[00:00] slangasek: yeah, we can't run all the tests at MP time, it takes waaaaay too long [00:00] infrastructure issues that we can solve by throwing 10,000 emulators at it? [00:00] so we do it pre-release [00:00] right... but then that's not continuous integration [00:00] slangasek: I don't know, sorry. veebers, can you provide more details maybe [00:00] it's.. integration by parts [00:00] slangasek: yeah [00:00] (or, insert your preferred corny calculus joke here) [00:01] slangasek: we've noticed the problems around the way we used to release things, we're being much more rigerous, and part of that is not releasing trunk after every merge [00:02] hmmm [00:02] so we'll try this, and if it solves our problem, then we'll run with it. Otherwise we'll try something else [00:02] that seems like a perverse outcome to me; I want the rigor, without reduction in release frequency ;) [00:03] yeah, me too, but it's just not practical in this instance [00:29] Hi thomi, slangasek: The infrastructure regarding the release is the jenkins jobs that do the process are erroring [00:30] slangasek: throwing emulators would potentially assist in the speeds running the tests, so would benefit the release indirectly [00:30] veebers: anything I can do to help? Are the right people working on fixing whatever the jenkins problem is? [00:30] (can you give me a pointer to the exact jenkins failure?) [00:31] we currently run autopilot test suite at MR time, it's running all the touch app acceptance tests that take hours (so we reserve that for the release candidate) [00:31] * slangasek nods [00:32] slangasek: Not sure if you can help, it's the CI-Train process which is pretty recent, didrocks is my only point of contact at this point [00:33] veebers: so do you have visibility on the actual jenkins failure? [00:34] slangasek: just looking now, I believe the actual issue lies with the CI-Train code, not jenkins or canonistack concerns. Link coming right up [00:34] slangasek: this is the latest error: http://162.213.34.102/job/landing-002-1-build/1/console [00:37] veebers: heh, ok [00:37] slangasek: thanks for the offer. I'll keep you in the loop re: release and potential improvements re: testing speeds and times [00:38] veebers: I guess this particular blockage will be the focus of quite a lot of attention between now and the end of next week, given the core team sprint that's about to happen :) [00:39] so I'll cool my heels for now and worry about other things [00:39] Has anyone with a N4 updated to r143? The OSK doesn't seem to come up. [00:57] vthompson, known issue, there's some missing packages in the build [00:57] AIUI, should be fixed in 144 [00:58] cwayne, ok. Thanks for the info! [00:59] vthompson, no problem! === jackson is now known as Guest1629 === jackson_ is now known as Noskcaj === chihchun_afk is now known as chihchun [04:04] cyphermox: hey! around? === jackson is now known as Guest7088 [04:14] good tut on ho to install on a nexus 7 2012 [04:33] Could i get the exact kernel source git tree that was used to generate: trusty-preinstalled-boot-armhf+maguro.img === duflu_ is now known as duflu === vying is now known as Guest99684 [08:12] good morning [08:38] To any of the KitKat porting team in here, congrats on the successful port, even if it does spell the end of support for my phone! [08:47] ogra_, is there a way to add a README directly under cdimage.../current, to describe the various img and zip fileswhich image [08:47] s/which image// [08:47] ogra_, I am writing a wikipage about it, but having it from the horses mouth would guarantee people find it easier [08:47] yes, but we should rather do it cleanly and add that to the header [08:48] ogra_, the header is fine to, although it has quite some blurb and am not sure people ever read that [08:48] janimo, btw, i have "rootstock-touch" building a rootfs locally (armhf on x86 only yet, but will add x86 on x86 too) [08:49] ogra_, nice. I got live-build to do it eventually too [08:49] (using the livecd-rootfs/live-build setup) [08:49] there is one last issue with the click mirror ... but i'll solve that today [08:50] ogra_, this is what I manages, hopefully yours is more complete though [08:50] https://github.com/janimo/ubuntu-touch-live-build [08:50] janimo, well, if you write a wiki you can add "read the header at ..." :) [08:50] not complete at all yet, but io plan to complete is :) [08:50] *it [08:51] ogra_, indeed, that is why I wanted it somewhere read-only and more visible, since the wikipages are prone to deletion and get rarely read :) [08:51] ogra_, so where can that header be modified? [08:51] i use the original config, no chnages [08:51] janimo, cdimage tree [08:52] (you seem to fiddle with the live-build config in yours [08:52] ) [08:53] ogra_, I found no way to tell it to cross-build for arm [08:53] to have the ARCH env var get used [08:53] qemu-arm-static [08:54] in stacked chroots [08:54] ogra_, I added that later, so maybe the initial one is no longer needed indeed [08:54] ogra_, actually I had to pass -a $ARCH when building i386 on amd64 [08:54] that works withouth qemu [08:54] but needed arch or it defaulted to ubuntu/amd64 [08:55] well, your outer chroot should already use qemu-arm-static ... [08:55] LB_BOOTSTRAP_QEMU_STATIC only copies the binary around [08:55] it doesnt install the package [08:55] ogra_, you mean running the build script inside a manually entered chroot? [08:56] wekll, i have my script set up a qemu.arm-static chroot ... install livecd-rootfs and run the build inside that [08:56] (same setup that we have on the buildds) [08:57] ogra_, I just used live-builds poorly documented qemu build abilities. I had not tested the resulting image yet [08:57] so no nested chroots, or at least not obvious when running the script [08:58] right [08:58] i want the same env the buildd has ... [08:59] anyway, i hope to integrate that later with stephanes system-image creation tool once he has a community variant [08:59] so that you can get a complete image [09:06] janimo, btw, you want to change the text around line 739 in lib/cdimage/tree.py [09:08] ogra_, yes, I saw that html inside python and ran away for a while [09:08] to gather courage [09:08] haha [09:08] I was expecting text files being catted to the header is good shell tradition :) [09:09] well, thanks to your pushing for a fully open cdimage tree we ended up with a complete python rewrite :) [09:10] no more shell in cdimage === deegee__ is now known as drussell [09:42] didrocks, ciao didier :) [09:42] didrocks, was wondering what you want to be tested now for the snap decisions [09:45] * OttOmanTR still waiting to see the first Ubuntu Touch phone :) [09:49] How do I go back to the previous activity in Ubuntu Touch ? [09:50] like in Android back button (software/hardware) [09:50] OttOmanTR: What are the plans for an Ubuntu Touch phone ? [09:50] OttOmanTR: this year? [09:50] cff, there should be a back button in the toolbar of the app ... swipe from the bottom [09:51] if you want to switch apps, swipe from the right ... ubuntu doesnt have that "one back button for all" concept android has [10:02] MacSlow, hey man [10:03] how would you test snap decisions for wifi? [10:06] Cimi, you mean the password-entry? [10:07] MacSlow, I believe so [10:07] MacSlow, waiting didrocks [10:07] ogra_: There are multi-window attempts in Android but Android dev says Android UI isn't designed to be multi-window supported. I wonder how Android legacy is in Ubuntu Touch. What about Multi-Window? [10:08] Cimi, try connecting to another hotspot [10:08] MacSlow, I mean, automated tests [10:09] MacSlow, autopilot [10:09] (back from meetings) [10:09] Cimi: yeah, what's up? [10:09] didrocks, I wanted to mak you guys happy and test what needed to be tested [10:09] Cimi: with nested Mir, the snap decision for network didn't show [10:09] (and the unity8 AP tests, AFAIK were all passing) [10:10] so we need a test ensuring that a snap decision will always be shown, even with this nested Mir [10:10] didrocks, we need an autopilot test for that? [10:10] ok [10:10] MacSlow, ^ [10:10] MacSlow, I can do it, if you guide me [10:10] good luck guys ;) [10:12] Cimi, you'd need to port the examples/sd-example-password-entry.py from lp:unity-notifications to an AP-test [10:12] MacSlow, ok === rachelliu_ is now known as rachelliu [10:37] ricmm_: we got a retrace from a maguro crash leading to qtubuntu initially, bug #1271879 [10:37] bug 1271879 in qtubuntu (Ubuntu) "qmlscene crashed with SIGABRT