[05:01] <Mirv> morning
[05:23] <pitti> Good morning
[08:03] <mlankhorst> morning
[09:00] <seb128> good morning desktopers!
[09:03] <Laney> hey!
[09:03] <seb128> Laney, hey, how are you?
[09:05] <Laney> hey seb128, I'm alright thanks, good climb last night ;-)
[09:05] <Laney> you?
[09:05] <seb128> I'm good thanks ;-)
[09:05] <seb128> how is your finger?
[09:05] <Laney> seems alright now
[09:06] <seb128> good
[09:11] <seb128> bah, freezes are boring, 4 updates this morning
[09:11]  * seb128 wants the good stuff to reach his update-manager ;-)
[09:16] <seb128> hey mvo ;-)
[09:16] <mvo> hey seb128!
[09:16] <seb128> mvo, wie gehts?
[09:16] <mvo> seb128: great, thanks! how are you?
[09:16] <seb128> I'm good thanks!
[09:17] <mvo> I'm on vacation currently, which means hacking time :) (and time to take care of the kids)
[09:17] <seb128> nice!
[09:17] <seb128> hacking time.... ;-)
[09:18] <seb128> mvo, speaking of which, aptdaemon decided to stop ftbfsing in trusty, which is good, but now britney is unhappy because some autopkgtest is red ... any chance you could have a look to it?
[09:19] <mvo> seb128: yeah, I was looking at the ftbfs some days ago and it was fine in my pbuilder so I assumed its cool
[09:19] <mvo> seb128: I'm looking at a apt adt failure currently, but aptdaemon can be next
[09:19] <seb128> great, thanks
[09:19] <seb128> that update includes the fix for one of the most reported e.u.c errors
[09:20] <mvo> seb128: yeah, I remember your MP for that
[09:20] <seb128> so it would be good to have it in the release pocket ;-)
[09:20] <mvo> seb128: indeed
[09:22] <mvo> seb128: I will try to reproduce in my newly acquired lxc testing container
[09:24] <seb128> "lxc testing container", nice
[09:26] <mvo> seb128: well, all credit to pitti who helped me get adt going
[09:26] <pitti> hey mvo, wie gehts?
[09:26] <pitti> bonjour seb128
[09:26] <seb128> pitti, salut, ça va ?
[09:26] <mvo> pitti: gut, danke!
[09:26] <seb128> pitti, see, a mvo around!
[09:26] <pitti> mvo: heh, I'm currently creating a doc/README.running-tests so that I won't have to any more in the future :)
[09:26] <seb128> isn't that nice? ;-)
[09:26] <pitti> much-needed introductory documentation anyway
[09:27] <pitti> seb128: en effet !
[09:27] <jibel> mlankhorst, seb128 morning, any news on bug 1284536?
[09:27] <ubot2`> Launchpad bug 1284536 in compiz (Ubuntu) "compiz crashed with SIGSEGV in two_way_long_needle()" [High,Confirmed] https://launchpad.net/bugs/1284536
[09:27] <mvo> pitti: hehe, yeah, that scales better :)
[09:27] <pitti> mvo: Attention ! Tu dois parler français ici, c'est la langue officielle d'équipe du bureau
[09:28] <seb128> jibel, hey, I think mlankhorst said it's not an issue in the software rendering stack
[09:28] <mvo> uhhh, so I need to fire up duolingo again ;)
[09:28] <jibel> guten Morgen mvo
[09:28] <mvo> hey jibel! great to see you
[09:28] <seb128> jibel, Trevinho said he can workaround it in unity the same way compiz workarounded it
[09:28] <jibel> mvo, nice to see you too :)
[09:28] <pitti> mvo: mon aussi -- ma derniére leçon est trop long
[09:29] <pitti> mvo: btw, I got the go-ahead for autopkgtest 2.9, I'll uplaod it today to Debian (after finishing the documentation) and sync it; then adt-build-lxc will be in the package, much easier
[09:29] <jibel> seb128, ah okay, there is no comment on the report.
[09:30] <seb128> jibel, I tried to get traction on IRC, mlankhorst worked on it
[09:30] <jibel> pitti, "dernière" "était" "longue"
[09:30] <jibel> ;)
[09:30] <seb128> jibel, is that creating issues for you/QA testing?
[09:30] <pitti> jibel: merci --- see, that's what I mean, need more practice :)
[09:31] <jibel> seb128, it's making installer tests really unstable and sometimes prevents ubiquity fails to start because of it.
[09:31] <jibel> s/prevents//
[09:32] <seb128> Trevinho, ^ you are working on that decoration/software-rendering issue right?
[09:36] <seb128> mvo, btw another "easy to trigger apport prompt" from update-manager is "close the polkit password prompt if you get one to install new packages" (or walk away from the computer until it hits the timeout, which is not that long)
[09:41] <Laney> oops, should remove landing PPAs after you're done with them
[09:56] <seb128> yeah, I got bitten by leftover landing ppa as well ;-)
[10:06] <mlankhorst> I'm tempted to upload a new synaptics
[10:07] <Laney> 'new'?
[10:07] <mlankhorst> seems to contain a bunch of fixes to make touchpad a whole lot less twitchy
[10:07] <mlankhorst> new = git snapshot of 1.8 branch
[10:09] <seb128> if it's bugfix only that's fine
[10:10] <mlankhorst> I don't think it counts as such
[10:10] <mlankhorst> but the new features are basically bug fixes that are slightly too big to fit in a point release, afaict from git history
[10:15] <seb128> mlankhorst, sounds like you would need a ffe for that update then
[10:16] <mlankhorst> yeah I'll do some more investigating first, seems a couple of the bug reports I'm hitting locally too
[10:33] <seb128> mvo, did you see https://code.launchpad.net/~brian-murray/aptdaemon/bug-1266844/+merge/207276 btw? another bug similar to the one I tried to fix
[11:06] <jibel> seb128, is anyone still actively maintaining firefox? autopkgtest is currently failing with bug 1285605, mozcrash is not in the archive or in debian and there is a force-badtest hint on firefox.
