=== shuduo_afk is now known as shuduo === _salem is now known as salem_ === salem_ is now known as _salem [05:02] nik90: hello, do you want to test drive the latest QtC plugin from the https://launchpad.net/~ubuntu-sdk-team/+archive/ubuntu/tools-development? === chihchun_afk is now known as chihchun [07:02] good morning [07:14] morning all [08:18] Good morning all; happy Friday, and happy SysAdmin Appreciation Day! :-D === renato is now known as Guest59000 [09:09] bzoltan: hey, good morning [09:09] bzoltan: I forgot to remove the tools-development ppa after the previous testing [09:09] bzoltan: so I must be running the latest QtC plugin already? [09:09] bzoltan: what would you like me to test specifically? [09:11] nik90: if you have run dist-upgrade in the last ~15 hours then you have it [09:12] nik90: mostly the the emulator startup, html5/webapp packaging and deployment, correct fw list on the Publish page. Generic stuff. [09:12] nik90: thanks for your help [09:12] bzoltan: will check it out [09:12] np [09:53] t1mp: hey, in your https://code.launchpad.net/~tpeeters/ubuntu-ui-toolkit/rtm-fit-finish/+merge/227696 branch are you also planning to change the color of the listitem text to midAubergine like in the design docs of apps? [09:53] t1mp: or would you provide a property to change the text color? [09:54] dholbach, I've been pinged on #ubuntu-app-devel to get gallery-app approved in the store, as popey is not there. Is this something you could do? I could have a look myself (I've got the permissions, I think), but you've done more reviews/approvals and know better what you're doing. [09:54] Let me know if it works for you [09:54] sorry, I meant I've been pinged on #ubuntu-ci-eng [09:55] dpm, I don't see it in the review queue [09:56] dpm, right now we basically just run the review tools on the packages [09:56] dpm, I just checked - I replied on it "2014-07-23 11:06 - 1 day, 22 hours ago" [09:57] dholbach, ah, so it's indeed in the queue? [09:57] no [09:57] it was [09:57] I sent it back because it used an old framework [09:58] dholbach, ah, thanks [10:05] nik90: good question. I have a review with design in a few minutes, I'll ask what should be the default. I'll get back to you [10:10] t1mp: thnx [10:12] mzanetti: I got an answer to your question you had yesterday about changing the time via the clock app [10:12] mzanetti: https://docs.google.com/presentation/d/1JvDyhsW17d1-Mz8OY1YMBKwfRI2z9qgyRjbujEsxEMk/edit#slide=id.g18895458d_024 [10:13] mzanetti: we will provide a settings option in the clock app for that. On clicking it, the user will be taken to the system settings date&time page where they can change it [10:14] nik90: nice! [10:15] :0 [10:16] When I make a scope app, I receive error" package 'libunity-scopes>=0.5.0' not found [10:16] " [10:19] help [11:41] nik90: text in list items is dark grey by default, so apps with purple need to override it [11:42] nik90: I guess that's not possible now, did you report a bug for it? [11:42] t1mp: no I haven't reported a bug about it. I wonder if exposing would be better or if using the style API is better [11:42] exposing via property or via the style api* [11:45] * nik90 is super excited about https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1254775 [11:45] Launchpad bug 1254775 in Unity 8 "Desktop file can specify a splash image and background color that shell will show until app is ready" [Medium,Triaged] [11:50] t1mp: reported bug at https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1348604 [11:50] Launchpad bug 1348604 in Ubuntu UI Toolkit "Provide a property to change the font color of listitems" [Undecided,New] [11:54] karni: oh btw, I cannot fix the docs with no images since those point to the original qt docs. Only mhall119 can fix that :) [11:54] right, thanks nik90 :) [11:55] karni: I am more into fixing our own SDK documentation [11:55] ack [12:20] rpadovani: hey, got a minute? [12:20] nik90, for you always ;-) [12:20] rpadovani: :) can you review https://code.launchpad.net/~nik90/ubuntu-clock-app/improve-hour-hand-accuracy/+merge/227885 [12:20] rpadovani: 1 liner [12:24] nik90, looks good, 3 miuntes that I think to corner cases... [12:24] *minutes [12:24] ok [12:28] nik90, just