will need an android rebuild as well [15:40] mhr3: I think that fginther said he had to re-approve it [15:40] from what pitti told, so I guess we won't have it in time [15:40] didrocks, oh ? [15:40] why would it need that [15:40] didrocks, systemd doesnt touch anything in the container [15:40] mfisch, right, and that was almost 2hours ago [15:40] didrocks, and the fix for /etc/writable is already in [15:41] barry: "This leads to an upgrade path of 0:200:201:304" would be more like "0:2" or something along those lines, but yes, if we have alias in channels.json and that doesn't match the value from channel.ini, then consider we're at build 0 and update with that [15:41] ogra_: ok, believe you then ;) [15:41] pitti, ^^^ ? [15:42] stgraber: sorry, those paths are taken from test data, so just assume that 200 == 2 or something :). the key point is that we're *not* filtering out deltas in this case. we use the normal upgrade calculation rules, except that we squash build number to 0. it would be exactly equivalent to calling s-i-cli --build 0 [15:42] stgraber: agreed? [15:43] barry: yep [15:43] stgraber: cool, thanks. [15:46] davmor2: sorry, I do not have that information [15:47] cyphermox: no worries sorted anyhow thanks [15:47] though perhaps I should go try to get a SIM from my local walmart to see if I can break NM with it [15:47] so; I will have a package to test about now for the auto reconnection [15:48] cyphermox: I hit the auto disconnect I've grabbed some logs but then done nothing more with them cause I fail and ran out of time :) [15:48] didrocks: do you have time to expand on what this loopback issue was the other day? [15:48] ev: not really today [15:49] didrocks: okay, no worries [15:49] stgraber: last question i swear! when might we expect build 75 [15:49] oSoMoN: i can't repro failures in pasting images in notes. can you provide a clean case that i can repro here ? number and text of notes might matter pershaps ? [15:49] cwayne: as soon as all the bits are built [15:49] nerochiaro: let me test again [15:49] ev: sorry ;) [15:49] davmor2: ha! yeah, i was looking for more of a time [15:50] didrocks: entirely understandable given the landing meeting and whatnot === gatox is now known as gatox_lunch [15:51] nerochiaro: ahah, somehow I just managed to copy your joker-like cat face 6 times in the same note, in just one click [15:51] cwayne: see this piece of string? about that long give or take a minute? is that any better? [15:51] davmor2: if you have time for testing the fix; I put the packages up on http://people.ubuntu.com/~mathieu-tl/network-manager/ [15:51] nerochiaro: so pasting works, even though the process is really clunky [15:51] awe_: ^^ [15:51] ChickenCutlass: ^^ [15:52] ogra_: what's up? (re from supermarket) [15:52] awe_: this may help with phonesim as well actually. [15:52] cyphermox: nice I'll give it a go once we have image 75 in play so everthing else works [15:52] awe [15:52] *aye [15:52] damn keyboard :) [15:52] oSoMoN: still that's not right [15:52] oSoMoN: means the click gets repeated or something [15:53] cyphermox: stop doing irc on the phone then ;) [15:53] didrocks: it's in! [15:53] oSoMoN: and you were not able to paste before [15:53] sil2100: phew ;) [15:53] nerochiaro: no it’s not indeed, are you in touch with the SDK team to address this popover issue? [15:53] davmor2: I wish [15:53] oSoMoN: i'm following the bug, just commented on it. and tomorrow i'll poke zsombi and try to get to the bottom of the issue with him [15:54] cyphermox, ack [15:54] oSoMoN: do you know who's responsible for the url dispatcher ? [15:54] nerochiaro: I think tedg is [15:55] ogra_: stgraber: what do I need to do to make the android container work? stick android-ramdisk.img into /boot/android-ramdisk.img and that's it? [15:56] beuno: hey, i just got a 500 error on uploading an update to the FB webapp; did something wrong? [15:57] tedg: do you know what's the proper way to use the url dispatcher to start the online account settings ? i try settings:///system/online-accounts but it's not working, it tells me it's not dispatchable [15:57] xnox, yeah, that should get the android initrd up [15:57] dbarth_, first guess is forgot to bump the version # [15:58] dbarth_, if not, for my second guess, I'll be looking at logs :) [15:58] ogra_: http://paste.ubuntu.com/6180111/ should be much more reliable (I couldn't actually reproduce the original problem directly on the porter box, but it avoids the things explicitly documented as unreliable in fakeroot(1)). Should I upload this? It's not needed right now, but there doesn't appear to be a VCS for me to commit it to and forget about it. [15:58] beuno: 1.0.3 -> 1.0.5, but let me know [15:58] cjwatson, Laney: https://launchpad.net/ubuntu/saucy/+queue?queue_state=1&queue_text=systemd is now diffy [15:58] ogra_: and then the system partition will be mounted inside it? [15:58] cyphermox, ack [15:59] cjwatson, Laney: tested with relative/absolute/no symlinks [15:59] dbarth_, I see the error but it looks like a transient network one. Can you try again? [15:59] cjwatson, please wait until stgraber uploaded android and the currently working initrd was picked up [15:59] ogra_: i think I need qemud socket from the android container to be available in ubuntu, such that ubuntu's adbd can talk to the host via it. [15:59] beuno: ok [15:59] beuno: you were right:) [16:00] cjwatson, we need a working image as highest prio, else all testing is stuck [16:00] pitti: Looks safer, thanks [16:00] dbarth_, I'll chase it though, both for retrying as well as the root cause. Thanks. [16:00] xnox, your vm should have its own /sys and its own kernel, no ? [16:00] ogra_: Sure, whatever [16:00] ogra_: android uploaded [16:00] xnox, why would you talk to the host for adbd [16:00] stgraber, yay, thanks [16:01] ogra_: I just don't want to lose the code. Perhaps you could institute a VCS? [16:01] cjwatson, hmm, yeah [16:01] cjwatson: I've been using ubuntu:initramfs-tools-ubuntu-touch personally [16:01] * ogra_ will try to find a minute to do that [16:01] I can just use lp:ubuntu/initramfs-tools-ubuntu-touch if you like [16:01] didrocks, ogra_: I don't believe the systemd upload will need an android rebuild; that's just runtime stuff [16:01] thats also fine, yeah [16:01] didrocks: ah right, what ogra said [16:01] pitti, thanks [16:02] ok ;) [16:02] I'll commit once the importer catches up, then [16:02] Saviq, sil2100: any news regarding the ap-issues on jenkins-ci? [16:02] ogra_: Should I approve this systemd change? [16:02] cjwatson, yes please [16:02] I'm happy with it but don't know if you want it in this image [16:02] MacSlow: not from my side sadly... [16:02] ogra_: qemu creates qemud serial device, inside the VM qemud multiplexers sits on it and allows mulptile things talk to qemu. On the host one can talk to qemu via that serial device. Thus to have working adb client, I need to launch adb under qemud in the emulator. [16:03] would be good to have it in the next build [16:03] OK, done [16:03] sil2100, ok [16:07] ogra_: \o/ ha, I haz telnet ;-) [16:08] yay [16:10] t...elnet [16:10] hehe [16:12] xnox: now test it with towel.blinkenlights.nl me hopes he got the starwars telnet address right :D [16:16] awe_: did you ever successfully use the phone simulator for a call that gets answered? [16:16] awe_: according to /usr/share/phonesim/default.xml that should work with dialing 166, but it doesn't [16:17] awe_: other "magic" numbers (like 199 which calls you back) work fine [16:20] jdstrand, hey, was there any progress on the usermetric apparmor rules? [16:21] pitti, no I haven't touched the phonesim at all... [16:22] awe_: ok, thanks [16:22] awe_: maybe we need the mock after all [16:22] pitti, been focusing all of my testing on direct on-phone testing === fmasi is now known as fmasi_afk [16:23] pitti, have you looked to see whether there's been any work on the phonesim driver upstream, and/or checked with any of the ofono devs? [16:23] awe_: so it seems with the current phonesim I can't test "outgoing call gets answered" and "receive SMS" (or perhaps I can hack up the XML for that, haven't dug into how that works yet) [16:23] awe_: no, not yet; I started with asking you, tiago, and boiko [16:23] sergiusens: http://launchpadlibrarian.net/151383351/qtmultimedia-opensource-src-touch_5.1.1%2Bgit20130920%2B5b12abb862-0ubuntu1_5.1.1%2Bgit20130920%2B5b12abb862-0ubuntu2.diff.gz - any reason you didn't just remove those from debian/control? Have I missed something subtle about how this package works? [16:24] sergiusens: The reason I ask is that having them still be in debian/control confuses some archive scripts [16:24] sergiusens: (That's why those packages are still in the archive at an older version and haven't been semi-automatically cleaned up) [16:24] pitti, when you say "maybe we need the mock after all", I assume you mean a dbus mock of ofono, not the umockdev approach? [16:24] awe_: right [16:24] cjwatson, oh, I kept it believing fail-missing wouldn't bail out on me [16:25] k [16:25] fginther: ping [16:25] awe_: tiago just points out that phonesim has a d-bus interface which can run scripts, that sounds promising [16:25] awe_: anyway, thanks! [16:25] mfisch, pong [16:25] sergiusens: Hmm, OK, well I can remove them manually, just wondered [16:25] pitti: I know phonesim has some scripting capabilities, but they are very poor if I remember correctly, but maybe it is worth a try? [16:26] boiko: yep, I'll poke that; seems much easier than driving the phonesim GUI through AP [16:26] cjwatson, we still build everything as in the original, but don't want those packages at all [16:26] sergiusens: Yep, removed them now [16:26] Just for future reference, things work more smoothly if you drop them from debian/control [16:26] pitti, yw [16:27] That way we know the binary packages are orphaned and safe to clean up [16:27] stgraber: where to the logs from the upgrade end? [16:27] I am not sure which log barry is referring to in bug 1233094 [16:27] bug 1233094 in system-image (Ubuntu) "Upgrade fail" [Undecided,Incomplete] https://launchpad.net/bugs/1233094 [16:28] cjwatson, ack, thanks [16:28] stgraber: forget it , I shaw it now [16:28] stgraber: I can reproduce the problem as long as I don't change the client.ini file [16:29] stgraber: when I increase the logging it doesn't happen [16:29] for some reason [16:29] boiko: how does the dialer app store the call log? (the test should clean that) [16:30] pitti: ~/.local/share/history-service/history.sqlite [16:30] ls [16:30] gema: can you reproduce this using system-image-cli on the command line too? [16:30] boiko: ah, thanks [16:30] pitti: IIRC om26er did some code to move that away and restore at the end of the test, can't remember if it was for dialer-app or messaging-app [16:30] stgraber: I haven't tried that, what is the command you want me to use to try? [16:30] stgraber: reflashing back atm [16:31] gema: otherwise it may be some dbus related issue or something to do with the downloader code which we're changing in the next image to use the download service instead (so your problem may go away then). Anyway barry knows that stuff way more than I do since he's the one working on the client... [16:31] boiko: easy enough; i. e. move file away and kill history-daemon in setUp() , restore it in tearDown() [16:31] pitti: yep [16:31] barry: ^^ [16:31] gema: just run "system-image-cli" as root that will figure out the upgrade path, download and reboot. It won't print anything though unless you pass -v (or some extra -v to get even more verbose output) [16:31] pitti: and kill history-service again in tearDown() just in case [16:31] right [16:32] boiko, pitti we do that for a few other apps as well. e.g calculator sudoku-app. we have not yet done that for dialer/messaging [16:32] stgraber: ack [16:32] boiko, you probably reviewed my branch for the calculator [16:32] mfisch, home-scope branch landed in trunk [16:32] gema: we also had a socket bug which slangasek fixed recently and that was causing the download to fail under some condition and may cause what you've been seeing (I had that happen to me on the command line a couple of times with pretty out of date devices) [16:32] om26er: yep, but I remember you asked me how to do that for the history-service stuff, I thought you had implemented that already :) [16:32] I did that in the connected tests for the phone-app in the past. [16:33] boiko, Yes, I was working on that when I was at the sprint in Lex but now someone else is working on it [16:33] stgraber: I am using the first image that shouldn't require a full upgrade, fwiw [16:33] stgraber: hum, I'm pretty sure I didn't fix any socket bugs except wrt the testsuite itself [16:33] om26er: ok, np, thanks [16:33] slangasek: we are talking about bug 1233094 [16:33] bug 1233094 in system-image (Ubuntu) "Upgrade fail" [Undecided,Incomplete] https://launchpad.net/bugs/1233094 [16:34] om26er, boiko: I implemented it now, thanks [16:34] mhr3: cool, now we just need it in the image! [16:34] mfisch, i wouldn't hold my breath there [16:35] yeah I'm not [16:35] we may just build it ourselves [16:35] mfisch, fwiw it should be in the daily build ppa soonish [16:35] gema: we've landed the new system-image architecture based on ubuntu-download-manager in part because of unreliabilities in the previous architecture, and most of these bugs aren't going to be solvable retroactively because the bugs are in the client in the old image. I think the only thing we can advise here is "reflash to an image that uses the new code". [16:36] mhr3: perfect then we dont even have to buld it [16:36] slangasek: oh, ok, didn't look at the actual code change. So anyway, I've had some weird timeout related issues here when an image is made of a lot of files, sometimes it'd take 2-3 tries to get everything downloaded properly. Hopefully the download maanger will fix that issue. [16:36] mfisch, soonish == ~12hours [16:36] gema: now, I think separate to that we should be thinking about having some tests which stress the downloader client by artificially feeding it a large