[11:06] <ubot2`> Launchpad bug 1285605 in firefox (Ubuntu) "autopkgtest fails with: ImportError: No module named mozcrash" [Undecided,New] https://launchpad.net/bugs/1285605
[11:06] <seb128> jibel, no, no firefox maintainer anymore (chrisccoulson does what he can but he's focussed on other things)
[11:06] <jibel> seb128, it is a rather big testsuite and use lot of resources
[11:07] <seb128> well, if it's not working and using resources, I would say to disable it
[11:07] <seb128> it's a bit of a shame but no point using resourcing for something which is not working...
[11:07] <Laney> https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/1274429
[11:07] <ubot2`> Launchpad bug 1274429 in firefox (Ubuntu) "Autopkgtests fail" [Undecided,New]
[11:08] <seb128> +1 for turning it off then
[11:09] <jibel> seb128, thanks, will do.
[11:09] <seb128> thank you
[11:15] <jibel> Sweetshark, hey, same for libreoffice, autopkgtest fails because a patch is rejected http://paste.ubuntu.com/7004529/
[11:15] <jibel> Sweetshark, is it something that you can fix? otherwise disable autopkgtest
[11:30] <Sweetshark> jibel: Ill look into fixing the auotpkgtests still. But do we have the infra to run them in jenkins anyway? previously we ran out of disc space for them ..
[11:33] <jibel> Sweetshark, we do, this has been fixed a while ago. The test timed out during the copy of the source tree which has also been fixed 3 weeks ago or so
[11:35] <jibel> Sweetshark, tests are now running on disk and can grow up to 40G of disk space
[11:39] <jibel> seb128, firefox tests have already been disabled in 28.0~b2+build1-0ubuntu1 (0ubuntu2 is in proposed). Once it is released I'll remove the package from jenkins.
[11:41] <seb128> jibel, ok, thanks
[11:41] <seb128> chrisccoulson, ^
[11:41] <seb128> jibel, can you update the bug Laney pointed with those info?
[11:42] <jibel> seb128, done
[11:43] <seb128> jibel, thanks
[11:58] <chrisccoulson> i would like to have working firefox tests again at some point, but running them just in the development release isn't much value to me :/
[11:58] <chrisccoulson> i need them https://launchpad.net/~ubuntu-mozilla-security/+archive/ppa and https://launchpad.net/~mozillateam/+archive/firefox-next really
[11:59] <chrisccoulson> we still waste a significant amount of time doing manual tests for updates
[12:02] <ricotz> chrisccoulson, hi :), are you about to push 28.0b6?
[12:03] <chrisccoulson> ricotz, it's quite low down on my list of priorities atm
[12:04] <chrisccoulson> ricotz, maybe later tonight
[12:04] <ricotz> chrisccoulson, oh, really?
[12:04] <ricotz> even thunderbird 24.3.0 didnt hit trusty yet
[12:04] <chrisccoulson> ricotz, yeah, i only touch it now when i get a few minutes in my spare time
[12:05] <ricotz> chrisccoulson, it would be great if you push those to the ppas if you arent comfortable having things in the main repo yet
[12:05] <chrisccoulson> ricotz, it should be in trusty. i just haven't had time ;)
[12:06] <ricotz> meaning if you haven't time to test them deeply
[12:06] <ricotz> i see
[12:08] <jibel> chrisccoulson, on which releases do you want to test firefox-next and security ppa? all stables + dev or only dev or only stables or only LTSes ? :)
[12:08] <chrisccoulson> jibel, it would be all stables (i don't use those PPA's for the dev release)
[12:09] <chrisccoulson> if we can do that, I'll spend some time getting the tests in order again, as that would save me quite a lot of time
[12:09] <jibel> chrisccoulson, ack, I'll see what I can do
[12:09] <chrisccoulson> thanks :)
[12:12] <seb128> chrisccoulson, well, at least infinity fixed the ppc build issue ;-)
[12:13] <seb128> chrisccoulson, so we should have an updated version in trusty after the beta freeze ends
[12:14] <Sweetshark> jibel: awesome. Ill see to get the autopkgtests back up then!
[12:55] <jdstrand> seb128, chrisccoulson, jibel: re firefox autopkgtests, there are only 60 something failures iirc, but thousands of tests that pass. wouldn't it be better to disable the 60 tests that can't import mozcrash?
[12:56] <jdstrand> until such time that someone can fix those tests
[12:57] <seb128> jdstrand, that would work for me, it's just that tests are red for ever and nobody did anything about it so far
[12:57] <jdstrand> chrisccoulson, jibel: +1000 on enabling autopkgtests in the ppas
[12:57] <seb128> but sure, if that works, +1
[12:57] <Laney> bah, I admit defeat
[12:57] <Laney> who wants to help me with some C++? :-)
[12:58] <Laney> it involves virtual functions
[12:58] <jdstrand> bug 1285605 seems to suggest it is python
[12:58] <ubot2`> Launchpad bug 1285605 in firefox (Ubuntu) "autopkgtest fails with: ImportError: No module named mozcrash" [Medium,Fix committed] https://launchpad.net/bugs/1285605
[12:59] <jdstrand> why not disable the script that runs /usr/lib/firefox/testing/testrunhelper.py?
[12:59] <chrisccoulson> that's used to run all tests
[12:59] <jdstrand> hmmm
[12:59] <jdstrand> then how are only 60 something tests failing?