to be sure, analogTime.getMinutes() / 2 becomes a float right? I mean, 00:00 has angle of 0 and 00:01 has angle of 0.5? [12:29] rpadovani: I would assume yes [12:29] rpadovani: since rotation is a float, it should convert it automatically [12:29] nik90, approved :-) [12:30] rpadovani: thnx [12:35] mzanetti: ping [12:43] nik90: pong === _salem is now known as salem_ [12:43] mzanetti: hey I did a small test where I changed the timezone in the system settings app and returned to the clock app. I noticed that the clock app time was incorrect. However on restarting the clock app, it then showed the correct time. [12:44] mzanetti: could this be the same bug you reported where timezone changes are not transfered properly to the javascript date object? [12:44] nik90: I don't think so... [12:44] nik90: but no idea what it is [12:45] mzanetti: hmm ok. I am going to make the plugin output the local time and see if that shows correctly after the timezone change. [12:45] yeah, try that.. [12:53] nik90: yeah, maybe using the style api is better [12:54] t1mp: is there a documentation for that or any examples that I can take a look at? [12:57] nik90: no, we didn't write new docs yet on the new (since one year?) theming engine [12:58] nik90: it is a bit confusing for me too, zsombor knows it all but he is away now [12:58] nik90: perhaps there are other apps that use a custom theme [12:58] nik90: oh wait I am talking nonsense [12:58] nik90: the list items are not themable [12:58] nik90: so we'll have to fix your bug [12:58] t1mp: :) [12:59] t1mp: I will see what the address-book is doing for instance [13:00] t1mp: address-book defined their own custom listitem since they needed one with the new delete swipe actions [13:21] nik90: yes, it is completely custom [13:21] nik90: zsombor is working on completely new list items for after RTM [13:22] t1mp: Yeah he told me...but its too late for me. I am thinking about temporary solutions for now. [13:38] dholbach: hi! [13:39] hey jdstrand [13:39] dholbach: ok, I think we are ready to cut click-reviewers-tools 0.8 [13:39] yeeehaw! [13:39] jdstrand, great work [13:39] jdstrand, shall I take care of it? [13:40] dholbach: snag r214 and do whatever testing you normally do. please ping me if I need to fix something [13:40] awesome [13:40] dholbach: oh, heh, yes please :) [13:40] will do [13:40] thanks! [13:40] thank you :) [13:40] dholbach: this should be the one that works without needing any apparmor deps [13:41] *\o/* [13:41] so that will be nice [13:41] trusty users, unite! [13:43] dholbach: I'm happy about that, but I'm most happy that now if an unrecognized hook is found, it will WARN [13:43] we ended up getting so far behind with all the new hooks [13:43] that's great! good thinking! [13:44] some had security impact too, so I'm glad we are caught up [13:57] t1mp, Hi! Is the current height of the header as designed ? With new fonts changes header just looks bigger [14:03] jdstrand, can you have a quick look at https://code.launchpad.net/~dholbach/click-reviewers-tools/pep8-fixes/+merge/228315? [14:03] om26er: the header icon sizes became smaller also, but they will become bigger again [14:03] om26er: the header is the same as before, but it will be 0.5 grid units smaller soon [14:04] t1mp, alright, great. I see the header size in dash is smaller than what we have in apps :) [14:08] om26er: the height and auto-hiding of the header in dash is something custom (only what's inside the header they took from sdk) [14:08] om26er: but we'll make it a little bit less high in the uitk [14:09] dholbach: huh-- are you running utopic? [14:10] dholbach: trusty's pep8 was clean [14:10] jdstrand, yes - let me give you the output [14:10] dholbach: that's ok, I have a schroot [14:11] oh, I fixed some of those in the past [14:11] (different files) [14:11] gerlowskija, get your adt-run question answered? [14:15] mzanetti: bug confirmed :/ .. the c++ plugin returns the correct time while the javascript date object that I use to display the time in the main clock fails miserable [14:15] mzanetti: atleast I know now how to fix it. I might create a new plugin whose only job is to return the local time [14:16] nik90: meh... if its different than my