[00:08] <balloons> phillw, see my mailing list post
[00:08] <balloons> elopio, I'm here atm.. likely you aren't :-)
[00:09] <elopio> balloons: I am.
[00:09] <elopio> I'm wondering about the filemanager case. If I patch the home with a directory in /home/phablet, the app can't be launched. But if I patch it with a dir in /tmp, it works.
[00:10] <elopio> so, why do we have to create all that complex dirs hierarchy instead of using /tmp ?
[00:13] <balloons> elopio, fm is unconfined.. it's not a good case
[00:13] <elopio> balloons: hum
[00:13] <phillw> balloons: I'll go dig it up... I've had the most amazing thing happen....
[00:13] <phillw>  Hi, Thanks! I want to learn a little more about this flavor of ubuntu! I am a teacher in a school that has many obsolete computers. I would like to use it there.
[00:13] <phillw> A teacher in Mexico. Our pre official 10.10 was the last time I was asked about using lubuntu on low spec kit.
[00:14] <balloons> phillw, that's an awesome story
[00:14] <balloons> I'm so happy you did the session, and really glad to hear the story!
[00:15] <phillw> And that, balloons, is why I stick with lubuntu. My local youth team have a chance to have a computer lab with two old computers.
[00:15] <balloons> elopio, so yea since fm is unconfined none of the apparmor stuff is needed; it can do whatever and apparmor won't complain
[00:15] <phillw> That man, will get what ever I can do for his school.
[00:15] <elopio> balloons: well, it crashes if we use a temp home inside the real home.
[00:18] <balloons> elopio, I seem to remember that as well.. not sure why
[00:19] <balloons> but yea, I left the apparmor stuff on the table.. it works atm for calendar.. it's in production even
[00:19] <balloons> music has been a complex pain. Filemanager should convert over painlessly I hope. Still it's pretty crazy for what we want to do
[00:21] <balloons> elopio, I'm off to walk the puppy. We can chat more about it, but suffice to say for now the calendar can serve as an example of what will work with apparmor
[00:22] <elopio> balloons: yes, don't worry. I was just wondering about the filemanager case. Your explanation was good, thanks.
[00:22] <balloons> I'm all for simplifying it, but Jamie explained how locked down the apps are, while still expecting a certain layout etc
[00:22] <balloons> the expected layout is what we are forced to reproduce since we are changing /home for them
[03:54] <pitti> Good morning
[09:24] <jibel> pitti, with latest autpkgtest there are again quoting issue with ssh because ssh executes the command with sh -c, it'll be the same with adb. But if I systematically quote the scripts it breaks the other drivers.
[09:25] <pitti> jibel: what does your auxverb look like now?
[09:25] <pitti> I suppose something like
[09:25] <pitti> #!/bin/sh -e
[09:25] <pitti> exec ssh ... "$@" ?
[09:25] <pitti> or does that need double quoting?
[09:45] <jibel> pitti, yes something like that but I probably did something wrong, rechecking.
[09:45] <pitti> jibel: ah, this would indeed need double quoting then
[09:45] <jibel> pitti, did you change something in the debug output of VirtSubproc, nothing is display with debug mode enabled
[09:45] <jibel> ?
[09:46] <jibel> for example: sudo ./run-from-checkout -dd libpng --- lxc -d adt-amd64
[09:46] <jibel> should display lines like <VirtSubproc> ...
[09:47] <pitti> jibel: I just committed some refactoring for debugging, yes (~ 10 mins ago)
[09:47] <pitti> so if you rebased in the past 10 mins, then yes
[09:47] <pitti> jibel: it's a prerequisite for refactoring the debian/test/control parsing; I need to put that into a separate class so that we can also get a parser for click metadata
[09:48] <pitti> jibel: it now displays the program name first, i. e. "adt-run:" (as it always used to), or "adt-virt-lxc: DBG:"
[09:49] <jibel> okay, I don't have the latest version, I'll update it
[09:49] <pitti> jibel: ah, if you don't, it should also show debug info of course
[09:49] <pitti> jibel: it would only affect your new ssh runner if that uses VirtSubproc.debug()
[09:49] <jibel> it didn't
[09:49] <pitti> don't worry about that for now, too many variables at once I suppose :)
[10:15] <jibel> pitti, so exec ssh -p 22 root@10.0.3.125 -- $(printf '%q ' "$@")
[10:15] <jibel> did the trick
[10:15] <pitti> jibel: ah, nice!