[13:00] <chrisccoulson> if there are only ~1000 tests passing, then that would suggest most things are broken, as there should be in the order of 30000 test results
[13:00] <jdstrand> well, let's look at the failures real quick
[13:00] <jdstrand> https://jenkins.qa.ubuntu.com/job/trusty-adt-firefox/150/ARCH=amd64,label=adt/testReport/
[13:00] <jdstrand> 34 failures (±0) , 879 skipped (±0) 21,286 tests (±0)
[13:01] <jibel> jdstrand, mochitest are not counted at all because they completely fail to run. 60 are only for the tests that ran
[13:01] <jdstrand> weird, I could've sworn that number was 60 something, not 30 something
[13:01] <jibel> jdstrand, if you compare to saucy it's more like 3k tests that did't run
[13:01] <Laney> bzr branch lp:~laney/ubuntu-system-settings/reset-api build, install, USS_SHOW_ALL_UI=1 ubuntu-system-settings reset → Reset all system settings
[13:02] <Laney> crashes because it can't resolve the virtual function
[13:02] <jdstrand> looking at https://jenkins.qa.ubuntu.com/job/trusty-adt-firefox/150/ARCH=amd64,label=adt/testReport/, I see that 21,286 tests are passing
[13:02] <Laney> lib/Systemsettings/plugin-interface.h src/plugin.c reset()
[13:03] <jdstrand> with 34 failures
[13:03] <jdstrand> is it not possible to just disable/skip the 34 failures for now and then we still have the 21,286 tests?
[13:05] <attente> Laney, should the return type match?
[13:05] <jdstrand> jibel, chrisccoulson: ^
[13:06] <attente> Laney, never mind, guess i'm looking at the wrong function
[13:08] <jdstrand> seb128: ^
[13:09] <Laney> attente: line 253 in plugin.cpp
[13:10] <chrisccoulson> disabling / skipping tests requires a source change to firefox in any case
[13:10] <chrisccoulson> but if mochitests are being skipped then the 34 failures is inaccurate. that number is really in the 10's of thousands
[13:11] <chrisccoulson> and mochitests are the ones that matter
[13:12] <jdstrand> chrisccoulson: hrmm
[13:12] <Laney> attente: the idea is that panels can either handle resetting in their plugin or in their page component
[13:13] <Laney> the bool tells you whether it was handled or not - I tried to add a default false case
[13:13]  * Laney sucks
[13:14] <jdstrand> chrisccoulson: what provides mozcrash?
[13:14] <chrisccoulson> jdstrand, firefox. it's in the package, it just doesn't find it
[13:14] <jdstrand> so, it is maybe something as simple as adjusting PYTHONPATH?
[13:15] <jdstrand> or the equivalent module setting in the script (I forget what it is otoh)
[13:18] <jdstrand> chrisccoulson: https://bugzilla.mozilla.org/show_bug.cgi?id=849900#c1
[13:18] <ubot2`> Mozilla bug 849900 in General "remove hacks to enable mozbase from the automation code when we are running all tests in a virtualenv" [Normal,New]
[13:19] <jdstrand> jibel: will you be the one implementing the autopkgtest runs in the ppas chrisccoulson mentioned?
[13:21] <jdstrand> chrisccoulson: as for manually testing of security updates due to lack of autopkgtests runs> I would think that if we ran the autopkgtests manually on stable releases, we could back off on the manual testing to the point that we would if the tests ran in the ppa, no?
[13:22] <jdstrand> chrisccoulson: therefore, fixing this would be worthwhile for the next update
[13:22] <jdstrand> chrisccoulson: oh, https://bugzilla.mozilla.org/show_bug.cgi?id=827446
[13:22] <ubot2`> Mozilla bug 827446 in General "update mochitest, reftest, xpcshell to use mozcrash" [Normal,Resolved: fixed]
[13:23] <jdstrand> chrisccoulson: see comment #22 (among others)
[13:24] <jdstrand> chrisccoulson: actually, starting at 15. seems that might provide some hints anyway
[13:24] <chrisccoulson> thanks. the real issue is the first bug, because upstream are now running all tests in a virtualenv
[13:24] <chrisccoulson> whereas we aren't
[13:24] <chrisccoulson> i need to rework our test harness to create one
[13:25] <jdstrand> I see
[13:25] <jdstrand> anyhoo-- does what I said about stables and the next security update make sense?
[13:25] <chrisccoulson> yeah, it does
[13:25] <jdstrand> chrisccoulson: so maybe it is ok to disable the autopkgtests for now in trusty, get things flowing again
[13:26] <jdstrand> chrisccoulson: then when we can do manual tests until we get the autopkgtests working again (eg, circa next security update or so)
[13:27] <jdstrand> chrisccoulson: and the 'we' for trusty testing is not necessarily 'you', but all of us
[13:27] <attente> Laney, should you try to cast those plugins to the plugin interface when you call plugin->reset in PluginManager::resetPlugins
[13:28] <attente> or is that not necessary here
[13:28] <jdstrand> chrisccoulson: how does that sound?
[13:28] <attente> Laney, sorry, i got my types confused again, never mind
[13:28] <chrisccoulson> yeah, that sounds fine
[13:30] <jdstrand> jibel, seb128: ok, so yeah, just disable the tests for now, but someone is going to need to watch the firefox bugs in trusty for regressions. when trusty is released, it'll be in the security team's hands and we will either do rather extensive manually testing or fix the autopkgtests. when we fix the autopkgtests, we'll push that up to the dev release
[13:30] <jdstrand> that may happen before trusty release
[13:31] <jdstrand> jibel: is there some way I can file a high priority request to get the autopkgtests working on the ppas chrisccoulson mentioned?
[13:31] <jdstrand> s/working/running/
[13:49] <jibel> jdstrand, done for firefox-next and security PPA on P and S. For Q I don't have a testing environment ready yet, I'll create one unless you think it is not worth it because it EOL in 6 weeks or so
[14:13] <Laney> attente: stupid c++ eh
[14:14] <larsu> Laney: you're complaining about c++ a lot today :P
[14:14] <Laney> it's the same problem :P
[14:14] <Laney> attente had a look at it
[14:15] <larsu> ah
[14:22] <seb128> Laney, you should maybe ask chrisccoulson, he knows c++ quite well, he might be able to help you
[14:24] <seb128> Laney, do you consider https://bugs.launchpad.net/ubuntu/+source/nautilus/+bug/1130722 a feature or an UI change? ;-)
[14:24] <ubot2`> Launchpad bug 1130722 in nautilus (Ubuntu) "[ffe] Restore a traditional menu under Unity" [Medium,In progress]
[14:25] <Laney> seb128: borderline-ish, but probably a bug fix imo
[14:25] <seb128> Laney, should I "un-ffe" the bug or do you just want to ack it? ;-)
[14:26] <larsu> clearly a design-bug if you ask me
[14:26] <larsu> instead of a feature
[14:26]  * larsu might be biased
[14:27] <Laney> I commented
[14:27] <Laney> FFes should be "New" ideally btw
[14:27] <larsu> why?