in qt_message_fatal()" [Undecided,New] https://launchpad.net/bugs/1271879 [11:01] Mirv: can you get the runenv from that? [11:01] envs and pid space === ricmm_ is now known as ricmm [11:01] Morning all [11:01] morning [11:01] moaning === chihchun is now known as chihchun_afk [11:10] ogra_: oh cheer up it can't be all that bad ;) [11:10] :) === _salem is now known as salem_ [11:45] psivaa: do you know which kind of runenv ricm ^ is after? (regarding http://q-jenkins.ubuntu-ci:8080/job/trusty-touch-maguro-smoke-ubuntu-clock-app-autopilot/146/artifact/clientlogs/) [11:52] Mirv: looking [11:52] I have seen the runenv mentioned before but I've never really learned what kind of stuff people want they want that :) === MacSlow is now known as MacSlow|lunch [11:56] Mirv: ricmm: i have the device where that crash occurred, if you could let me know what i should do to get the envs and pid space, i could get tat [11:56] s/tat/that [12:02] I've 404 when browsing documentation, espacially the "implementation examples", like this one: http://developer.ubuntu.com/api/ubuntu-12.10/qml/mobile/qml-ubuntu-components0-textfield.html [12:03] Anyone know what to change in the URL to get the expected content? [12:03] I've tried to switch to 13.10 or 14.04 but same [12:03] dpm, ^^ [12:06] psivaa: thanks, let's see and learn what ricmm would want from there. and if it's not already in the artifact collection, maybe could be added. [12:07] Mirv: ack [12:20] ybon: remove the ubuntu-12.10 completely [12:21] ybon: http://developer.ubuntu.com/api/qml/sdk-1.0/ is how the api for qml starts for me [12:22] ybon: http://developer.ubuntu.com/api/qml/sdk-1.0/Ubuntu.Components.TextField/ === alan_g is now known as alan_g|lunch [12:22] yay, thanks davmor2 :) [12:22] ybon: if you saved a bookmark maybe they updated the site since then and your path is nolonger correct [12:22] nope [12:22] it's from the bottom of this page: http://design.ubuntu.com/apps/building-blocks/text-field [12:24] ybon: ah then it is the bookmark that design used which possibly worked till they updated the site [12:24] possibly yep :) [12:26] ybon: your best bet is to look on developer.ubuntu.com when you see links like that now there is a search box so you can just type in the TextField and it should just appear :) [12:27] ah, nice tip, thanks :) === k1l_ is now known as k1l [12:45] mardy: ping [12:58] tsdgeos: hi! === MacSlow|lunch is now known as MacSlow === dandrader is now known as dandrader|afk [13:13] Seems that if I put two Action in Page.actions, qmlscene crashes (when using from QT Creator at least). Anyone has noticed this? [13:19] how can I click an option in an ItemSelector with autopilot? [13:19] I've clicked it and made it expand but not sure how to select the one I want === alan_g|lunch is now known as alan_g === dandrader|afk is now known as dandrader [14:01] mardy: i can't login with canonical.com account into google code, can you? [14:10] xnox, qmake is not used directly, but QT_INSTALL_QML is [14:11] sergiusens: can you show me where QT_INSTALL_QML variable is used. [14:12] ? [14:12] sergiusens: or how. [14:12] xnox, hmmm, right, doesn't seem to be the case here; was just inferring since all these apps have this trick and it is used for the install path [14:13] sergiusens: maybe i'm missing something, but it compiles and installs correctly without that variable. [14:13] sergiusens: here is cross-compile build-log http://paste.ubuntu.com/6803096/ [14:14] xnox, this will actually allow me to cross compile as click, so I am all for it :-) [14:15] sergiusens: and all QML stuff is now installed into /usr/share/dialer-app/* I guess also in preparatation to click conversion ;-) [14:15] sergiusens: yeah, #sdk people were pinging me about it re: cross-compiling in click chroot. [14:15] sergiusens: you really want https://launchpad.net/ubuntu/+source/cmake/2.8.12.1-1ubuntu4 however =) [14:16] xnox, hmmm, well the one that is click ready is addressbook app,camera, gallery, notes, reminders (and require compilation) [14:16] xnox, I want to add the click chroot stuff into click-buddy, was just waiting for the sprint to pester you ;-) [14:16] xnox, the build error in your build is not a build error [14:17] it's a infra error [14:18] xnox, shell