[10:17] <jibel> I still have an error: "sh: 3: cd: can't cd to ."  when it executes the script that fetches the source, searching why . is not accessible
[10:41] <jibel> ah, it's because 'cd' is executed via su and ubuntu doesn't have access to /root
[10:48] <pitti> jibel: whoops, that sounds like something which should be fixed in adt-run proper?
[10:48] <jibel> pitti, not sure, it's happening only with ssh
[10:53] <pitti> jibel: ah, that's because ssh defaults to cd $HOME, while lxc-attach and friends don't; but schroot ought to have the same problem, hmm
[10:54] <pitti> su ought to change it though
[10:59] <jibel> pitti, su -l fixes it
[10:59] <jibel> testbed command ['sh', '-ec', 'su -l -s /bin/sh ubuntu -c \'exec  ...
[11:00] <jibel> and can probably remove -s /bin/sh too unless there is a reason to force the shell
[11:01] <pitti> jibel: probably, yes; that runner shouldn't run as "nobody" or similar disabled accounts
[13:50] <balloons> brendand, time to setup your session :)
[13:51] <balloons> brendand, have a look at https://wiki.ubuntu.com/UDS/Sessions. you can do it all yourself
[22:02] <elopio> ping ubuntu-qa: somebody available for a review?
[22:02] <elopio> https://code.launchpad.net/~elopio/ubuntu-calculator-app/swipe_to_delete/+merge/223001
[22:05] <veebers> elopio: is self.waitingConfirmationForRemoval.wait_for(True) doing the same as screen.confirm_removal() ?
[22:05] <elopio> veebers: no. The first one waits for the confirmation to appear.
[22:05] <elopio> the  second one clicks it.
[22:05] <veebers> ah ok, makes sense
[22:06] <elopio> veebers: well, it could be clearer. But that comes from the QML code.
[22:06] <veebers> ack
[22:06] <veebers> elopio: where do the numbers on diff line 75-77 come from / are calculated?
[22:07] <veebers> hmm nvm
[22:07] <elopio> veebers: trial and error. The labels section takes less than 1/3 of the screen.
[22:07] <elopio> and you have to swipe for at least half of the width.
[22:07] <elopio> so 1/3 to 1/3 + 1/2
[22:08] <veebers> elopio: trial and error foe those numbers or for working out how much of a swipe to make?
[22:09] <elopio> veebers: for both. I just found that the default which is: swipe from almost the left side to almost the right side doesn't work for the calculator.
[22:09] <elopio> so I moved a little to the left, and a little to the right until I understood how it works.
[22:09] <veebers> elopio: would that indicate an issue with the calc app? i.e. if a user is used to swiping in one way then tries it with calc and it doesn't work?
[22:10] <elopio> veebers: yes, probably. It was really hard for me to do it.
[22:10] <elopio> now that I understand it, it makes sense. But you are right.
[22:10] <elopio> I'll report a bug to UX.
[22:11] <elopio> I suppose it's possible to identify if it's a drag instead of a tap.
[22:12] <veebers> elopio: cool, other than that the code looks good to me, I'm not setup to run and test the code right now though sorry :-\ But I can give my LGTM if you like
[22:12] <elopio> veebers: yes, that's fine. Thanks for your review.
[22:12] <elopio> I'll wait for jenkins to confirm before approving it.
[22:12] <veebers> elopio: although, if you're filing a bug to UX you should link it in a comment for the "_drag_pointing_device_to_delete" method
[22:12] <robotfuel> elopio: I marked it as lgtm
[22:13] <elopio> veebers: I'll do that.
[22:13] <veebers> elopio: awesome :-)
[22:13] <veebers> hey robotfuel how goes things?
[22:13] <elopio> thanks robotfuel.
[22:14] <robotfuel> hi veebers, better than yesterday.
[22:14] <veebers> robotfuel: that's good to hear. Hopefully you feel 100% again real soon