[14:27] <Laney> so that you can tell in the list if they need attention
[14:29] <seb128> oh, right, sorry about that
[14:29] <Laney> nm, it's just a small detail
[14:30] <Laney> trying to keep on top of email but eventually I'll fall behind and try to use the web ui
[14:34] <tjaalton> so the test to check if gtk-doc-tools fails on launchpad, succeeds in sbuild (gstreamer-vaapi)
[14:34] <tjaalton> *if ... is installed
[14:34] <seb128> is your chroot uptodate?
[14:34] <seb128> we had a new gtk-doc-tools recently
[14:34] <jdstrand> jibel: awesome, that was fast :)
[14:34] <seb128> which might create issues
[14:35] <tjaalton> seb128: yes, updated daily
[14:35] <tjaalton> the test just does: ifeq ($(shell test "`dpkg -l gtk-doc-tools | grep ^ii`" && echo binary-indep),binary-indep)
[14:36] <jdstrand> chrisccoulson: ^ jibel enabled jenkins for firefox-next for and ubuntu-mozilla-security
[14:36] <seb128> tjaalton, :/
[14:36] <seb128> jdstrand, jibel, chrisccoulson: nice!
[14:36] <tjaalton> so thinking of just enabling it unconditionally
[14:37] <attente> Laney, i guess it's aborting early from the PluginPrivate::ensureLoaded function
[14:37] <attente> so m_plugin points to garbage address
[14:37] <jdstrand> jibel: is there a way to make sure that new releases have jenkins enabled?
[14:38] <attente> Laney, if you add m_plugin = NULL at the top, it seems to work
[14:38] <Laney> attente: at the top where?
[14:38] <Laney> I think that it can't resolve the virtual reset function
[14:38] <attente> near the top of PluginPrivate::ensureLoaded
[14:38] <jdstrand> jibel: I wonder if it makes sense to have trusty enabled now in both ppas... how will we ensure 'u' gets enabled?
[14:39] <Laney> if it's null then if (d->m_plugin && ...) makes the problematic call not happen
[14:40] <attente> right, but m_plugin doesn't seem to start initialized to NULL
[14:43] <jibel> jdstrand, task 45 of https://wiki.ubuntu.com/NewReleaseCycleProcess . You could add a link to a specific document for the security team
[14:44] <jibel> task 45 of the section "Previous release plus 1 day"
[15:10] <happyaron> seb128: hey
[15:10] <seb128> happyaron, hey
[15:10] <happyaron> seb128: ibus-pinyin is still in -proposed
[15:11] <happyaron> why it's stucked there?
[15:20] <attente> Laney, i think it's just a problem of ensureLoaded returning early before m_plugin is initialized
[15:21] <Laney> attente: yeah?
[15:22] <attente> Laney, guess so...
[15:22] <Laney> I don't see the problem
[15:23] <Laney> can you be more detailed? :P
[15:24] <jdstrand> jibel: fyi, "46. Notify Jean-Baptiste Lallement (jibel) or Martin Pitt (pitti) to initialize autopkgtest in the ubuntu-mozilla-security and firefox-next PPAs for the new release."
[15:24] <jdstrand> jibel: thanks again
[15:24] <attente> Laney, in src/plugin, add at line 90: m_plugin = NULL;
[15:29] <Laney> ubuntu-system-settings currently only matters on phone/tablet
[15:29] <Laney> NO
[15:30] <Laney> om26er: I'm not really cool with doing that
[15:31] <Laney> it leads us down a path of being crappy on the desktop, but we want to be used there
[15:31] <seb128> shrug, got disconnected from that server
[15:31] <seb128> Laney, doing what?
[15:31] <Laney> disabling tests on desktop only
[15:31] <seb128> we shouldn't disable tests
[15:31] <om26er> Laney, you can ignore that commit message, we are going to enable tests on desktop setup as soon as there is a fix for apport window poping up in that environment
[15:31] <desrt> *cough*
[15:32] <seb128> attente, what are the keybinding issues you mentionned during the meeting?
[15:32] <Laney> I don't think it's okay to ignore tests there because something is broken
[15:32] <Laney> we should fix the thing that is broken
[15:32] <Laney> that's what would happen if it was a problem on the phone, right?
[15:32] <Laney> also I've never seen this on my desktop
[15:32] <seb128> attente, is that the enable-mnemonic issue, or something else?
[15:32] <om26er> Laney, its much better than not running them at all, I think since I enabled them last week none of your CI showed success
[15:32] <attente> seb128, it's the ctrl+print problem
[15:33] <seb128> oh
[15:33] <seb128> if that's the only one that's pretty minor
[15:33] <seb128> did you look at it?
[15:33] <Laney> the same tests that all pass for me and seb128
[15:33] <attente> seb128, i did, but it's a really weird problem
[15:34] <om26er> Laney, the test environment needs to disable 'apport report a problem' window
[15:34] <attente> it's almost as if printscreen/sysrq can't be modified by ctrl
[15:34] <om26er> fginther, ^ :)
[15:35] <attente> but it seems to work under xev, so...
[15:36] <seb128> attente, does it work under gnome-shell?
[15:36] <seb128> attente, also the keybindings do work by default, it's only if print is unset that it triggers the other one
[15:38] <attente> seb128, ctrl+print doesn't seem to do anything for me by default
[15:38] <seb128> attente, it copies to the clipboard, did you try to go to e.g gimp or inkscape and paste there?