seems to have died mid test; let me just retrigger that [14:18] sergiusens: right. another part what i wanted to tackle at the sprint is setting up auto-cross builder to continiously check that things are all. [14:18] ok. [14:19] in terms of not-regressing cross-compilation. [14:19] sergiusens: who shall i pester to get access to any jenkins instances of ours to prototype and launch my own jobs. (just any i386/amd64 workers should be fine) [14:19] xnox, well I plan to cross build all the clicks, and it's an agenda item to tackle click building during merges at the sprint with fginther [14:19] xnox, that would be fginther [14:20] sergiusens: hm. i think i have slightly different plans. [14:20] xnox, like which ones? [14:20] sergiusens: i want to test the archive, and archive-proposed, to make sure the _toolchain_ is not regressing in terms of cross-compiling. [14:20] xnox, ah, that's ok; I thought you meant conflicting plans ;-) [14:20] and to do so, i want to trigger throw-away cross-compilations of what's in the archive / current click. [14:21] cjwatson, so i heard thre is a requirement to roll images with PPAs included for the new CI stuff, have you been involved with that, are there any concepts yet (i assume we will need a new image type or so in cdimage for this) [14:24] xnox, sounds good to me [14:25] ogra_: first I've heard of it [14:25] cjwatson, ok, probably a topic for next week then [14:25] ogra_: or, well, that's not quite true, I think it was mentioned at some point at the client sprint, but no work has been done on it [14:26] k [14:26] ogra_: I hope to make progress next week on the master side of moving livefs building into Launchpad (the slave side is at least theoretically done and in the launchpad-buildd version that's about to be deployed) at the sprint - I think that should help [14:27] ogra_: indeed I think that's a requirement [14:27] yeah [14:27] we've pretty much designed it, just have a pile of database patches and code to write [14:27] I'm rooming with wgrant which should help ;-) [14:27] the plan is obviously to have images built from CI PPAs so they are close to identical to what we ship [14:27] heh [14:28] which i guess means you want an output path per image in this case or some such [14:29] where utah then can pull from [14:42] kalikiana, ping [14:47] aquarius: pong [14:47] kalikiana, I am having weird failures with indexes in u1db-qml. Do you have a few to help me out? [14:48] aquarius: shoot [14:49] kalikiana, see http://pastebin.ubuntu.com/6803263/ [14:49] tsdgeos: never tried, I use my personal account [14:49] kalikiana, indexes don't seem to work on top-level fields in the document. [14:50] mardy: ok, i guess i use my personal one too for that index range patch [14:50] kalikiana, the examples show indexing on, say, hello.world.message in a doc that looks like {hello: {world: {message: "Hello"}}} and that works, but indexing on a field at the top level doesn't. Or am I going mad? [14:51] kalikiana, is there some way I can get debugging information out of u1db-qt so that I can see what it's doing? [14:51] Cimi, heyo [14:52] Cimi, sorry, I thought I mentioned I had solved that problem in yesterday's standup, but it didn't occur to me to check in with you because I didn't know there was a desire to reduce the scope of the test [14:52] tsdgeos: if that bothers you, you can just send me the patch as git formatted patch by email and I'll apply it (there's no need to pass through the issue tracker for such a small change) [14:52] Cimi, so the problem was that the unity8 session wasn't being marked as 'active' by logind, which affects a lot of dbus permission states [14:52] mardy: ok [14:53] Cimi, and seemingly the NM daemon woudn't talk to the session over dbus in this state [14:53] didrocks, we're talking about issue with nested mir ^ [14:53] Cimi, so if we want to properly test that code path, the AP test needs to actually talk to NM [14:53] so, it seems the valid test wouldn't be the snap decision, but the real network test [14:54] (as suggested at first) [14:55] aquarius: running the example nothing shows up, probably expected. I'm not sure what you mean by top level here [14:55] I see db.putDoc({type: "place", placename: inp.text}) here [14:55] kalikiana, hang on, let me create a more obvious example :) [14:58] Cimi, so your first approach, with dbusmock. Would that be a proper test? Or does it mock out the NM daemon? [15:00] mardy: ok, sent you the email, tell me if that's enough for you [15:01] Mirv: https://code.launchpad.net/~aacid/qtubuntu-sensors/temporary_open_file/+merge/202874 [15:01] tvoss_: ↑↑↑ === rachelliu_ is now known as rachelliu [15:06] mterry, Wellark said is too complicated now [15:07] kalikiana, see http://pastebin.ubuntu.com/6803336/ for an example. You'll observe that there are two listviews; the one on the left finds documents, the one on the right doesn't. THey both are tied to queries; the queries are tied to indexes; the indexes don't do anything different except that one of the indexes is on a field at the root of the document and one is on a nested field. [15:08] Cimi, well whatever didrocks is happy with is good with me [15:08] Cimi: well, seems that from what mterry told me, this won't test the failure with current nested more [15:08] quite easy, ogra_ can provide you a package to enable nested Mir [15:09] I just want an AP failing in the same or similar case than the one I described with NM [15:10] didrocks, not from my disk, i would have to build it again [15:10] ogra_: maybe push that to a ppa for Cimi? [15:11] ok [15:13] mterry, how would you test this ^ [15:14] Cimi, does NM let you set up mock access points? [15:14] pitti, ^ ? [15:14] tvoss_, Talking with kenvandine about content-hub with the embedded sources. [15:14] trusted sessions [15:15] mterry, Cimi: no, I don't think so [15:15] tvoss_, So it seems that there'll be a second instance of the app running. Or will there always be one instance. i.e. it can be embedded or not, but always one. [15:15] mterry: well, dbusmock's NM template obviously does, but not NM itself [15:15] pitti, how do they do tests? :) [15:15] mterry: upstream does them manually [15:15] guh, sounds awful [15:15] mterry: but I wrote some automated tests using mac80211_hwsim [15:16] tedg, there will be a special embedded version of the app running per content-picking operation [15:16] mterry: they are in NM's autopkgtest (and in an upstream bug report) [15:16] mterry: mac80211_hwsim is quite nice for that [15:16] Cimi, didrocks: We could go for a simpler test. Try to query NM over dbus and see if we get an error or a response of any kind... let me see which methods have the right permissions to test this [15:16] bug 1271963 [15:16] bug 1271963 in unity8 (Ubuntu) "Blank icons in "More suggestions" when no internet connection available" [Undecided,New] https://launchpad.net/bugs/1271963 [15:16] mterry: it's great for testing NM itself, but quite a bit too complicated IMHO if you want to test something that talks to NM; mocking NM is easier/faster/more flexible for that [15:16] (if anyone can reproduce/confirm) [15:16] tvoss_, So will that version of the app have a different app id? [15:17] I guess it would have to. [15:17] pitti, yeah, but in this case we really kind of want to test NM's policykit permissions are working in normal case [15:17] tedg, hmmm, why would it? === dpm_ is now known as dpm [15:17] tvoss_, Where is the second Exec line defined to run the app in that mode? [15:18] tedg, that would go in the app armor profile and the desktop file entry [15:18] tvoss_, Figured it'd have to be a separate desktop file (or other format), thus a different stanza in hooks. [15:18] mterry: ah; well, you'll need root privileges then, a bunch of dependencies (hostap etc.), and do that setup [15:18] mterry: although that really sounds like it should be tested in NM itself, not in an application.. [15:18] tedg, my idea was that we should have a special exec line for that in the same desktop file as the app's one [15:19] tedg, it really isn't a different app, just a different point of entry to the app [15:19] tvoss_, Seems to me that it'd be simpler as a "second app" in the Click package. That way the apparmor profile could be different, etc. [15:19] pitti, it's sort of an integration issue. Like, we enabled nested mode for Mir, which accidentally marked the unity8 session as inactive, which made NM deny us. So we just want to test somewhere in that stack to avoid