download (~200M) so that we can verify it's handled correctly; but we need to run such tests against current client code, not old stuff [16:36] slangasek: ok, so what is the first image with the new architecture? [16:36] gema: 73 [16:37] (not yet promoted) [16:37] slangasek: ok, so I should be testing from 73 onwards [16:37] which hasn't even landed [16:37] gema: yes please :) [16:37] it's still in devel-proposed, yes [16:37] slangasek: you do realise that we are going to have almost no change of catching this kind of bug because our upgrade path is not going to be big enough, right? [16:37] slangasek: this is a calculated risk, I assume? [16:38] mfisch, but if you ask someone nicely they might be able to publish it there manually [16:38] are we going to get contacts syncing again anytime soon? [16:38] gema: as I said, I think we should have tests with artificially-large updates [16:38] slangasek: do you have some of those? [16:38] gema: the upgrade path from v. 5 to v. 70 (the current version on the devel channel) is going to go straight to 70 and download it in one large block; we can simulate that [16:38] mhr3: we're waiting for a bunch of other stuff, so I can wait [16:39] mhr3: is it the usual daily PPA or a special unity one? [16:39] stgraber: ^^ could you prepare a stand-alone artificially large system-image to use for upgrade testing? [16:39] cwayne: We'll have a deb soonish to put apps at the top of the home scope [16:39] slangasek: ack, that sounds good [16:39] mfisch, this one https://launchpad.net/~ubuntu-unity/+archive/daily-build [16:39] mhr3: perfect [16:39] stgraber: actually, I think it would be sufficient to prepare an artificial channel that always forces a full download [16:40] slangasek: well, just call system-image-cli with -b 0 then and you'll get a full ;) [16:40] slangasek: or even better, with -f full which will ignore any existing delta image in the resolution [16:41] stgraber: not what I am trying to test, sorry [16:41] stgraber: I am trying to test a different upgrade path [16:41] and make sure the UI does the right thing [16:41] etc [16:42] yay, android built [16:42] so what you could do is use -5 but first upgrade system-image to the latest version on the device [16:42] stgraber: so I'd do a phablet-flash ubuntu-touch --revision -5 from 73, right? [16:42] so switch to writable mode, update the system-image packages with apt, switch back to read-only, reboot the device, run the update [16:43] yep, same thing as you were testing so far, except that you remount you / writable, update system-image and reboot before proceeding with the update [16:43] that way you'll actually be testing the current system-image client code [16:43] something along the lines of: [16:43] adb shell mount -o remount,rw / [16:43] adb shell apt-get update [16:43] I know what you mean [16:44] I don't think it is going to work [16:44] it's not what I am trying to test [16:44] adb shell apt-get install system-image-cli system-image-common system-image-dbus [16:44] adb shell reboot [16:44] then test as usual [16:44] gema: what are you trying to test then? [16:45] stgraber: usually I'd flash X images back and try to upgrade to current, first test case is going back as many as it takes to trigger the full update [16:45] stgraber: we need a standing test case based on recent images, not retrofitting system-image packages onto a version 5 image [16:45] stgraber: then one less [16:45] stgraber: and it needs to work from the UI, not just from system-image-cli [16:45] um, I'm running ubuntu-system daily stable images......why is apt-cache running and using up my CPU? [16:45] stgraber: and make sure in the first case I get a full image upgrade [16:45] and in the second I don't [16:46] sergiusens: yes-- I am waiting on bug #1231863 to land in the archive [16:46] bug 1231863 in ubuntu-ui-toolkit (Ubuntu Saucy) "Local Sqlite databases are still created/stored in incorrect location" [Critical,Triaged] https://launchpad.net/bugs/1231863 [16:46] stgraber: so I think this could be done by forcing /etc/ubuntu-build on the filesystem, right? [16:46] mhall119: use "ps axf" to find out what's calling it [16:46] sergiusens: it is unrelated to the usermetrics, but because it is difficult getting things into the image, I am gathering up the various bug fixes into one upload [16:46] mhall119: there's still an apt cache on ubuntu-system images even though it's RO [16:47] cjwatson: well it's not running anymore, but if I see it again I'll do that [16:47] cjwatson: right, but it shouldn't be using 50% of my CPU [16:47] Sure === gatox_lunch is now known as gatox [16:47] slangasek: hmm, then for each image you need to have a server import that, generate a bunch of big updates in a way that you result with X deltas before the upgrader decides to use a full and have your test device use that. It's definitely possible but we'd first need to get the ports/oem server work done as that'd need something similar for the import+signature and then probably a few more days of work to generate the right "fake" updates [16:47] Just saying that it's kind of orthogonal to whether you're on ubuntu-system or not [16:48] * mhall119 nods [16:49] stgraber: why would you need to generate a bunch of big updates? [16:49] jdstrand, I feel your pain :-) [16:49] stgraber: install version n-1; set /etc/ubuntu-build to 0; upgrade [16:49] slangasek: because what's failing for gema is not a full update, it's a massive amount of mid-size updates [16:49] stgraber: no, the path from 5 to 70 is certainly going to be a full update [16:50] unless she's done something to override [16:50] slangasek: she's testing with -5 [16:50] slangasek: which means 5 versions before the current one and that's most likely to be a delta path, not a full [16:50] oh? [16:50] then I misunderstood [16:51] slangasek: I am testing from 57 to 70 [16:51] slangasek: I edited the bug to make it clearer [16:51] slangasek: just big enough delta - 1 [16:51] stgraber: are you sure the upgrade path from 57 to 70 pulls deltas, not a full update? [16:51] ah, ok [16:51] slangasek: I am positive [16:51] testing a full update is really quite simple indeed, just change channel.ini to force a lower version, run the code again, done (or use system-image-cli -f full for a cleaner solution). Simulating a case where you need to download 25 small files in parallel is tricky (and that's the case that's been failing for me here with weird timeouts) [16:51] right, I completely misunderstood the problem, sorry [16:52] so yeah, we need to test both of those cases [16:52] slangasek: well, since they wrote a script to tell you which version to take to get the biggest possible delta, I sure hope so ;) [16:52] slangasek: http://bazaar.launchpad.net/~javier.collado/+junk/upgrade-tests/view/35/scripts/full-update.py [16:52] but we *definitely* need to test the "Multiple small updates" case using the new code, and not the code that was in 57 [16:52] I think the right way to do this is with a synthetic channel [16:52] oh, actually, faking the version number