/srv/irclogs.ubuntu.com/2015/03/19/#ubuntu-quality.txt

* ToyKeeper o/ a bit late...00:15
ToyKeepern00:17
ToyKeeper... would mean [n]ext song, if I hadn't moved the cursor right before pressing enter.00:18
knomewell it created nice legs for your smiley.00:18
ToyKeeper:)00:21
elopioo/00:52
elopioam I too late?00:52
veeberselopio: afraid so, your ubuntu\-qa badge has been revoked :-(00:53
knomeand all the cookies and pizza are gone00:53
knomethere's only one salty stick left but the dog licked it00:54
elopio:'(00:54
knomewell cheer up, at least it has some natural flavors to it from... nevermind.00:54
* elopio goes away to cry on the bed, and enjoy the extended nap :)00:54
knomesleep tight elopio00:54
veeberselopio: I'll be asking you for a review shortly if you'll be around00:59
elopioveebers: I was joking.00:59
elopioI'll be around.00:59
elopioI can't even go to the gym, today I can't raise my arms higher than the keyboard.01:00
veeberselopio: heh, you feeling the burn?01:00
elopioyes, I took the wrong class01:01
veebersI should have gone today as my gym buddy won't be there tomorrow and the gym is a scary place for a scrawny skinny guy :-|01:01
veeberselopio: what did you take?01:01
elopiovveebers: it was a kettelbells class. We learned how to hit our forearms and shoulders with iron balls.01:03
veeberselopio: holy crap, that does sound intense :-P01:07
elopioveebers: when you have some time, can you take a look to the packaging card? I've tested it and +1. It would be nice to get the second approval.01:08
veeberselopio: ack, can do. That's actually next on my things to do :-)01:13
elopio:)01:20
elopioI think I got the dbus upgrade working!01:35
veeberselopio: awesome!01:41
veeberselopio: hey, I've reviewed the packaging branch, Needs Info, due to the lack of -B in the readme (works for me with -B, I get apt 404s on the device otherwise)01:42
elopioveebers: did you phablet-network?01:42
elopioveebers: for some reason, -B works for me but fails for barry. without -B it fails for me with a missing fakeroot error, but works for barry.01:44
elopionow you are throwing a new error to the mix :(01:45
veeberselopio: really odd. Yeah the network is setup, it's that the image is out of date and needs an apt-get update. The problem is that even w/ a fresh image we have seen the apt cache out of date01:45
elopioright. I'm all for -B, as that's how we got the sanity working. But without understanding barry's error, I'm not sure which one to recommend.01:48
elopioveebers: do you know how to check the version of a py package if it has no __version__ attribute?01:52
veeberselopio: no I do not, sorry01:58
elopioveebers: that last card for xlib update just requires a top-approval on the autopilot branch, right?02:06
veeberselopio: if the packging has been acked, yes. We'll be doing a release tomorrow02:09
elopiook, top-approving then.02:10
* elopio goes for food02:36
elopiobbs.02:36
veeberselopio: if you're still around can you review the reflash MP for me please?03:33
elopioveebers: sure.03:33
veebersthanks03:38
elopioveebers: I think find_adt_test_file_names is complex. What do you think of this:03:38
elopioif we put all the upgrade tests in  ubuntu_ota_tests.tests, then we can use discover or something like autopilot list to find all of them.03:39
elopiothen, we can define the Test-Command like python3 -m unittest $TEST_NAME03:39
veeberselopio: if we want to use --testname to run a test then flash the device they need to be named in a Tests: stanza, using Test-Command: doesn't work (doesn't have a name)03:40
elopioand we could use something like: adt-run --testname command1 --setup-commands "export $TEST_NAME=ubuntu_ota_tests.tests.test_basic_upgrade" ...03:40
veebersalso, I think it only _looks_ complex, it's not really it's just parsing a text file03:41
elopioveebers: Test-Command gets the name command{index}03:41
veebersah right03:41
elopiomaybe what I'm proposing is more complex. I'm just thinking here.03:41
veeberselopio: I like the spitballing. where would we get the name to export in TESTNAME?03:42
elopioveebers: we need something like autopilot does when listing the tests.03:43
elopiowe do a loop over that list, running adt-run once per test.03:43
veeberswell, find_adt_test_file_names does that kind of, assuming the test is defined in d/t/control :-)03:43
veeberssounds like what you're proposing is pretty much what we have there, just actually running them is slightly different03:44
veebersi.e. parse what tests are available, and run them individually.03:44
veebersAs it is currently the tests can be shell scripts or python scripts etc. no restrictions there03:44
elopioright.03:44