[15:38] <seb128> attente, it does the screenshot animation for me
[15:38] <seb128> then I've the image in the clipboard
[15:39] <attente> seb128, ctrl+print didn't do anything for me :(
[15:39] <attente> and if i do 'dbus-monitor member=AcceleratorActivated', i get nothing on ctrl+prtsc
[15:39] <om26er> Laney, your tests are actually not failing, autopilot have a way to check snapshot of running applications before and after a test is run, so if during that time a new window pops up autopilot believes that it started some window and failed to close it. Thats pretty much whats happening for our tests in CI
[15:40] <Laney> can we see what the apport crash is for?
[15:40] <attente> seb128, in xev, what's the output of doing ctrl+print for you?
[15:40] <fginther> om26er, the changes to remove the desktop test are in place. I spent some time debugging the problem yesterday, but my attempts to disable apport did not solve the problem so far
[15:41] <seb128> attente, http://paste.ubuntu.com/7005625/
[15:42] <seb128> http://paste.ubuntu.com/7005635/ for a bit more content
[15:42] <om26er> fginther, is there a way to get the apport crash file out of otto ?
[15:43] <attente> seb128, yeah... different from mine: http://paste.ubuntu.com/7005649/
[15:43] <fginther> om26er, yes, it should be linked to the jenkins job as an artifact
[15:44] <seb128> attente, keyboards :/
[15:44] <attente> seb128, indeed...
[15:44] <seb128> attente, well, don't bother much about that one, as said it's a corner case
[15:45] <om26er> Laney, the attached .crash files link to bug 1260237 and bug 1277589
[15:45] <ubot2`> Launchpad bug 1260237 in system-image (Ubuntu) "system-image-dbus crashed with PermissionError in initialize(): [Errno 13] Permission denied: '/var/log/system-image/client.log'" [Medium,New] https://launchpad.net/bugs/1260237
[15:45] <ubot2`> Launchpad bug 1277589 in ubuntu-download-manager "Better protection against concurrent access" [Critical,In progress] https://launchpad.net/bugs/1277589
[15:45] <om26er> fginther, thanks
[15:46] <Laney> hah, that first bug is from me
[15:53] <mvo> pitti: silly question, where does the  ./run-from-checkout --built-tree=~/devel/apt/mvo.git --- lxc -es adt-trusty store the generated logs?
[15:53] <pitti> mvo: it doesn't by default :) but you can tell it with -l /path/to/logfile or -o /path/to/output-dir/
[15:54] <mvo> pitti: aha, thanks
[15:55] <om26er> Laney, do you know the reason of that bug ? like what causes it
[15:57] <mvo> pitti: I think I have the apt failure under control, looks like SIGPIPE is behaving subtly different inside lxc, but I will simply rewrite the test to be more robust
[15:57] <pitti> mvo: *hug*
[15:57] <mvo> (and rely less on crazy shell constructs)
[15:58] <Laney> om26er: the log directory has weird permissions so the client cannot open the file
[15:58] <om26er> Laney, it only happens once if I am not mistaken ?
[15:59] <Laney> you won't get the same crash more than once from apport
[16:01] <Laney> om26er: system-image is maintained by barry - you could ask him if there's any plans to fix it
[16:03] <om26er> Laney, that could be the cause because we have a clean container every test run
[16:03] <Laney> makes sense
[16:27] <seb128> mpt, I know you are busy but I would appreciate if you could review those
[16:27] <seb128> https://bugs.launchpad.net/ubuntu/+source/unity-control-center/+bug/1285021
[16:27] <ubot2`> Launchpad bug 1285021 in unity-control-center (Ubuntu) "What's UI Scale?" [Low,Triaged]
[16:27] <seb128> https://bugs.launchpad.net/ubuntu/+source/unity-control-center/+bug/1284221
[16:27] <ubot2`> Launchpad bug 1284221 in unity-control-center (Ubuntu) "Shouldn't it be "menu bar" rather than "top bar"?" [Medium,New]
[16:27] <seb128> https://bugs.launchpad.net/ubuntu/+source/unity-control-center/+bug/1283150
[16:27] <ubot2`> Launchpad bug 1283150 in unity-control-center (Ubuntu) "Missing article in the new LIM option dialog" [Low,Triaged]
[16:28] <seb128> mpt, they are pretty trivial, just issues about the wording on the new options for unity (local menus and UI scaling for Hi-DPI)
[16:28] <mpt> ogra, sorry, I have no idea how the phone HUD gets its commands, or why Quit is anything other than a troubleshooting function. tedg would be able to point you in the right direction.
[16:28] <ogra> mpt, well, i'm seeing MPs everywhere where apps implement the HUD quit function now
[16:29] <ogra> i thought there was a design decision that they should re-introduce that (just judging by the amount of MPs)
[16:29] <tedg> mpt, It was a request so that there is a searchable command to quit.  When it is in the toolbar (which it will be) it won't be on the first list.
[16:29] <tedg> ogra, It was a design decision by our lead designer.
[16:30] <ogra> we have someone who designs lead ?
[16:30] <ogra> is there also a silve and gold designer ?
[16:30] <mpt> tedg, we’re getting rid of the toolbar, so I doubt that
[16:30] <ogra> :P
[16:33] <tedg> mpt, Ah, hadn't heard that part of the story yet.
[16:34] <tedg> If only we had something written down, we could call it a specification, that we could get notified when it changes.
[16:34]  * tedg dreams of the future of technology
[16:34] <ogra> that would kill all these little surprises !
[16:34] <ogra> how boring
[16:34] <mpt> tedg, I’ve been working on it and mentioned it last week IIRC. The new header element will allow buttons, so we can put Back in the header instead of in the toolbar, so we’ll barely need the toolbar any more.
[16:35] <ogra> mpt, header like in ... at the top of the screen ? i.e. the space i cant reach with my thumb ?
[16:35] <tedg> mpt, Toolbar of the app or toolbar of the HUD?
[16:35] <mpt> tedg, the HUD has its own toolbar??
[16:36]  * ogra prefers the toolbar at the bottom to not have to use two hands ... 