would work in either case [16:53] true [16:53] because if you're on rev 71 and change channel.ini to say it's 57 [16:53] it'll resolve a path using a bunch of deltas === dandrader|lunch is now known as dandrader [16:53] which will just end up replacing files you alrady have, no big deal, same end result [16:53] so that'd work [16:53] yep - but the deltas may have different characteristics from one version to the next, that doesn't make for a very reliable test case [16:53] how so? [16:54] we really want the integration tests for the *updater* to be based on fixed data [16:54] stgraber: one day, the max deltas to download might be 3; another, it might be 6; how will the behavior change, and what are we trying to test? [16:55] if it's a problem with parallel downloads, we don't want to overlook a regression because a particular day's delta was large and cause a smaller number of parallel downloads [16:56] right but doing that needs a separate server with fixed size pointless deltas that are properly signed and trusted on the device [16:56] that's definitely possible and moderately easy to do once we have the oem/port code implemented in the server side code, but don't count on it for at least another 2 weeks to a month [16:57] so it may be better to have a slightly less optimal testcase today than try and aim for a perfect one and run with none until then [16:58] stgraber: jcollado is working on that [16:58] stgraber: not sure if he has spoken to you yet or not [16:58] stgraber: we shouldn't spend a lot of energy on a test case that's not going to test the code we care about [16:58] stgraber: I am not even sure he is aware of this rearchitectured version of the upgrader [16:59] we care about testing that upgrades *will* work from the current version of the image, before we promote it... because that's the point where we have the opportunity to fix bugs [16:59] learning that there were bugs in an image from two weeks ago really doesn't help us ;) [16:59] gema: the upgrader hasn't change in the past month, the client and server did (sorry, I'm picky about terminology but it's easy to misunderstand what we're talking about if we're not using the right terms) [16:59] gema: and yeah, he talked to me a bit a few weeks back, haven't heard much since [16:59] stgraber: you are well right to be picky, I don't really understand what changes you guys are talking about [16:59] and while installing s-i + u-d-m on an old image and testing that is better than nothing, it's not a true integration test - let's make sure we're targeting the right thing [17:00] slangasek: that's why using the latest image and faking channel.ini seems reasonable to me [17:00] as a first pass, yes [17:00] slangasek: so you don't care that you can upgrade from a week ago to today's update, you care that today's image can upgrade to future, or do you care about both? [17:00] slangasek: that'll have the current code of the latest image try to do a full update and a large series of delta updates at pretty much no implementation cost [17:01] gema: I care about things I can fix; if the upgrader in the already-released image is broken, I can't fix that [17:01] slangasek: because if you produce an upgrade that is broken somehow, or that is not backwards compatible, you want to know [17:01] slangasek: the upgrade may be wrong/broken [17:01] dpm, hi.... do you know where the winners of the app showdown are going to be announced? [17:01] slangasek: well, we can fix an upgrade path from an old image on the server side. === alan_g is now known as alan_g|EOD [17:02] gema: so integration testing of the *upgrader* should test from the current upgrade with a mocked channel.ini; integration tests of the *upgrades* should test with a previous image using the real channel [17:02] slangasek: if today's upgrade renders millions of phones unusable for whatever reason, you definitely don't want to promote that upgrade [17:02] slangasek: if we notice that a particular delta path will fail, we can break it and force a full for those. So we should definitely test that so we can know about it before we get user reports. === bfiller is now known as bfiller_afk [17:02] slangasek: agreed [17:02] slangasek: so we are talking about two things at once [17:02] gema: and I thought what we discussed previously was to test upgrades from n-2 to n-1 and n-1 to n. What's driving the test from n-5 here? [17:03] slangasek: n-2 is not good enough to trigger a full upgrade and I wanted to test that [17:03] slangasek: automation is not in place yet, so I am doing manually [17:03] stgraber: I'm not happy for us to be hands-on managing that on the server. We should be setting a high bar for the reliability of the upgrader, which makes this issue moot [17:03] sil2100: ping [17:04] gema: right, but if you want to trigger a full upgrade, just install n-1 and set /etc/ubuntu-build to 0 [17:04] slangasek: sure, I still think it's worth knowing if we messed up so we can do something about it rather than panic when we get users report after releasing the new image [17:04] slangasek: I am seeing problems that only happen if the ro partition hasn't been touched [17:05] slangasek: so doing that is not a real life scenario, which is what I am trying to test [17:05] gema, slangasek: /etc/system-image/channel.ini, not /etc/ubuntu-build, /etc/ubuntu-build is just there for backward compatibility, it's not read by the client at all! [17:05] gema: there's nothing magical about the filesystem that it knows when it's touched [17:05] slangasek: I am sitting here with shrodinger's cat [17:05] stgraber: ah, ok [17:06] slangasek: I know, still bugs stop being reproducible after that [17:06] slangasek, stgraber I am being called for dinner, will have to continue this conversation tomorrow, I will read backlog anyway [17:08] gema: well, testing with versions of code that we don't care about and don't tell us anything about the current code is not a good workaround for the fact that some bugs aren't reproducible if the filesystem has been touched. I'm happy to help root-cause *those* bugs if it would help, but we still need the test I'm describing to answer the question of whether the upgrader in the to-be-promoted version works [17:08] gema: ok, buen aprovecho :) [17:10] slangasek: I agree with you, we need that testing as well and I am going to make sure jcollado is on the same page tomorrow as well [17:10] slangasek: thanks! [17:12] gema: hi. still need help? [17:27] popey, Is there a bug tracking the alarm stuff? [17:29] Hello all, I've just installed saucy touch nightly (downloaded today) on my nexus 4. I'm unable to get a data connection over 3G. Calling works fine. I've taken a snip of /var/log/syslog showing Network Manager trying to connect but failing. It's here http://pastebin.com/GJ52g5J8 [17:30] ogra_, stgraber: did image 75 happen yet? [17:31] davmor2, not yet [17:31] nik90, ^ question for popey above, do you know? [17:32] tedg: can you copy paste the question again? [17:32] tedg: I am viewing through office irc account [17:33] nik90|Office, Is