veeberselopio: oh on that note, brendand tells me that the check accounts test needs to be a shellscript (not python) due to some permissions issues he came across03:45
veebersso we can't say it _has_ to be a python unittest file03:45
elopiough, I hate that.03:45
veebersyeah, I wonder if a little more investigation we could make it a python script, but for now . . .03:46
elopioveebers: so, on your branch we will need to move the basic test out of ubuntu_sanity_tests.tests and put it into debian/tests, right?03:46
veeberselopio: hmm, we should be able to keep it there and have a test script that calls python3 -m unittest ub. . . .03:47
veebersthat sounds like a bit of a work around though03:47
veebershmm, although I'm not to worried about that, it's moved a command from the control file into it's own standalone file which isn't too much of a bad thing03:48
elopioveebers: I won't mind about moving it out. I think your branch is right, just not what I had in mind.03:48
elopiobut it should work, lets give it a try.03:48
veeberselopio: aye, fair enough. If you think it should be (and can be) improved lets discuss it. Perhaps its a case of (and I say this a lot ;-)) that we iterate on it03:49
elopioveebers: I don't understand this comment: Hmm, apparently the revision argument is no more :-\03:49
elopioI actually like when you say "lets iterate on it" :)03:49
elopioveebers: I have latest vivid, and the revision argument is in there.03:50
veeberselopio: heh, that's a stale comment. The argument does work03:50
elopiocool.03:50
veebersits just that it needs to be before 'touch'03:50
veebersSo that's the 2nd time in as many days that I've been caught out by an agument being position dependant'03:50
elopioveebers: so I like the branch, I would like it even better with tests.  Like putting it in ubuntu_ota_tests/command_line and make some selftests, like we did for sanity.03:51
elopiobut the branch is already too big. I would suggest to split it: a first branch which only does one flash and runs adt-run ...03:52
elopiothis one will work on the current branch, where we have only one test for upgrade.03:52
elopiothen a second branch which does the test discovery, and puts the flashing in a loop.03:52
veeberselopio: I agree that its stupid that it lacks tests. I got a little carried away getting POC up on that branch before actually testing.03:59
veebersI disagree with the size, it's not that big and will take a while to take apart to make smaller branches that are all related anyway.04:00
elopioveebers: ok. I can help adding the missing tests, I still have a couple of hours before going to bed.04:00
veebersif you're really keen on that I can revert to an earlier rev where it's really simple and propose that, but then I would be fixing things in that branch that are already fixed anyway. I guess I'm just being difficult thought :-)04:01
veeberselopio: coolio, I'm confident that it's written in a a pretty testable way. Let me move things around a little so we can test it etc.04:02
veeberselopio: rats, actually I want to rebase it on barrys branch, but that hasn't landed yet. I don't want merge conflicts wasting our time either04:02
elopioveebers: splitting the branch was just a suggestion. If you think there will be no problem adding tests with the branch as is, I'm ok with that.04:02
elopioveebers: about barry's branch, I suggest to land it.04:03
elopiothere's a card about updating the readme, where we can investigate about -B or not -B04:03
veebersso on that note I've bottom approved.04:03
veebersdo you mind if I top  . . . you've already suggested that04:03
elopioveebers: please, go aheaad.04:04
veeberselopio: done. I'm going to grab trunk + barrys branch, merge that all in and move things around to get testing sorted04:04
elopiowhat we can do for now is to pass the -B from your script to adt-run.04:04
elopioone more crazy error in here. With v2.5, check_upgrade returns that there is no upgrade available when there is one.04:06
veebersugh, any idea why that is ?04:09
elopioveebers: no clue. For now I'll just skip the assertion.04:12
elopioactually, I'll let it fail.04:12
elopiothat's what the suite is for. There is a bug, it must ffail.04:13
veebersah I see, you're saying there is an actual bug, not with the test04:13
elopioyes.04:14
veebersWell thats good I guess :-) Already being useful04:15
veeberselopio: you'll be happy to see* that I've added another 100+ lines to that already big diff  :-)05:18
veebers(*note you may not actually be happy)05:18
veeberselopio: tests added (should have been there from the start, I know) and I'm going to go have tea. I'll be back on later tonight though to clear up anything05:19
elopioveebers: I'm happy.05:20
elopioI'll take a look.05:20
elopioveebers: I found that revision is actually not working for me.05:21