example, feel free to add more info to the bug report [14:16] mzanetti: ok [14:16] mzanetti: I have one other performance issue with our timezone model [14:17] nik90: performance issue? [14:17] mzanetti: yeah. The xmllistmodel that we feed into the c++ plugin now has about 150 cities. So when I load it, it takes about 5 seconds to display the page [14:17] mardy, any luck having a look at https://bugs.launchpad.net/reminders-app/+bug/1347905? [14:17] Launchpad bug 1347905 in Ubuntu Reminders app "Autopilot tests fail to create fake account on device" [Critical,Triaged] [14:18] mzanetti: I will still need to do some debugging to see if it might be a bottleneck elsewhere [14:18] dholbach: approved. thanks! [14:18] nik90: parsing that xml model takes 5 secs? [14:18] wow... seems I messed up badly then [14:19] nik90: can you please add this at selected places of the code: [14:19] mzanetti: it could be my fault as well since what I do now is send the results of our timezonemodel into a sorted model provided by the sdk. The sorted model sorts it alphabetically. [14:19] QDateTime start = QDateTime::currentDateTime(); [14:19] and later [14:20] qDebug() << "operation took" << start.msecsTo(QDateTime::currentDateTime()); [14:20] then you can find out where it takes long [14:20] mzanetti: do you want that to be added inside the data() function? [14:20] nik90: I'd suspect the xml parsing code [14:21] nik90: but yeah... beginning and end of each function that could potentially be the bottleneck [14:21] mzanetti: ok will do and let you know the results [14:21] data() might be a bit odd for that, as its called rowCount * rolesCount times [14:21] so will spam quite a bit [14:22] mzanetti: yeah I think the loadTimeZonesFromXml() might be a good starting point [14:24] balloons: nope, sorry, was busiy with something else. Monday morning, I promise :-) [14:26] jdstrand, all done [14:26] popey, click-reviewers-tools should be happy on trusty again [14:27] \o/ [14:28] mardy, :-) === daker_ is now known as daker [14:35] mzanetti: confirmed. the bottleneck is loadFromXmlModel() which takes about 4.772 secs on the phone. [14:35] mzanetti: on the emulator it takes about 0.7 secs [14:36] nik90: really weird... [14:37] nik90: ok. need to debug that... don't have an idea atm [14:37] mzanetti: ok, I will also do some reading about this === boiko_ is now known as boiko [14:58] balloons: Nope, not yet. Still around? (and if so, up for an adt-run question? ;-p) [15:00] gerlowskija, better to ask the question, than ask to ask it ;) [15:10] just checking you were still around. adt-run has a bunch of options for specifying the source of the tests to run, and the click/deb package to run them against. But I was having trouble putting together a command to run the use a local checkout both for the source of the tests, and of the app-code. [15:11] i.e. What I want to do is bzr pull lp:~gerlowskija/ubuntu-calendar-app/my-branch-blah-blah [15:11] and then use the version of both the tests, and the app that are in that checkout [15:11] I had a couple examples of things that I'd tried and their results, but those are on my home machine. [15:13] jdstrand, I re-read the bug about terminal/file manager security, as there was something I didn't quite understand. Does this diagram reflect the workflow in the requirements? https://ubuntu.mybalsamiq.com/projects/ubuntuphonecoreapps/Terminal%20security [15:14] dpm, just imagine it runs under gksudo now ;) [15:14] (even thought it doesnt, since it doesnt elevate privileges, just allows access at all) [15:15] ogra_, that's what I thought, but I'm trying to figure out how to do authentication in the Qt world. We can check if there is a password via dbus, but I don't know how to prompt and check for the password from a Qt app [15:16] dpm, there is some code you can use as example in system-settings ... also mterry should be able to help you [15:27] balloons: oops, meant to tag you in one of those posts ^^ === chihchun is now known as chihchun_afk [15:30] gerlowskija, you would build the click file and pass that [15:30] so bzr branch lp:ubuntu-calendar-app [15:30] click-buddy --dir ubuntu-calendar-app [15:31] adt-run ubuntu-calendar-app com.ubuntu.calendar*.click . . . [15:32] balloons, thanks for the workaround for bug 1347905 - have you had to talk to mardy for the proper fix? [15:32] bug 1347905 in Ubuntu Reminders app "Autopilot tests fail to create fake account on device" [Critical,Triaged] https://launchpad.net/bugs/1347905 [15:33] gerlowskija, hey, thanks a lot for your awesome work and your patience with our tools! When your calendar branch lands we need to celebrate :) [15:36] dpm, I spent some time looking at it and prodding mardy about it. He's promised to have a look monday [15:36] balloons, i commented :) [15:37] balloons, you don't actually have to have the trailing ) i just found out [15:37] balloons, it can go at the end of the second line too [15:37] brendand, :-) The reason I wasn't concerned about the line is it shouldn't be there. It should be a method call to the class [15:38] balloons, do you think we can get reminders on the dashboard already, with your workaround? [15:38] see the commented line above, it gives an error and I wasn't sure why lasy night [15:38] dpm, we should be able to move it over [15:38] excellent [15:39] balloons, independently of the dashboard, could you organize a reminders upload? A critical bug fix (camera not saving pictures) landed, and it'd be good to get it in the store [15:42] dpm, sure thing [15:43] brendand, if you look at the MP can you understand why the self.app.main_view.no_account_dialog.open_account_settings() doesn't work? [15:43] we can land it as-is, but I'd like to understand that. Also a review from you so it can land would be helpful.. poor Leo has enough to look at [15:44] awesome, thanks balloons! === gatox is now known as gatox_lunch [15:47] balloons: I think I tried something similar yesterday evening when I was playing with adt-run. It saw the click, but reported that there were no tests in the manifest for the click, and exited [15:47] balloons: errr...actually forget I said that. I might've made some other mistake in my command...i'm useless w/o my bash history around ;-p [15:53] gerlowskija, well :-) [15:53] balloons, i'll have a link. elopio isn't working today anyway [15:53] s/link/look/ [15:54] brendand, excellent thanks. In the interim in case we land it as-is, I'll fix the flake8 issue [16:08] dpm: "Does the device have a password set? No" -> terminal *starts* [16:10] jdstrand, do I get this right? So if my phone doesn't have a password set and my little cousin grabs it he can just do whatever he wants with the terminal? [16:10] dpm: I think you want to use 'unix_chkpwd' if you have to write your own, but might check with mdeslaur [16:10] balloons: belated thanks balloons; I'll try your steps as soon as I get home. [16:10] dpm: that is correct [16:11] if you don't want your little cousin to have access, set a pin/password or keep it out of his hands :) [16:11] https://wiki.ubuntu.com/SecurityAndPrivacySettings/ProtectingUserData [16:11] better said than done ;) [16:11] ok, let me update the diagram [16:12] dpm: I don't understand the note "Or it could remain in the password prompt to allow retries" [16:12] dpm: we definitely want to allow retries [16:13] jdstrand, ah, so in that case the app should not quit, but it should remain open with the password prompt dialog [16:13] dpm: it should probably follow what we end up using in https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1348251 [16:13] Launchpad bug 1348251 in lightdm (Ubuntu) "please make use of pam_tally2 for Touch login and screenunlock" [High,Incomplete] [16:13] dpm, the new reminders is in the store. We'll try and push yet another one today with the workaround [16:13] dpm: well, it could quit. that meets the requirement [16:13] dpm: it seems not as nice though [16:14] dpm: we should not allow arbitrary retries without the application quitting [16:14] ie, if you are going to allow retries, it should quite after a while [16:14] quit* [16:15] jdstrand, something like that? https://ubuntu.mybalsamiq.com/projects/ubuntuphonecoreapps/Terminal%20security [16:15] awesome, thanks balloons [16:16] dpm: looks good (did you see my comment on unix_chkpwd?) [16:17] jdstrand, I saw it, I was going to ask you, I'd rather use code than calling binaries in the app. What step were you