[16:36] <ogra> (even though i admit its butt ugly)
[16:36] <tedg> mpt, What do you call the buttons at the bottom of the HUD?
[16:36] <mpt> ogra, if there are parts of the screen you can’t reach with your thumb, you have bigger problems than just the Back button
[16:36] <mpt> tedg, lost.
[16:36] <ogra> mpt, i cant reach most of the top of the screen with my thumb when holding the N4 naturally
[16:37] <seb128> no indicators for you then!
[16:37] <ogra> i probably have an unusually short thumb, dunno
[16:37] <tedg> mpt, Heh, but that's what I was meaning by having Quit "in the toolbar" not the app one.
[16:37] <ogra> seb128, i need the second hand for them usually
[16:37] <seb128> nowadays phone are ridiculous
[16:37] <ogra> yeah
[16:37] <seb128> it took me ages to find a 4" one
[16:37] <ogra> and meizu and bq are even bigger than N4
[16:37] <seb128> I don't want 4.5" or 5"
[16:37]  * ogra is fine with "up to 4.5"
[16:38] <seb128> 4.3" ftw
[16:38] <ogra> but 4.7 (like the N4) or 5 i cant easily reach the top panel anymore
[16:38] <seb128> but as mpt said, you have other issues than the toolbar then
[16:39] <seb128> you have no access to indicators either, which have lot of the common functions you need
[16:39] <ogra> right
[16:39] <seb128> e.g calendar events, messages etc
[16:39] <ogra> well, i do if i use a second hand
[16:39] <ogra> but it gets uncomfortable
[16:39] <mpt> tedg, ok. I guess I need a “How to design for the HUD” tutorial.
[16:40] <seb128> mpt, btw did you see the bugs I listed? no hurry, just making sure they didn't get lost in the ping flood that is hitting you ;-)
[16:40] <qengho> Weeeeird.  I had an alarm on my trusty-Gnome3 machine, and I don't know what's making the sound or how to turn it off.
[16:41] <seb128> GNOME3, like gnome-shell?
[16:41] <qengho> Something calling itself "alarm" plays something through pulse every 3 seconds or so. Sounds like a telephone ringing.
[16:41] <qengho> seb128: yes, shell.
[16:41] <seb128> is indicator-datetime-service running for you?
[16:41] <seb128> (it shouldn't under shell)
[16:42] <qengho> Yes, it is.
[16:42] <seb128> how come?
[16:42] <seb128> that's what is playing the sound
[16:42] <seb128> you should have received a notification that let you dismiss it though
[16:42] <qengho> I don't know why it is. I didn't run it manually.
[16:42]  * qengho greps around.
[16:42] <seb128> did you check your shell notifications (they are hidding at the bottom of the screen I think)
[16:44] <qengho> seb128: yes, there was something in bottom left. Nothing like "dismiss". Just an on/off, as I recall. Not well labeled, I think.
[16:44] <Laney> It's probably because the xdg autostart file is dodgy atm
[16:45] <Laney> Exec=/usr/lib/x86_64-linux-gnu/indicator-datetime/indicator-datetime-service
[16:45] <Laney> NotShowIn=Unity;
[16:54] <mpt> seb128, sorry, ran out of time, I’ll look tomorrow
[16:57] <seb128> mpt, sure, thanks, have a nice evening!
[17:31]  * Laney blurgs
[17:31] <Laney> finally it works
[18:00] <seb128> Laney, well done!
[18:00] <Laney> it ate too much time
[18:00] <Laney> but yeah, good to win in the end
[18:04] <seb128> I know that sort of feeling
[21:33] <tedg> Is there a way I can configure pulse to not stop when I switch to the greeter?
[21:33] <tedg> i.e. Rhythmbox can continue to play
[21:36] <tedg> mterry, Do you know what's going to happen on the phone with the split greeter there? ^
[21:37] <ochosi> oh, that'd be interesting to know, light-locker (default on xubuntu) is also somewhat affected by this
[21:37] <ochosi> (one of the more inconvenient downsides of using the greeter as lockscreen)
[21:37] <seb128> tedg, that's the main reason why we didn't change to use lightdm as a lock screen on the desktop btw
[21:37] <ochosi> seb128: security-wise it isn't so bad though to have the greeter/lockscreen on a separate VT
[21:37] <tedg> seb128, Does pulse have to be a system service then?
[21:38] <tedg> Or can we whitelist lightdm user?
[21:38] <seb128> tedg, I guess it would (it has a mode for that iirc, but it's not recommended to use it for reason I don't remember)
[21:38] <tedg> i.e. only pause if going to a user session other than lightdm.
[21:38] <ochosi> tedg: what happens if >1 users have a pulse-session running?
[21:38] <seb128> well, the issue is that your user session/seat stops being active
[21:39] <seb128> I don't think we have code to handle "unactive but still active"
[21:39] <tedg> ochosi, Last one continues to play.
[21:39] <seb128> mdeslaur might know more about it
[21:39] <seb128> we discussed it a bit in the past
[21:39] <tedg> Interesting, so we don't necessarily know why it became inactive.
[21:40] <tedg> I'm sure mdeslaur wants to write a really cool apparmor profile for us.
[21:40] <ochosi> have to say i see benefits of stopping audio-playback on locking (especially in a multi-user or office-like environment)
[21:40] <tedg> :-)
[21:40] <tedg> ochosi, We're putting the controls for the audio player on the greeter (that's what I'm working on) :-)
[21:40] <seb128> tedg, do you know what GNOME is doing?
[21:41] <seb128> it's working for them
[21:41] <tedg> seb128, Well, not really a gnome issue... it'd be a fedora one.
[21:41] <seb128> no, it's gdm
[21:41] <ochosi> tedg: well it works already with unity-greeter, but something in indicator-sound is fishy. using unity-greeter in xubuntu i can start gmusicbrowser in unity-greeter(!)
[21:41] <seb128> that's working in Ubuntu GNOME with gdm afaik
[21:41] <tedg> Hmm, interesting.