elopioI tell it to flash 152, and it flashes 153.05:21
veebersoh? ugh hmm05:29
elopioveebers: http://paste.ubuntu.com/10625388/05:29
elopioI'm not doing anything stupid, right?05:30
elopiomight be that it's late.05:30
elopiooh, I was doing something stupid.05:31
elopio:)05:31
elopioI forgot the --.05:31
elopiolets blame the hour.05:31
veeberselopio: heh, you need to get some rest :-)05:32
elopioI really do. I spent like an hour wondering why the fuck it was saying that there were no updates.05:33
veebersand it turned out to be a bug?05:35
elopioveebers: no, it turned out to be me being stupid for an hour.05:35
elopioveebers: your branch looks nice, and I like the tests. But I will leave the detailed review to the EU team.05:36
elopioI better go to bed. I'll just try to leave my branch in a decent state, in case one of them can finish it.05:36
veeberselopio: sounds good, thanks for your help tonight. See you tomorrow o/05:37
veeberselopio: let me know if there is anything I might be able to do with your branch, I may have some spare cycles this evening05:37
elopioveebers: you can give it a review, and give it a try.05:40
elopiohttps://code.launchpad.net/~canonical-platform-qa/ubuntu-ota-tests/dbus-upgrade/+merge/25342305:40
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
elopiono more meetings for... almost one hour.14:42
elopiobrendand: is there something useful I can do before the meeting?14:42
brendandelopio, i wanted to know if you tested barrys packaging changes?14:43
elopiobrendand: I did.14:43
brendandelopio, if you leave out -B it tries to build the package but it fails because of lack of fakeroot on the device14:43
elopiobrendand: yes. And if you add -B, it fails for barry but works for the rest.14:44
elopiobrendand: have you tried with -B ?14:44
brendandelopio, -B means not to build anything14:45
elopiobrendand: yes, but the ubuntu_ota_tests do not need to be build14:45
barryi have apparently bricked my phone :/14:45
elopiothey are on the current working directory.14:45
brendandelopio, hmmm. yes but only for the phablet user14:46
brendandelopio, it fails for commands run with sudo (the import)14:47
brendandelopio, at least when using Tests: (i was using Test-Command:)14:47
elopiobrendand: this is the command I'm currently using:14:49
elopioTest-Command: PYTHONPATH=$PYTHONPATH:$(pwd) python3 -m unittest ubuntu_ota_test\14:49
elopios.tests.test_basic_upgrade14:49
elopioonce we figure out how to make it work for everybody without -B, exporting patching the python path won't be needed.14:49
elopiobrendand: for your case, can you patch the pythonpath ?14:49
barryright, we really should not need to hack sys.path14:50
barrythat's kind of the whole point ;)14:50
brendandbarry, the build doesn't work for me unless i install fakeroot on the device (and specify --gpg-dir)14:51
elopiobarry: we need to figure out what was special about your device that didn't fail with fakeroot. But now you bricked it :)14:51
elopiobarry: were you running adt-run while the partition was r/w?14:51
barryyeah ;)14:51
barryno, i flashed it to -1 on ubuntu-touch/devel-proposed :(14:51
barrynow i'm trying to find a working revision14:52
elopioI'm using 150. Newer ones are bombing in weird ways.14:52
barryheh, i skipped over 150, let me try that one14:53
brendandelopio, barry - don't go near 15314:54
barryspinny ubuntu symbol isn't spinny14:55
barryyeah, even 150 asks for my passphrase but gives me no digits to push14:58
zsombiping ubuntu-qa15:00
brendandzsombi, hey15:01
zsombiI need some help in creating an autopilot test case for a sample test app which supposed to be an UriHandler15:01
zsombibrendand: hello :)15:01
brendandzsombi, ok, what problem do you have exactly?15:02
zsombibrendand: so, UITK autopilot test, must launch a test app with a given set of URIs15:02
zsombibrendand: the test must be an app which url_dispatch can call15:02
brendandelopio, i can't rely on PYTHONPATH because i need to use sudo to call the upgrade commands and it's not passing the environment through (even if i use -E)15:02
zsombibrendand: the instructions I see here https://developer.ubuntu.com/en/apps/platform/guides/url-dispatcher-guide/ state that I'd need to get 2 JSON files, but these are for clicks, and we all have fake desktop files, without any proper executable line...15:03
brendandzsombi, so you need to make a minimal app which can handle url-dispatcher requests?15:03
brendandzsombi, ok15:04
zsombibrendand: that's the smallest pain, I need to get that app to be registered as URL handler15:04
zsombibrendand: the app is fine, we have the QML elements to do the job15:04
brendandzsombi, ok. i'm in a bit of a tangle at the moment so i'll get back to you15:05