suggesting it for? For "password is set" or "password prompt"? [16:17] dpm, jdstrand: someone needs to write a library to check the password using pam that can be used from qml [16:18] dpm: password is set is in the bug. password prompt is what I was talking about [16:19] mdeslaur: unix_chkpwd should be using PAM, but I agree, a library is the way to go [16:19] mdeslaur, jdstrand, if that is the case, that's a big job, not sure we should be able to write a QML authentication library from the community core apps side of things [16:19] Initially I saw we could do the password check via DBUS [16:19] jdstrand: no, unix_chkpwd is what pam_unix uses, you can't use that directly [16:19] but password prompt and check is an entirely different beast [16:19] oh I see that now in the man page [16:27] jdstrand: is there any reason why this check and prompt can't be done via unity-app-launch of apparmor itself for unconfined apps? [16:28] apparmor can't prompt [16:28] why not? [16:28] it is all kernel [16:29] ual could, I guess [16:29] jdstrand: don't we prefix an app's Exec= with aa-exec or something like that? [16:29] mhall119: that would only work on non-unity8 [16:30] non? [16:30] jus write a QTsudo and hook that into the .desktop Exec line ? [16:30] ogra_: we don't want to run them as root [16:30] unity8 doesn't use those desktop files [16:31] jdstrand: it doesn't use ~/.local/share/applications? [16:31] Exec=aa-exec-click -p com.ubuntu.developer.mhall119.ureadit_ureadit_0.9.8~1 -- qmlscene @$ uReadIt.qml [16:32] mhall119: no, it doesn't [16:32] that is for non-unity8 [16:32] so..what does Unity 8 use? Saviq ? [16:32] if the application itself isn't doing it, then you need trust session support [16:35] balloons, reminders app seems to be broken more fundamentally [16:35] balloons, it's using the wrong desktop file to launch [16:35] balloons, ap tries to use /usr/share/application/reminders.desktop [16:35] balloons, which doesn't exist [16:36] brendand, well on the device / dash we'll be using launch_test_click [16:37] mhall119, perhaps there is something we can use from Qt or directly from C++ - ogra_ was pointing to the ubuntu-system-settings code for an example, but I've not yet had a look and I'm not too familiar with the code [16:37] mdeslaur: so you are saying that rather than using unix_chkpwd directly, we should use the PAM library (which will ultimately use unix_chkpwd itself), and all that should be in a library so it can be abstracted from the terminal and filemanager [16:38] brendand, that said, it seems the deb build the .desktop file as /usr/share/applications/com.ubuntu.reminders.desktop [16:41] jdstrand: yes [16:42] brendand, I can fix that with a simple cmake change [16:42] jdstrand: perhaps reusing this: http://bazaar.launchpad.net/~unity-team/unity/trunk/view/head:/lockscreen/UserAuthenticatorPam.cpp [16:42] or we just update the test [16:43] dpm: ^ [16:43] mdeslaur: that is the bit that sarnold reviewed? [16:43] jdstrand: I have no idea if sarnold reviewed that or not [16:43] no, that is unity [16:43] yeah, that's what the unity lock screen does [16:44] it can be used as an example [16:44] cool [16:44] thanks [16:45] mzanetti, quick question as a Qt expert: do you think this http://bazaar.launchpad.net/~unity-team/unity/trunk/view/head:/lockscreen/UserAuthenticatorPam.cpp could be made into a QML authentication plugin that the Terminal and File Manager app could use? The context is that we need to add a password prompt/check for those 2 apps [16:46] dpm: I think you would want something based on that broken out into a library/plugin, not that specifically [16:47] but yeah-- that is definitely the approach to take (I added a comment to the bug) [16:47] jdstrand, yeah, that's my question. Something based on that made into a QML plugin or a C++ wrapper that the apps can use [16:48] dpm: yes, but you'd need to get away with things like "const pam_message** msg" in the api and use Qt data types instead [16:48] but otherwise sure. I'm not entirely sure what it does though === gatox_lunch is now known as gatox [16:49] dpm: yeah, seems rather simple [16:50] mzanetti, ok thanks. As per what it does, I assume it does password authentication using PAM, I don't