[21:41] <seb128> but I think their lock screen is on the user session
[21:41] <ochosi> yup
[21:41] <ochosi> gdm is on the same VT afaik
[21:42] <tedg> Ah, so they're just running gdm in the user session.
[21:42]  * ochosi still wonders how they do it
[21:42] <seb128> "just"
[21:42] <seb128> I'm not sure it's that simple
[21:42] <ochosi> frankly, we're considering to run greeters in the user-session with light-locker too
[21:42] <ochosi> that would also solve that problem
[21:42] <seb128> ochosi, how would you do that?
[21:42]  * tedg hand waves a couple more solutions
[21:42] <ochosi> seb128: basically emulate lightdm
[21:43] <ochosi> (not saying this is easy or done yet, so far there's only a prototype)
[21:43] <seb128> (right)
[21:43] <seb128> well it's a problem that's common to most desktops
[21:44] <seb128> so we should share the solution
[21:44] <ochosi> absolutely
[21:44] <seb128> we just need the right people to design the architecture
[21:44] <ochosi> if lightdm could provide a sort of liblightdm ...
[21:44] <seb128> it does?
[21:45] <seb128> https://launchpad.net/ubuntu/+source/lightdm
[21:45] <seb128>  liblightdm-gobject-1-0: LightDM GObject client library
[21:45] <seb128>  liblightdm-qt-3-0: LightDM Qt client library
[21:45] <seb128>  liblightdm-qt5-3-0: LightDM Qt 5 client library
[21:45] <seb128> that's one the initial reasons for lightdm
[21:45] <seb128> have the greeters/UI properly split for the server
[21:45] <ochosi> hm, but does that allow you to do what we want?
[21:45] <seb128> with a well defined interface to writer greeters
[21:45] <ochosi> (frankly haven't looked at that yet)
[21:46] <TheMuso> If pulse is run in system mode, it doesn't use shm between server and client, I believe there may also be realtime priority issues there too, although not 100% sure.
[21:46] <TheMuso> In any case, system mode is meant to be used only in specific use cases.
[21:47] <tedg> TheMuso, Seems it could be layered, like a system compositor type solution?
[21:47] <ochosi> seb128: slightly OT, but how did it go with your new unity-lockscreen? did everything land in time?
[21:47] <TheMuso> tedg: Urm, I'
[21:47] <TheMuso> tedg: Urm, I'm not sure how that would work...
[21:48] <mterry> tedg, uh, no...  I thought the phablet user had explicit audio group permissions for that reason.  Let me test
[21:48] <TheMuso> The phablet user should not be in teh audio group, logind handles sound device node permissions.
[21:49] <mterry> TheMuso, I know.  But it was originally in the audio group because we weren't using logind in early days.  Now it is still in the group for that historic reasons.  But it might be useful in allowing it to keep playing audio
[21:49] <TheMuso> But would that not be a possible security risk, particularly if an input device is active?
[21:50] <seb128_> sorry, got disconnected
[21:50] <seb128_> not sure was got through, I was writing
 (frankly haven't looked at that yet)
 well, why would you need to run anything
[21:50] <seb128_>  you are just saying "unity-greeter should run as $user rather than as $lightm-user"
[21:50] <seb128_>  you can run the greeter in test-mode in your session
[21:50] <seb128_>  not sure if that has limitation for e.g pam use though
[21:50] <seb128_>  you also need to make sure that the UI can't ever loose focus
[21:50] <ochosi> oh, that didn't end up here anymore, thanks for repasting
[21:51] <ochosi> indeed, those things all have to be taken into consideration
[21:51] <seb128> robert_ancell is off today, but he probably know better about the topic
[21:51] <ochosi> i think we have a slight head-start on the "can't ever lose focus", because light-locker is a gnome-screensaver fork
[21:52] <ochosi> so the stuff to keep a window "on top of everything, no matter what" is already implemented
[21:52] <seb128> rigt
[21:52] <seb128> h
[21:52] <seb128> shrug
[21:52] <seb128> rigtht, but it's quite a bunch of hacks
[21:52] <ochosi> i guess some PAM stuff needs to be don
[21:52] <ochosi> e
[21:52] <seb128> ideally the lock screen would be part of the wm/compositor
[21:53] <ochosi> that would mean waiting for wayland/MIR
[21:53] <tedg> http://www.freedesktop.org/wiki/Software/PulseAudio/Documentation/User/WhatIsWrongWithSystemWide/
[21:53] <ochosi> which is not ideal for xubuntu at least
[21:53] <seb128> which is what https://code.launchpad.net/~andyrock/unity/lockscreen/+merge/206291 is doing
[21:53] <ochosi> (after all, xfce is still mostly gtk2)
[21:55] <ochosi> i actually asked you about the progress of unity-lockscreen before you got disconnected :)
[22:04] <ochosi> seb128: so what's the status on the unity-lockscreen? the MR still seems unmerged?
[22:05] <seb128> ochosi, it was still being worked on this week
[22:05] <ochosi> oh, ok, so big FFe
[22:05] <seb128> Laney asked for more details on the ffe but andyrocks did provide them yet
[22:06] <seb128> bregma is tracking it I guess
[22:06] <seb128> or Trevinho
[22:06] <ochosi> mkay
[22:07] <bregma> andyrock was having a bit of build trouble with his changes, we'll see what tomorrow brings
[22:07] <ochosi> keeping my fingers crossed for you guys
[22:08] <seb128> thanks
[22:08] <seb128> (didn't provide*)
[22:09] <ochosi> btw, i saw that larsu landed a nice nautilus patch today, bringing back menus, unfortunately only for Unity
[22:09] <bregma> ochosi, we'll get the problems resolved, we're not pushing hard on that feature this week because of other things going on
[22:09] <ochosi> wouldn't !gnome* make more sense?
[22:09] <ochosi> bregma: great, that's nice to hear!