brendandzsombi, but i understand your problem and acknowledge your request15:05
zsombibrendand: ok, thx, you might get me on mail eventually, my day is pretty much done :(15:05
brendandballoons, issue with this page: https://developer.ubuntu.com/en/apps/platform/guides/url-dispatcher-guide/15:07
brendandballoons, the section on 'temporarily hardcoded urls' is false15:07
balloonsbrendand, looking15:09
brendandelopio, thoughts on zsombi's request?15:09
brendandballoons, that was reverted i distinctly remember, but confirm with ted15:09
balloonsbrendand, ahh.. so there are no special urls?15:09
brendandballoons, support for file:// urls was completely removed15:10
brendandballoons, again i'd confirm with ted though15:11
balloonsright, will do.. thanks15:11
elopiobrendand: zsombi: we need ted involved in this discussion. How to register a click app as url handler without doing the click install ?15:12
brendandelopio, yeah15:12
tedgHowdy15:13
tedgThere is still a hardcoded URL for apps, but not the music or video ones.15:13
tedgbrendand, ^15:13
brendandtedg, so that page is wrong?15:13
tedgOut of date, yes.15:13
jfunkubuntu-qa -- new QA Engineer role is now up15:14
jfunkso tell your friends15:14
brendandtedg, we also had a question, how would one register an app as a url handler without actually doing the click install?15:14
balloonstedg, I'd be happy to help get the page updated. Is there a source for up to date information?15:14
davmor2jfunk: you think we have friends15:14
brendandtedg, for test purposes for example15:14
om26erdavmor2, talk for yourself :p15:14
brendandjfunk, all my friends are already QA engineer :)15:15
brendandjfunk, or should that be :(15:15
rvrdavmor2: Silos are very friendly to me ;)15:15
tedgbrendand, You drop a url-dispatcher file in ~/.config/url-dispatcher/urls/15:15
tedgbrendand, The name of the file is $(appid).url-dispatcher15:15
om26erdavmor2, I have 200 friends on facebook15:15
tedgbrendand, It should then detect the file and update the DB.15:15
tedgbrendand, BTW, that feature has a unit test, but I don't think anyone has used it regularly, so there might be bugs.15:16
davmor2om26er: now remove family and people that work at canonical and how many then ;)15:16
tedgballoons, No, just delete the "Temporary" section.15:16
brendandtedg, where does the appid come from? considering this wouldn't actually be an installed click15:16
tedgbrendand, From the desktop file for the application. Which would be $(appid).desktop15:17
brendandtedg, so as long as it matches the .desktop file that's ok?15:17
tedgbrendand, That would probably be in ~/.local/share/applications15:17
elopiobrendand: we are using a temp file for that, so it's like tmp123p1masd.desktop, but it's something we can get.15:17
elopiobarry, brendand:15:17
elopio$ bzr bd15:17
elopio$ adt-run --built-tree=. --- ssh -s adb15:17
elopioThat should work for read-only.15:17
tedgbrendand, Correct, basically URL dispatcher will pass that to UAL.15:17
tedgbrendand, So if UAL can launch it, you're good.15:18
brendandzsombi, does tedg's comments help?15:18
barryelopio: that makes sense i think15:19
barryfwiw, i'll need to do a quick reboot in a few minutes15:19
elopiotedg: I have doubts about this: https://developer.ubuntu.com/api/qml/sdk-14.10/Ubuntu.Components.UriHandler/15:20
elopiohow do you pass those URIs to the app through ual ?15:20
tedgelopio, Magic :-)15:20
tedgelopio, They replace the %u on the Exec line in the desktop file15:20
tedgelopio, Or if the app is running it is sent via dbus.15:21
elopioahhh15:21
balloonstedg, brendand section removed ty15:21
tedgAwesome, thanks balloons15:22
elopiorhuddie: did you work on the noop test, or was it federico? I think it needs to be skipped when there is an update.15:22
elopiomy phone has just upgraded during selftests.15:22
brendandelopio, that didn't work for me15:22
brendandelopio, i still get the fakeroot error15:22
elopiobrendand: with --built-tree? That's not right, it shouldn't build anything.15:23
rhuddieelopio, that one was federico's, I saw it upgrader earlier too15:23
elopiorhuddie: do you think it's right to skip it? Just like on your test but the opposite? :)15:24
brendandelopio, see i look at buil-tree and automatically read unbuilt-tree!15:24
brendandwhich could have been the problem all along...15:24
rhuddieelopio, well, given that its a self test it shouldn't be doing the actual update... let me have quick look15:24
elopiobrendand: the readme currently says unbuilt-tree. So you were right before.15:25
brendandelopio, maybe that was a typo15:25
brendandelopio, so we update the README too?15:25
elopiobrendand: no, for some reason it was working for barry.15:25
elopiobrendand: there's a card for updating the README. But the instructions depend on whether we land veebers branch or don't.15:26