this. We could query logind directly, or try a similar NM dbus probe or something. We don't *need* to mock NM on the backend [15:19] kalikiana, I think this is a bug, unless I'm doing somethnig massively obviously wrong, but I can't work out how to get debugging information. (Also, nothing shows up in index_definitions in the actual sqlite file, even for working indexes?) [15:19] tvoss_, It could call the same executable, but perhaps with different params and different confinement. [15:20] tvoss_, We could for instance not want networking in the content selection case, but want it in the main app. [15:21] mterry: ah, I see [15:21] tvoss_, This is basically what we're looking at doing for the infographic-visualizer case. So you could potentially have an app and visualizer in the same click package even though the visualizer would be far more restricted. [15:21] tedg, we do agree about the individual steps :) But I think it's still the same app and thus we should just have a section in the desktop file, something like [Content Picking] [15:21] mterry: well, if you want an integration test between app, mir, NM, and polkit, then mac80211_hwsim indeed sounds nice [15:21] Cimi, didrocks: the simplest test would be to just query logind and see if the session is active... [15:22] Cimi, didrocks: the simplest test would be to just query logind and see if the session is active... [15:22] mterry: or more appropriately, calling pkcheck with the privilege you want? [15:22] mterry, I'm not practical at all of those things... [15:22] pitti, ah... I forgot about pkcheck [15:22] mterry: (don't assume logind or anything which is an implementation detail) [15:22] tedg, or [Infographic] [15:23] aquarius: hmm I would expect ["toplevelfield"] there in the second index… I played a bit with the values just now, and it feels like a non-obvious bug is somwhere there - you won't see them in sqlite because it's all in qt data structures only – now there is a bug that *may* be related but I'm not certain if this is the same thing [15:23] https://bugs.launchpad.net/u1db-qt/+bug/1266478 – wrt debugging I wish there was a way, but I don't know what a good debugging feature would look like when it comes to queries/ indexing; if you have any user-side ideas I'd love to brainstorm because it could help fix bugs quicker [15:23] Ubuntu bug 1266478 in U1DB Qt/ QML "Query model doesn't work properly with a multilevel document list" [Undecided,New] [15:23] tvoss_, At least with infographic we want drastically different confinement, so I don't think that one makes sense. Content selection is fine, I like the separate App ID better, but don't see a huge difference. === salem_ is now known as _salem [15:24] tvoss_, But, perhaps more importantly, who are you expecting to parse that desktop file :-) [15:24] * tedg hides [15:24] kalikiana, also, we're not returning the contents of the document; we're only returning the fields that are in the index. [15:24] kalikiana, which means that I have to index eveything, which is really wasteful :( [15:25] kalikiana, also also, if you store everything in qt data structures then what does "create: true" do? And other u1db apps not using QML can't read the same database! [15:25] tedg, why does a different confinement scenario require modelling it as a different app? I'm a bit confused about that [15:25] kalikiana, I'll file a bug about this indexing thing. [15:25] Cimi, should be something like this: [15:25] pkcheck --action-id org.freedesktop.NetworkManager.enable-disable-network --process $$ [15:25] @parsing: I think ual is the system's point of entry to that whole world ... so I would say it should live there [15:26] Cimi, if that command succeeds, we're good [15:26] aquarius: I've recently worked to improve the Synchronizer.sync_output which collects different "messages" on what happens or doesn't happen - I wonder if this would also be a way for query debugging. not sure if you've seen that api [15:26] tvoss_, Because the apparmor profile's name is the appid. I guess we could create "$(appid)-profile" but that would require changing the apparmor tools to look into the desktop file to decide what profiles to generate. [15:27] tvoss_, Had a feeling that was your answer :-) [15:27] tedg, I