there a bug tracking the alarm stuff? [17:33] atomic0x: i believe that should be fixed with the next image [17:33] tedg: https://bugs.launchpad.net/ubuntu-clock-app/+bug/1233176 [17:33] Ubuntu bug 1233176 in Ubuntu Clock App "Alarm notifications do not appear when an alarm is triggered" [Critical,Triaged] [17:33] nik90|Office, Great! Thanks! [17:34] charles, ^ [17:34] tedg: although that bug is localised to the clock app..so you may want other bugs in respective projects for other stuff like indicator-datetime and sdk [17:35] Excellent. I'll try again in a day or two :) thanks! [17:35] nik90|Office, I added bug tasks, I think it's close enough. [17:35] tedg: ah perfect. thnx [17:36] tedg: in your MP, why does it say .app_id = "calendar-app" instead of clock-app? [17:37] nik90|Office, Cut and paste bug :_) [17:37] :-) [17:37] nik90|Office, Fixing [17:37] ;) [17:38] tedg: also fyi for the clock app, the id is applicationName: "ubuntu-clock-app" [17:38] that's how it is defined in the main view [17:38] I am not sure if that helps or not [17:38] nik90|Office, What is the name of the desktop file? [17:39] tedg: ubuntu-clock-app.desktop [17:39] tedg: it executes Exec=qmlscene /usr/share/ubuntu-clock-app/ubuntu-clock-app.qml [17:39] nik90|Office, Yup, found it on my phone. [17:43] om26er: ping [17:44] cwayne, pong [17:46] om26er: hey, your touchshot script is pretty useful, any plan on including it in phablet-tools? [17:46] cwayne, I would but its not going to work in a few days as we move to Mir :/ [17:46] screencap is SF specific [17:46] om26er: ah, i gotcha === jhodapp is now known as jhodapp|lunch [17:49] bfiller_afk: when you have some time please email me the next priorities. What remains from your last email is either waiting on something/someone or almost done, so it would be nice to have some other things to fill the gaps and to start on next week. Thanks [17:51] Is there anyone online xD? [17:52] no [17:53] only robots [17:53] some have attained self-awareness though [17:53] call me robot? [18:07] stgraber: will images in /stable be safe from pruning? [18:08] mfisch: no [18:08] mfisch: /stable is an alias for /saucy, /saucy contains up to 10 images then prunes the old ones [18:09] stgraber: okay, thanks [18:10] dobey: isn't that the start of Terminator? [18:11] dobey: or the matrix [18:12] davmor2: and all the religious books in the world. [18:14] dobey: Pretty sure that god is a robot whose been dealing with 3d printing for a while :D [18:20] dobey: btw updated the MR with more info === cwayne is now known as cwayne-afk [18:26] build 75 done ... [18:27] ogra_: need testing? [18:27] we wont release it tonight ... [18:28] but yeah, go ahead if you feel like [18:28] ok, will do [18:28] TZ selection should work now [18:28] oh, good good [18:28] and the breakages should be gone again [18:28] * ogra_ crosses fingers [18:28] next step is to get mir on it, i guess [18:32] ogra_: meh I just got 74 again [18:33] popey: check it's 75 you get [18:34] cwayne-afk: that also doesn't solve the problems [18:34] yip yip [18:34] davmor2, cdimage is done, system-image might still be processing [18:35] ogra_: right [18:35] yeah, its still grabbing previous image [18:35] phablet-flash should report the image number [18:35] imo [18:35] file a whishlist bug [18:35] popey: +1 [18:35] then pause 3 seconds so you can CTRL+C it ☻ [18:36] popey: better still would be a confirmation to continue [18:36] hmm, i prefer no confirmation [18:36] figlet build number 74 [18:36] like that [18:36] so i see it and then stab ctrl+c ☻ [18:37] popey: agreed plus it is one less thing to automate if you run the script automagically [18:40] bug 1233800 [18:40] bug 1233800 in Phablet Tools "phablet-flash should report human readable image before flashing" [Wishlist,New] https://launchpad.net/bugs/1233800 [18:44] popey: +1'd [18:44] why are so many of the icons not on a square canvas? === jhodapp|lunch is now known as jhodapp [18:47] ogra_: looks like the import is running at the moment [18:47] stgraber, yeah, i was suspecting that [18:49] I'm trying to run an app (just a "Simple Touch UI" template from Ubuntu SDK) on a phone. Have you guys seen this message? [18:50] mkdir: cannot create directory ‘/home/phablet/.local/share/applications’: Permission denied [18:52] davmor2: got news for me? :) [18:53] I can't seem to find a way to break it so far [18:53] cyphermox: just waiting on system image to catch up [18:53] cyphermox: then I have 75 and can test [18:56] ogra_: any idea how long system image should take to play catchup at all? [18:57] ogra_: ignore that looks like it's there now [18:57] * popey cancells the bug and just gets davmor2 to check the status of system image for me [18:58] popey: tsk I might be on holiday then what do you do? [19:01] ricmm, sergiusens (or someone else non-English native), do you guys translate "WEP Enterprise" in UIs? [19:02] tedg, my env is in English :-) [19:02] cyphermox: as soon as image 75 is installed I'll get your mod installed and leave it on 3g till tomorrow, if it's still connected you win, if not you LOSE ;) [19:03] sergiusens, ah, that's no fun! [19:04] cyphermox: that will hopefully give it enough time that if the connection is gonna drop it will of dropped :) [19:08] davmor2: get your phone number from the directory ☻ === bfiller_afk is now known as bfiller [19:08] stgraber: are there any examples on s-i.u.c where an image has a phased-percentage key? [19:11] ogra_: What was the URL for the changelogs? I've seem to lost it somehow [19:11] barry: not at this point, I actually still have to implement the feature server side (though I expect it to take 30min at most, it's pretty trivial on my side) [19:11] cwayne-afk: pong [19:12] stgraber: okay. so just so i understand how it's going to work: only an image with the highest version number can have a phased-percentage key, and if that exists and it is < hash, we essentially treat that image as not existing when we calculate all the upgrade paths [19:13] Hey everyone ! Do you know if we have a way to know the header height of the tabs ? [19:13] When I ask the height of an element, it always give me the total height of the screen, not the height without the header [19:13] barry: correct [19:13] ogra_: worrtingly I have mtp conected but a black screen [19:13] ogra_: nevermind it finally woke up [19:13] but i won't enforce that "must be on the highest version", so if for some reason an intermediate image has a non-matching phased-percentage, i'd ignore that image but it probably wouldn't matter as the next delta would get you past that anyway [19:14] stgraber: ^^. if so, i think my logic is pretty simple [19:14] barry: yep, that's perfectly fine. On the server side, I'll change the function that actually writes and signs the index.json to strip phased-percentage from all images but the latest, so we'll never have that case in production anyway [19:15] (and .set_phased_percentage() will fail on older images too so people won't get confused when the flag won't stick ;)) [19:19] stgraber: +1 [19:26] ogra_: still no incoming call tone === rpadovani is now known as WebbyIT === Dazzozo` is now known as Dazzozo [19:34] hello hashcode [19:36] I just wounder how the hdmi driver is comming along? === dandrader is now known as dandrader|afk === Kyle_ is now known as Kyle === popey_ is now known as popey === tsimpson_ is now known as tsimpson [19:48] does anyone know how mir compares to x? (speed, security, etc) === dandrader|afk is now known as dandrader === kdub is now known as kdublunch === josepht_ is now known as josepht [19:58] tedg, mhall119, know anything about locales? [19:58] looking at setting a search path for them [19:58] sergiusens, tedg, my env is in English :-) [19:59] Search path like where the .mo files are? [19:59] tedg, I know, that's why I don't know how to do this :-) Turns out the apps have translations now and I don't know how to 'find' them from the click pkgdir [20:00] tedg, if it's an envvar, can we com up with a convention in upstart-app-launch? [20:01] Yeah, I think that would make sense. [20:01] yay time/date setting works now! [20:01] Perhaps it would make sense to put in the manifest. So, in theory, people could change it. [20:02] I've got a TEXTDOMAINDIR as an env. [20:02] Oh, I bet seb128 knows. [20:02] tedg, that works for me ... jdstrand || cjwatson ^^ ? [20:02] tedg, what? [20:02] sergiusens: what about locales? [20:02] seb128, Trying to figure out locale stuff for click packages. [20:03] seb128, Is there an env we should set so they know to look in their install directory? [20:03] mhall119, just figuring out how to find them from a click pkgdir [20:03] tedg, I don't think so, for langpacks we patched libc [20:03] seb128, It turns out that even though he speaks non-English, sergiusens doesn't use it. [20:03] sergiusens: wouldn't they be in ./po/ in the click install dir? [20:04] seb128, Oh, wow. I didn't realize it went that far. [20:04] mhall119, and that _just works_ TM? :-) [20:04] sergiusens: of course, because it's built by a team of talented engineers who wouldn't date ship broken code [20:05] translation: I have no idea if it works, but it should [20:05] I don't think you want po files, you want mo files. [20:08] Uhm... eglibc is the most complex packaging I think I've seen. [20:08] mhall119, i have a rough WIP implementation of content transfers initiated by an exporter [20:08] mhall119, no promises on this landing for 13.10, this was deferred [20:09] tedg, are we discarding TEXTDOMAINDIR? [20:09] mhall119, but i wanted to spend a little time on it now to see how the API would behave, it might be doable for 13.10... [20:10] tedg: given that we likely won't have content-hub doing what we need for docviewer-app before 13.10, can I submit an MP to url-dispatcher to do the work in the mean time? [20:10] kenvandine: thanks [20:10] kenvandine: remember that 13.10 is in 2 weeks :) [20:11] sergiusens, Looking at the eglibc patch... seeing if that's right. [20:11] sergiusens, Not looking good. [20:11] mhall119, That's fine. Just know it'll break when we switch to registering URL types. [20:11] tedg, I'll send an email to the list to see if someone has more ideas [20:11] tedg, clearly not for 13.10, but maybe url-dispatcher could support finding the app that owns content based on a file URI and invoking it to initiate an export [20:12] so filemanager-app could just use url-dispatcher, which would query the hub for a handler to start a transfer [20:12] kenvandine, Not sure I'd like to complicate url-dispatcher for a corner case like filemanager.... [20:12] tedg, note currently we have no way to query to find what app owns a file... but maybe we need to think about that [20:13] kenvandine, Seems like filemanager should be complicated in itself. [20:13] kenvandine, remember file manager wont even be on the consumer devices [20:13] kenvandine, That way if most people don't use it, they're not encumbered by the complexity. [20:13] oh true... i guess filemanager might be the only app to use it that way [20:13] pmcgowan, understood [20:15] sergiusens, Looking around I don't see anything. [20:16] tedg, ack, I just sent a bunch of questions to the appstore list [20:16] sergiusens, Perhaps look at how the 0install folks do it. [20:16] tedg, 0install? [20:17] oh, maybe typo :-) [20:17] sergiusens, http://0install.net/ [20:17] tedg, well I know cwayne-afk and mfisch sent an email with customization and stuff, they may have looked at it already [20:20] sergiusens, In Inkscape when doing relocatable binaries we change what we pass to bindtextdomain() [20:21] jhodapp: mediaplayer seems to be exhibiting different behavior on mako vs. maguro. It is failing 1 test on maguro with build 75 (passed on 73, 74) but on mako has been failing 4-5 tests on all 3 of the latest builds [20:22] tedg, hmmm, bindtextdomain can work for compiled apps [20:24] plars, thanks for the heads up...I don't have a mako but rsalveti does [20:24] sergiusens, patch qmlscene ? [20:25] jhodapp: if time allows, will try to do some more retries, and try locally too, but trying to get through a lot of the other tests right now also [20:25] tedg, to take a param with the domaindir and domainname? [20:25] jhodapp: I did retry once with the same results [20:26] tedg, if there is no better option this is as good as it gets [20:26] sergiusens, No, to set bindtextdomain with the correct value depending on the env. [20:27] jhodapp: ah, there's a .crash file on both of them too: http://reports.qa.ubuntu.com/smokeng/saucy/touch_ro/4512/mediaplayer-app-autopilot/ [20:27] plars, could you send an email to rsalveti and myself detailing which tests are failing and the output of them running with GST_DEBUG=*:3,*amc*:5,mirsink:5 as well as the logcat output? [20:28] jhodapp: there's a link to the logcat and a list of the failing tests at the link I just posted, I'll need a bit before I can try it with the debug settings though. Where do I set those? [20:29] tedg, I'll branch the qmlscene code (which is probably in a huge code base :-/) and see what I can do [20:30] plars, just set it on the command line before the test's executable name [20:30] sergiusens, Cool, check for UBUNTU_APPLICATION_ISOLATION=1 and then use the current working directory. [20:30] plars, for example: GST_DEBUG=*:3 ./my_media_tests [20:31] tedg, current working directory + rel path or just a hardcoded locale dir? [20:31] jhodapp: the tests start the mediaplayer themselves though, don't they? [20:31] sergiusens, I imagine "$cwd/locale" would be best. [20:32] jhodapp: so I don't think setting that on my side is going to get that downstream to the app [20:32] tedg, nah... -translation ............ Set the language to run in [20:32] sergiusens, ? [20:32] tedg, that already in qmlscene [20:33] How useless [20:33] I guess qmlscene was only ever built as a demo. So I shouldn't give it too hard a time. === kdublunch is now known as kdub [20:34] tedg, we were never supposed to use qmlscene, not sure why it was kept as the way to go [20:34] sergiusens, The Europeans are sleeping, let's blame them! [20:34] :-) [20:35] The qmlscene utility is meant to be used for testing your QML applications, and not as a launcher in a production environment. To launch a QML application in a production environment, develop a custom C++ application or bundle the QML file in a module [20:35] plars, it will, that's how I run the mediaplayer [20:35] that's from http://qt-project.org/doc/qt-5.0/qtquick/qtquick-qmlscene.html [20:35] plars, gstreamer is in-process to the mediaplayer [20:35] sergiusens, upstream is working on something new and shinier, so we didnt want to duplicate that effort [20:36] jhodapp: so I guess you aren't using phablet-test-run right? [20:36] pmcgowan, makes sense [20:36] dobey, i'm sorry to keep harping on this, but i don't understand the issue int he icon-fix MR. it's following the same convention as the other ubuntu-mobile apps, and we really need this icon fix for when we demo this [20:36] tedg, can we reuse that or should I still patch qmlscene? [20:37] sergiusens, We can't reuse that, we'd have to do too much special casing. [20:38] tedg, ack, I'll work on a patch [20:40] plars, no I'm not [20:41] cwayne: if other apps are doing it, then they are doing it wrong too. [20:41] cwayne: but i'm not blocking that branch afaik [20:42] dobey, that's where every other app's icon lives. is literally every single app doing it wrong? [20:43] dobey, it still needs top-approval, not sure who would do that? [20:43] balloons: ping [20:43] m-b-o, howdy [20:44] I'm fighting with a testing the updated removable list ite, [20:44] m [20:44] balloons: swipe to delete, where you have to click the trashcan icon [20:45] cwayne: i think you're overstepping the amount of exaggeration you need to provide in your argument. [20:45] cwayne: and the mobile apps are supposed to be migrated to click packages anyway, not remain debs, so yes, they would be doing it wrong in that respect as well. [20:47] cwayne: i don't know who all has privileges to set the global approve on those branches. sil2100 can at least. but i'm not going to. :) [20:47] m-b-o, ohh, what's wrong with it? [20:48] balloons: how to write an autopilot test for it [20:49] barry: if you want to confirm that your parser is correct, I've marked image 75 as phased at 50% in saucy-proposed [20:52] m-b-o, well swipe as usual, and if you have to confirm it, then do so [20:52] I think the clock and rss reader both do that [20:52] are you stuck on a piece of it? [20:53] balloons: it's acm ewith an update som e hours ago, so I dont' know, if we are talking about the same [20:54] jhodapp: plars: I'll take a look, was planning to do that today anyway [20:54] I know some will indeed fail [20:55] as we're using the hw accelerated based media stack at today's image [20:55] ah [20:56] plars: we're working on the remaining issues, but we decided to push it yesterday as that would finally bring the missing feature [20:58] balloons: https://code.launchpad.net/~martin-borho/ubuntu-weather-app/LocationLookup === davmor2_ is now known as davmor2 [21:09] stgraber: cool, thanks [21:13] stgraber, image 75 looking *much* better btw [21:13] m-b-o, can I have a look at this tomorrow? are you needing to land the branch now? [21:14] m-b-o, I see the diff. You do the drag and then click the trashcan [21:15] ballons: yes, and somehow it's going ahead too fast [21:15] or the click gets somehow dismissed === bfiller is now known as bfiller_afk [21:15] m-b-o, ok let me just visually see it :-) [21:15] one sec [21:16] m-b-o, I got brokenness off the bat [21:20] Laney, heya! timezone worked when set from cli with the gdbus call! although still not from system settings for me [21:20] slangasek, stgraber I'm having a bit of issues with the upstart-local-bridge; on the android side I have around a 100 events and on the local bridge on the ubuntu side I only get 4 [21:21] sergiusens: I don't know this code at all; your best bet is to ask jodh when he's around during the UK workday [21:21] slangasek, ok, I'll ask him early tomorrow [21:21] sergiusens: I can try to reproduce here the issues you're seeing and maybe take a shot at debugging, but I'm coming in cold [21:22] slangasek, no worries, I'm waiting for salveti to apply my patches so the bridge is setup on boot so tomorrow jodh can have an easier debug session [21:26] barry: note that those will automatically get flushed when 76 is released (unless I messed up the code, but I'm pretty sure that's tested ;)) [21:27] stgraber: not to worry. i now have test data with that value, so i'm relatively confident that the live site can be handled [21:29] so, I saw someone mention the other day this blank 'update manager' icon... anyone know what that's about? I can't imagine that we're meant to have update-manager on the device, and package lists confirm it's not there. [21:34] slangasek: not idea, but I've seen that icon on my devices for a few days now [21:36] slangasek, stgraber that is the click app updater app [21:36] ah [21:36] WIP then? :) [21:36] there should be a fix for the icon in the next promoted image, and a fix for it to "actually work" in the next image [21:36] slangasek, well, I would say "bugs" more then WIP [21:37] I suspect it will be fixed in 76, which I suspect will be promoted tomorrow [21:37] ok [21:37] and image promotion is still all manual overriding of test failures? [21:37] slangasek, maybe we should rename it to "app updater" or something? [21:37] slangasek, aiui, image promotion involves didrock's dipping the image in some kind of holy water [21:37] rickspencer3: might be good to avoid confusion with the unrelated update-manager command on teh desktop [21:38] rickspencer3: right, so still manual; my understanding was that the goal is to have images passing all tests on a standing basis so that they can be autopromoted, I wonder how close we are to that [21:38] slangasek, I'll log that bug, but I don't know what to log it against! [21:38] heh [21:38] slangasek, we are not close [21:39] and when we turn on Mir, I assume we will get further from that [21:39] apparently the package name is 'click-update-manager' [21:40] slangasek, https://bugs.launchpad.net/click-update-manager/+bug/1233863 [21:40] Ubuntu bug 1233863 in Click Update Manager "app should be renamed to avoid confusion with system updates" [Undecided,New] [21:41] rickspencer3: coolio [21:43] sergiusens,tedg: Patching eglibc to handle this in click packages would be pretty insane I think. Setting TEXTDOMAINDIR seems reasonable. How about doing it in aa-exec-click, which already sets various other environment variables? [21:43] slangasek, rickspencer3 shouldn't that live in settings too? [21:43] sergiusens,tedg: I don't like the idea of having to pass extra arguments when there's already a perfectly good environment variable. [21:43] sergiusens, it's a super temporary app [21:44] rickspencer3, ack [21:44] the functionality will be taken over by the apps scope asap [21:44] [systemimage] Oct 01 21:43:53 2013 (18303) Upgrade path is 17:18:19:20:21:22:23:24:25:26:27:28:29 [21:44] barry: that's quite an update path ^ [21:44] cjwatson, I don't see that variable get used in eglibc [21:44] cjwatson, sure, I'll confirm that TEXTDOMAINDIR works [21:44] tedg: I was going off the conversation above; I didn't check the name but I'm pretty sure one exists [21:44] stgraber: indeed. turn on verbose logging for a full dump of the candidates and scores [21:44] cjwatson, the next thing would be to see if we want a hardcoded location or something we can setup in the manifest === salem_ is now known as _salem [21:45] cjwatson, I thought so too, but I couldn't find one with grep :-/ [21:45] barry: oh, I'm sure it's right, I was just surprised to see one that long ;) [21:45] Maybe I'm remembering a conventional automake variable or something instead ... [21:45] And it seems that most projects set it in their build. [21:46] barry: that's devel-proposed-customized which gets quite a few more updates than devel-proposed most of which are tiny [21:46] * barry nods [21:47] tedg: Hmph. Maybe you're right. :-( [21:48] tedg: So what calls bindtextdomain in a qmlscene app, if anything? [21:48] Something in Qt I suppose [21:48] I'm assuming it uses gettext rather than being wrong in some different way [21:49] Heh, I'm not sure I'd assume that :-) [21:49] Oh God, .ts files [21:49] I hope we're not using those [21:50] It'd be nice if we could find a way were, for instance, GEdit could repackage as a click and things would generally work. [21:50] Not sure if that's possible, but I think it'd be a good goal. [21:50] Way out of scope for anything I care about [21:51] And I think it's the wrong goal - focus on the things we can't already do rather than furtling about with the things we can [21:51] (I understand where you're going but I think this is a fundamental misfocus that we're prone to) [21:51] https://bugreports.qt-project.org/browse/QTBUG-2404, grr [21:52] Sure, and I understand your point. It just seems like "something that should work" [21:52] So what are we doing right now? [21:52] I mean, do we have qmlscene-ish translated apps at all, even non-click? [21:52] Not sure, sergiusens ? ^ [21:52] tedg, cjwatson ... yes, dpm made me look yesterday... he has a rule that looks like this http://paste.ubuntu.com/6181462/ [21:53] cjwatson: clock has translations.. https://translations.launchpad.net/ubuntu-clock-app [21:53] *targe in rules [21:53] We could LD_PRELOAD bindtextdomain with a thing that tries the current directory first [21:53] You may vomit now [21:53] cjwatson, so it's mo [21:53] sergiusens: Is that in ubuntu-clock-app or something else? [21:53] ./debian/ubuntu-clock-app/usr/share/locale/en_GB/LC_MESSAGES/ubuntu-clock-app.mo [21:53] Since I'm more interested in the app's setup code for it [21:53] has lots of mo files [21:54] cjwatson, that's actaully the clock, yeah [21:54] cjwatson, but all apps have translations [21:54] cjwatson, I think patching eglibc makes more sense than LD_PRELOAD ;-) [21:54] sergiusens, heyo, i made phablet-config work with the new (right!) way to change timezones: https://code.launchpad.net/~cwayne18/phablet-tools/phablet-config-timezone-fix/+merge/188727 [21:54] tedg: Patching eglibc should be a last resort - high bar [21:54] Maybe LD_PRELOAD is a laster resort :-) [21:55] cwayne, goodie [21:55] popey: Hm. So, all well and good, but what loads them? [21:56] cjwatson: I think we recommend using the gettext bindings for QML apps [21:56] Has this been seen to actually load translations? :-) [21:56] cjwatson: congratulations on reaching the limit of my knowledge on the subject. have a cookie. [21:56] Heh [21:57] i18n.tr() calls [21:57] Oh, so it's in the UI toolkit [21:57] interesting [21:58] pmcgowan, do you know how the translation features are supposed to work? [21:58] sergiusens, not exactly but I thought it was pretty transparent [21:58] So we could make the UI toolkit do this pretty easily [21:58] whats the issue? [21:58] sergiusens, this was worked out some time ago bit I dont have the details [21:58] pmcgowan: Never mind, I've found the code now [21:59] cjwatson: that's *if* apps are using i18n.tr() calls, which is recommended, but I'm not sure they are doing [21:59] pmcgowan, just loading translations from non conventional paths as a click install path for a package ;-) [21:59] lool: clock is, anyway [21:59] cool [21:59] lool, pretty sure they are [21:59] So there's an i18n.bindtextdomain interface to set the domain, but clock doesn't use it [21:59] pmcgowan, what about ours? [21:59] core apps team and our own guys do [21:59] pmcgowan, good [21:59] So it'll just get the default [21:59] we had them all showing chinese at some point [21:59] i.e. we can't assume that right now core apps are doing something we can easily override for click [22:00] cjwatson, hmm, just found this https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1220395 [22:00] Ubuntu bug 1220395 in ubuntu-ui-toolkit (Ubuntu) "translations are not updated after calling i18n.bindtextdomain" [Undecided,New] [22:00] it's click + translations [22:02] cjwatson: I just did a quick grep to see whether we were calling bindtextdomain by default and it seems not; the i18n plugin isin ubuntu-ui-toolkit/modules/Ubuntu/Components/plugin/i18n.* [22:02] And https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1233071 [22:02] Ubuntu bug 1233071 in Ubuntu UI Toolkit "Set up the translation domain automatically for click packages" [Undecided,New] [22:03] That looks like this is in progress [22:03] So I'm inclined not to try to dive in :) [22:04] cjwatson, yeah, too bad it seems we are in silos at times [22:04] In general it's not like you can take an arbitrary program configured with prefix=/usr and relocate it to be a click package anyway; takes work [22:05] yeah, i wanted to clickify irssi for my phone but it looked like a lot of work ☹ [22:05] cowsay on the other hand... [22:06] yeah, not that much once you have all the workings in your mind [22:06] translations was new to me though [22:06] Certainly if it doesn't touch the filesystem then it's trivial, but not the general case [22:06] Even say man-db isn't going to work if you just shove it into a different shape on the filesystem since it knows about paths to bits of itself [22:07] Common [22:07] heh, luckily man doesn't have a reason to be in clicks... yet [22:16] sergiusens: I'm using it as an example of a not fundamentally all that complicated program that isn't relocatable, to try to deflate the implicit assumption I see going around that it might be straightforward to relocate existing packages without rebuilding. [22:18] (And, as I say, I think it'd basically be a waste of effort better spent elsewhere) [22:20] cjwatson, I'm not sure I follow, is that in general or specific to these apps? [22:20] What apps? [22:20] I'm off on a tangent, not talking about translation of core apps :) [22:21] Never mind [22:21] cjwatson, ok, then I did follow perfectly fine ;-) === mspencer is now known as iBelieve [23:31] having some difficulty getting the sdk setup properly. I dont see any "ubuntu touch" project templates. any thoughts?