elopioat this point, I'm guessing we won't land it.15:26
brendandelopio, isn't that card for sanity tests?15:26
brendandelopio, who added it?15:26
elopiobrendand: no, for ota tests. I did, when I noticed the README made no sense with what we have now.15:26
brendandelopio, it shouldn't really be extra then15:27
brendandelopio, it's related to a story15:27
elopiorhuddie: I would like to have fake upgrades for these tests, but that will take some time to write.15:27
elopiobrendand: it's related to all the cards, but we forgot to edit it in all the cards.15:27
elopioso I made a new one so we didn't forget at the end. I will update it.15:28
rhuddieelopio, yes, I agree, some local server would be much better.15:29
elopiorhuddie: maybe even a dbus mock. If you agree, I will make the skip on this README test, so it makes sense to say that we should start by flashing an old version.15:30
elopios/README test/README branch15:31
rhuddieelopio, yes that sounds good to me.15:32
brendandelopio, so with that command the existing tests work but that seems to be a coincidence15:32
brendandelopio, since i don't think it actually installs python3-ubuntu-ota-tests15:32
elopiobrendand: it does. At least, that's what I understand from the docs.15:32
brendandelopio, unless of course it's specified as a dependency in a test15:33
brendandelopio, and when i do that it it says 'E: Unable to locate package python3-ubuntu-ota-tests'15:33
elopioI will try that.15:34
brendand'Note that all test dependencies are then satisā€15:36
brendand              fied by archive packages, unless you explicitly specify  locally15:36
brendand              built .debs with --binary.'15:36
elopiobrendand: where are you getting that from?15:37
brendandelopio, from the adt-run manpage15:37
elopioI see. brendand and does that work for you?15:37
brendandelopio, i'm trying it15:38
brendandelopio, it doesn't seem to do anything though15:41
elopiobrendand: did you pass the --binary before the build-tree?15:49
brendandelopio, strange that makes a difference15:50
brendandelopio, now it fails because of read-only15:51
elopiobrendand: well, it gives a different error :)15:51
elopioI will report a bug to adt-run, because it should install the binaries in the tmp.15:51
elopiobrendand: so for your branch, what about cd into the source directory after switching the user?15:53
brendandelopio, i don't actually switch the user15:54
brendandelopio, that actually breaks adt-run15:54
brendanddoing sudo -u in a test script causes it to hang15:55
elopiobrendand: I know, but like: sudo cd /tmp/blah; sh debian/test/upgrade15:55
brendandelopio, i could try that15:56
elopiobrendand: there's also the option to install fakeroot with a setup command, which I presume is the problem without -B.15:56
brendandelopio, yeah15:56
brendandelopio, it's a 'pity' pitti is not here15:57
elopioand anyway, all this would be temp solutions until we get the bug solved.15:57
elopiobrendand: also, I think you should report a bug for not being able to switch the user.15:58
elopiobarry: brendand: https://bugs.launchpad.net/ubuntu/+source/autopkgtest/+bug/143416316:50
ubot5Ubuntu bug 1434163 in autopkgtest (Ubuntu) "Can't run tests that require build in a readonly testbed" [Undecided,New]16:50
barryelopio: i wonder if i wasn't hitting that bug because i made my root fs writable?17:56
balloonsla_juyis, fyi, the help app is in the store if you want to check it out on your device. I'm curious again about any video content you might have17:56
elopiobarry: yes, the problem is that with r/o the fakeroot is not properly installed.18:06
=== daz is now known as Guest10803
Guest10803I submitted my first merge request!19:41
Guest10803https://code.launchpad.net/~darran-kelinske/music-app/lp_bug_1428454/+merge/25356619:41
=== Guest10803 is now known as dazza5000
dazza5000thank you to Nicholas Skaggs for the documentation on how to write autopilot tests!19:42
alesageballoons ^^ woo!19:44
dazza5000me happy19:44
dazza5000Is anyone running the ubuntu-calendar-app on vivid?20:12
dazza5000When I try to add a calendar I get this in syslog: Mar 19 15:12:36 localhost gnome-session[1833]: gvfs-open: settings:///online-accounts: error opening location: The specified location is not supported20:14
dobeydazza5000: url-dispatcher and all that stuff only works on the phone images with mir currently; tried doing it under unity8 preview session instead?20:24
dazza5000not yet20:25
dazza5000haven't run a unity8 preview session before20:25
dazza5000if you have to make changes to a merge request20:49
balloonsdazza5000, glad you liked the docs20:59
Letozaf_balloons, hey21:09
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!