actually like the appid-profile proposal [15:27] jdstrand, mdeslaur ^ [15:27] fyi, a click package can have two apps in it [15:27] pkgname_appname_version [15:27] that is the APP_ID [15:27] tvoss_, What it breaks is that currently the different click hooks are independent, this would make the apparmor hook work with the entire manifest. [15:27] the appname is the name of the hook [15:28] or rather, the a key in the hooks db [15:28] hooks: { [15:28] appname1: ... [15:28] appname2: ... [15:28] aquarius: "everything" here is indexing/query results only; create:true decides if you want to add documents that don't exist, if it's false they won't be added [15:28] whoops [15:28] hooks: { [15:28] appname1: {} [15:28] appname2: {} [15:28] } [15:28] so, you get two APP_IDs [15:29] pkgname_appname1_version [15:29] pkgname_appname2_version [15:29] tvoss_, FYI, this is what we've discussed: https://wiki.ubuntu.com/SecurityTeam/Specifications/InfographicConfinement [15:29] mterry, ok [15:29] kalikiana, indexes have create:true too. Index definitions are supposed to be saved to the sqlite file so that all u1db apps are saving in the same format -- it's not critical, but it would be nice if the python library could open u1db.qt-created u1db files... [15:29] aquarius: please do file each of the points you're making. just pragmatically something like reading python u1db documents has been unneeded so far and thus not high on the list, even if I theoretically agree [15:30] kalikiana, *nod* I am filing bugs as we speak. :) [15:30] jdstrand, so we could "abuse" pkgname as profilename? [15:30] jdstrand, sorry, appname as profile name [15:30] tvoss_: no [15:30] tvoss_, We'd be saying that each appname could only be for one profile. [15:30] it isn't abuse-- it is intentional [15:31] tvoss_, So you could be an app or a infographic visualization tool, but not both. [15:31] https://wiki.ubuntu.com/SecurityTeam/Specifications/ApplicationConfinement/Manifest#Click [15:31] tvoss_: ^ [15:31] hello [15:32] I've got a question: [15:32] tedg, jdstrand grabbing coffee and reading the wiki, gimme a few [15:33] aquarius: incidentally, you don't happen to be savvy on u1db.one.ubuntu.com/kalikiana and oauth? I have a branch that is supposed to make sync with a real account work, but I'm struggling to get auth working https://code.launchpad.net/~kalikiana/u1db-qt/syncWithU1/+merge/202508 [15:34] I'm finding myself in the position where oauth with one.ubuntu.com/api/account appears to be fine, but not with the u1db api === alan_g is now known as alan_g|tea [15:34] I try to install Ubuntu Touch on a Nexus 4, but it fails at boot, it always shows the "Google"-Logo with the opened lock. So it doesn't boot it. Why? Which file do i have to flash? [...]armel+mako.zip, [...]armhf.zip or both? [15:34] Tried with phablet-flash and manual method [15:37] kalikiana, I don't know *enough* detail about it to help debug it, I don't think. Working out why oauth doesn't work is a pain :( [15:37] kalikiana, I probably did know enough, but I'm a year off the pace now :) [15:37] kalikiana, two bugs filed, anyway [15:39] XorgMC, phablet-flash ubuntu-syystem -c trusty [15:40] XorgMC, that should just work (it definitely does here) [15:41] okay, i'll try. and it isn't because kitkat, i heared kitkat makes hard to root === alan_g|tea is now known as alan_g === gatox is now known as gatox_lunch [15:42] aquarius: no worries. thanks for the bugs! [15:42] XorgMC, its a nexus ... they are built with rooting in mind [15:43] thats why we chose them [15:43] Should i try it as root? [15:43] no, should just work [15:43] "phablet-flash: error: unrecognized arguments: -c trusty" ? [15:44] oh [15:44] try --channel [15:44] i thought the short option works too [15:44] kalikiana, also, how do you delete a document from u1db? [15:44] That works, downloads stuff now. Takes some time (600KB/sec) [15:45] aquarius: would you like to review this? https://code.launchpad.net/~kalikiana/u1db-qt/removeDoc/+merge/196301 [15:46] in short: there's no distinct api for it, but you can by setting it to an empty value [15:46] the above branch adds the api [15:48] hi ogra_ , just a quick question: any chance the sources will be updated to 4.3 or 4.4? [15:48] kalikiana, have added