understand the code myself [16:50] dpm: also not using callback like this does, but switching it over to signals/slots [16:51] after reading a bit it just seems some glue code to talk to pam and ask a gui for input and display [16:51] yeah. shouldn't be too hard I suppose [16:51] mzanetti, do you think it would make more sense to make it into a QML plugin, or to keep it only accessible from C++? Both apps have some C++ code already, but I guess it'd be nicer to use it from QML [16:52] dpm: I'd definitely vote for both... but if you're doing it a bit clever I guess you can get a nice class that can be used in qml and c++ the same way [16:53] so you only write one class that c++ can include and is also qmlRegisterType'd [16:53] ok, thanks [16:56] dpm: is the intention that we would make this available to other apps? [16:57] mhall119, I think so, otherwise each app should ship a copy of this plugin. We should probably make it a QML plugin installed by default, that those apps requiring authentication can use [16:58] jdstrand: would we want to allow apps to call PAM? are there any potential security concerns to allowing that? [17:00] mhall119: it is fine cause they can't authenticate anyone else but the user [17:01] mhall119: and confined apps won't be able to do it [17:01] so a confined app wouldn't be able to brute force the password, for example [17:02] dpm: do keep that in mind ^ (only unconfined apps could leverage the plugin) [17:02] oh, good point about confinement, so perhaps we should not make it a system-wide plugin [17:02] but having terminal and filemanager need it is justification alone [17:03] we could, apparmor would just block it [17:03] ok [17:03] just like if a C++ app tried to access the PAM library itself [17:05] that didn't come out perfectly. if a C++ app tried to use the PAM library, there would be denials at some point that would prevent it from succeeding [17:06] there's no point in making it available when most apps won't be able to use it [17:06] especially when it's easy enough to just include it with the terminal and filemanager packages [17:08] * jdstrand doesn't have an opinion [17:08] whatever you decide-- it really shouldn't be hardcoded in both though. ie, embed the library if you must, but use a library :) [17:15] mhall119, yeah, you've got a point. I think I'd favour that too, although we'll have to think how to deal with code duplication. If we create a separate project for the pam plugin, I'm not sure how Jenkins will deal with fetching the code from there to put it in each app's click package [17:28] brendand, this should be ready for a final review: https://code.launchpad.net/~nskaggs/reminders-app/workaround-1347905/+merge/228213. I ended up removing the code that caused the flake8 discussion [17:31] balloons, i need to look at it more closely and unfortunately don't have the time now. i'll try and look at it later, but you might have to get elopio to look at it on monday (i'm off till wednesday) [17:31] brendand, ahh, ty [17:33] mzanetti, are you still about? [17:33] balloons: yeah [17:34] mzanetti, want to give https://code.launchpad.net/~nskaggs/reminders-app/workaround-1347905/+merge/228213 a quick peek? It's basically skip the tests that involve faking the account and services in an effort to get reminders onto the dashboard [17:34] I'd like to land this today so we can get it on the dash and monitored [17:34] things keep breaking out from under us [17:36] balloons: interesting... for me it fails exactly the other way round [17:36] balloons: not that code snippet, but creating accounts etc [17:37] balloons: do you want to disable test_go_to_account_settings everywhere? [17:37] mzanetti, yea.. I'm giving up on getting it to work on the device as well [17:38] sadly had these been running this month, we would have been informed when they broke [17:38] balloons: added a small comment [17:39] balloons: not saying you need to fix it. If you're still sure this is how you want it, I'm ok with it [17:39] ideally I want everything to work, but yes, I'd like it to get in today, so having some of the tests running is better than nothing :-) [17:41] mzanetti, replied. thanks [21:09] hi guys === salem_ is now known as _salem