[22:11] <seb128> ochosi, !gnome would be fine, it's just not our role to decide for others what they want
[22:11] <seb128> but feel free to open a bug/send a patch for that
[22:12] <ochosi> well you brought back menus for other apps as well ;)
[22:12] <ochosi> (evince, gedit, file-roller...)
[22:12] <ochosi> and those also didn't happen only in Unity, that's mainly why i was asking
[22:12] <seb128> we do it for Unity, the implementation might be inconsistent though
[22:12] <ochosi> but you're right, i should've opened a bugreport about it
[22:17] <seb128> bregma, is the focus for your team on bug fix until the LTS?
[22:18] <bregma> seb128, absolutely, there are no features left to add to Unity7 :)
[22:18] <seb128> \o/
[22:18]  * seb128 just noticed that alt-tab flickers now
[22:19] <bregma> there are a lot of bugs :(
[22:19] <seb128> yeah
[22:19] <bregma> high-DPI is going to need a lot of tweaking once people start playing with it
[22:20] <seb128> well, that one is "move the mouse pointer over a launcher icon, so it displays a tooltip, then alt-tab"
[22:20] <seb128> the background flickers on every switch then
[22:20] <seb128> bregma, do you get that one as well?
[22:20] <bregma> just checked on 13.10, not there
[22:21] <seb128> on trusty?
[22:21] <bregma> definitely get it on Trusty
[22:22] <seb128> k, so worth reporting I guess
[22:22]  * seb128 checks launchpad for bugs before doing that
[22:28] <seb128> bregma, Trevinho: https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1285889
[22:28] <ubot2`> Launchpad bug 1285889 in unity (Ubuntu) "rendering issues in alt-tab when launcher tooltips are displayed" [Undecided,New]
[22:36] <mterry> TheMuso, I'm not familiar with the security implications
[22:39] <tedg> mterry, http://www.freedesktop.org/wiki/Software/PulseAudio/Documentation/User/WhatIsWrongWithSystemWide/
[22:39] <mterry> tedg, how does this work on desktop?  Music plays when a user is locked, right?
[22:40] <tedg> mterry, Yeah, but as soon as you click "Switch User" it's gone.
[22:40] <tedg> mterry, Can't quite figure out what's doing that yet.
[22:40] <tedg> Grepping though the pulse source, but not finding any hints.
[22:41] <mterry> tedg, policykit / udev rules about "inactive" vs "active" logins?
[22:42] <mterry> tedg, system mode is different than a user being always in the audio group, right?
[22:42] <tedg> mterry, I figured that the session pulse had to be listening to something to know when it's inactive.
[22:42] <tedg> Yeah, well kinda, you need both.
[22:43] <mterry> tedg, I believe logind adds acls to the audio devices as the user switches active status
[22:43] <mterry> tedg, so it might remove them
[22:43] <tedg> Oh, wow.  That's kinda brutal.
[22:44] <tedg> I thought if you already had the socket open, it'd stay open though.
[22:44] <TheMuso> How so? Pulse has code to check in with logind to monitor sessions of users etc.
[22:44] <mterry> tedg, I don't know about that
[22:44] <tedg> TheMuso, I can't find it :-)
[22:44]  * TheMuso has a loook...
[22:45] <tedg> Or more correctly, I foudn the module, I can't figure out where it shuts off audio.
[22:45] <TheMuso> Hrm I'm thinking of consolekit/.
[22:45] <tedg> I think it's creating a client per user session.  Not looking up.
[22:46] <tedg> So if you have multiple logins there's one sound.
[22:46] <tedg> Curious if ALSA is involved in the VT switching?
[22:46] <TheMuso> Yeah, I think thats what the module is for to keep things unified if there is multiple sessions for the one user.
[22:46] <TheMuso> No, afaik its not.
[22:48] <tedg> "Why" here is some explaination: https://wiki.ubuntu.com/Audio/TheAudioGroup#Why
[22:48] <tedg> Looks like mterry might be correct, that can't be true!
[22:48] <tedg> :-)
[22:49] <mterry> :)
[22:50] <sarnold> I believe those are posix -acls- that logind adds and removes to the filesystem; you should be able to use the traditional unix permissions to do what you want, just like we used to do back in 1995
[22:50] <mterry> tedg, but on phone, if user is in audio group already, that would solve the problem?
[22:50] <TheMuso> Yes but what about a tablet with multiple users? Or even a phone?
[22:51] <tedg> Yeah, seeming so.  Looking some more.
[22:51] <tedg> We might need to make logind a bit more sophisticated here.
[22:53] <tedg> br
[22:53] <tedg> brb
[22:56] <xnox> mterry: i thought just audio group is not enough.
[22:57] <xnox> mterry: play some music, switch user, the music is blocked (it's still playing), since the cgroup based access to the device is revoked (as far as i understand) log back in, and the music continues.
[22:57] <xnox> mterry: thus you need to have the active seat as well, or "be special" like the login manager semantics / hints to pam_systemd
[22:58]  * xnox wishes i understood logind...
[23:05] <mterry> xnox, the acl stuff is removed because logind does it.  But if you were in the audio group all along, I'd expect it to work
[23:05] <mterry> But TheMuso is right that this needs proper thinking for a multi-user environment
[23:05] <mterry> But we might be able to slide by on the phone
[23:05] <mterry> Needs testing
[23:05] <mdeslaur> you definitely don't want the inactive user to still be able to capture audio
[23:05] <mdeslaur> for example
[23:08] <sarnold> or maybe I do :)
[23:09] <mdeslaur> sure, and give everyone root!
[23:09] <mdeslaur> :)
[23:10] <sarnold> no more wheel group either! that's a tool of the man!
[23:11] <mdeslaur> heh
[23:11] <xnox> mdeslaur: well phablet user does have root...
[23:11] <mdeslaur> xnox: what do you mean?
[23:11] <mdeslaur> xnox: you mean the phablet user is in the sudo group?
[23:12] <xnox> mdeslaur: yeah.
[23:12] <mdeslaur> yeah, that's fine...so you don't want _non-admin_ users to be able to capture audio... :)
[23:12] <xnox> ok =)