a note that it's called the wrong thing ;) More importantly, is "set the content to empty" what the reference implementation does when you call delete_doc? === dandrader is now known as dandrader|lunch [15:50] mamenyaka, https://plus.google.com/100264483712374857174/posts/hiefUmmUupj [15:50] mamenyaka, in the works :) [15:51] mamenyaka, target is end of the month [15:51] ogra_, great! [15:52] aquarius: in python ap it'S None, the closest to that is '' in qml [15:52] *api [15:52] ogra_, do you know anything about dualboot? will it support the unofficial ports? [15:52] kalikiana, ah, what I mean is, is that all it does? it doesn't create tombstones, flag the doc as deleted in some way, etc? [15:52] kalikiana, the test suite will test that stuff [15:53] hi tedg, I'm not sure if you saw my e-mail, but would you have a couple of minutes to discuss enabling translations in the indicators' upstream projects? [15:53] aquarius: "will" assuming somebody has tons of time to make it work [15:54] unfortunately that somebody isn't around [15:54] ogra_: ERROR:phablet-flash:Backup requested but cannot be completed succesfully, try with --system-image-ready if the system is already on an Image Based Ubuntu System to force it or use --bootstrap if data saving is not important or the system is not already on an Image Based Ubuntu System. [15:54] XorgMC, well, it tells you waht to do :) [15:55] use one of these options [15:55] * ogra_ would go with --bootstrap [15:55] dpm, Yeah, I haven't had a chance to look into it. Have you talked with didrocks about how that works with the landing process? If trunk == archive, not sure how LP merging in translations works. [15:55] aquarius: tombstone is the same as "is None", though, on the python side [15:55] it's not so clear in the docs but that's literally what the code does [15:56] aquarius: u1db-qt has no special tombstone api, all it does is skip such docs when using Query [15:57] kalikiana, ok. [15:57] XorgMC, btw, this is all described on https://wiki.ubuntu.com/Touch/Install [15:57] tedg, thanks, I haven't checked with him, but I can do. We discussed enabling upstream projects translations with seb128, and we already do it for system apps, unity and other projects [15:57] aquarius: but you probably asked for deleting the same reason others did, it's more ensuring to have a function than being told to set the value to empty [15:57] dpm, Yeah, everything is changing right now, so I'm not sure how all that fits together yet. [15:57] kalikiana, I did. :) [15:58] tedg, let me put together another e-mail trying to answer the questions and take it from there [15:58] bfiller, hi [15:58] attente: hey [15:58] aquarius: I'll rename it to deleteDoc, I agree with that point [15:58] dpm, Great, thank you! [15:59] bfiller, if the user switches display languages, should we also update the enabled-languages to append it as well? [15:59] ie update enabled-languages as well as active-language [15:59] kalikiana, gnaaah. How do I get the id of a document returned from a query? === jono is now known as Guest95543 [15:59] kalikiana, this is why query needs to return full documents, not just the indexed fields :) [15:59] ogra_ i've tried it like it was described there, it didn't work [15:59] attente: yes makes sense, as long as we first check that a keyboard layout exists for that lang [16:00] bfiller, ok. and what should we do if the user removes a keyboard layout which happens to be active? [16:00] aquarius: docId in the delegare [16:00] attente: that's a good question [16:01] attente: maybe set the previous one in the list to active? or set en as active? [16:01] kalikiana, ah, my apologies; it returns a subset of *contents*, but the acual docId is there, thank you. [16:01] aquarius: it's a "role" in qml speak, like "contents" [16:01] XorgMC, https://wiki.ubuntu.com/Touch/Install#Step_4_-_Downloading_.26_Deploying_Image_to_Device [16:02] XorgMC, says: "phablet-flash ubuntu-system --channel devel --bootstrap" [16:02] kalikiana, only returning a subset of contents is really annoying, but I can work around it by indexing all the fields I need. [16:02] bfiller, ok, i was wondering if setting to the first in the list was an option, but previous in the list sounds better [16:02]