[04:53] <pitti> Good morning
[05:26] <didrocks> good morning
[06:11] <pitti_> ça va didrocks, as-tu eu un bon we ?
[06:12] <didrocks> pitti_: très bien, on est allé en extérieur et pas mal marché. On a planifié nos vacances grâce au portable + 3G en profitant du soleil :)
[06:12] <didrocks> pitti_: et toi ?
[06:14] <pitti_> didrocks: ça semble bien !
[06:14] <pitti_> didrocks: nous avons été à Dresden à nouveau
[06:14] <pitti_> didrocks: je suis encode là, j'ai un rendez-vous ici ce matin
[06:14] <didrocks> ah, tu rentres quand ?
[06:16] <pitti_> didrocks: midi, après le rendez-vous
[06:16] <pitti_> je vais travailler dans le train
[06:18] <didrocks> comfy train ;)
[06:19] <pitti_> didrocks: mon bureau mobiles :)
[06:22] <didrocks> pitti_: waiting for you to plug a monitor and a keyboard to it then! :)
[06:22] <pitti_> didrocks: not that comfy yet :) but power plugs and wifi
[06:22] <didrocks> well, wifi -> 3G, right?
[06:22] <didrocks> or real wifi?
[06:25] <pitti_> didrocks: I suppose they have a 3G connection in the train of some sort
[06:25] <pitti_> maybe satellite, but I don't know
[06:25] <pitti_> I let Telekom do their stuff, as long as the bits are flowing :)
[06:26] <pitti_> as the user you just see wifi
[06:29] <didrocks> pitti_: and the connexion is decent?
[06:29] <pitti_> didrocks: not perfect of course, but much better than with 2g/3g tethering from my mobile
[06:30] <didrocks> still nice to have that service :)
[07:37] <seb128> good morning desktopers
[07:38] <didrocks> re seb128
[07:38] <seb128> lut didrocks ;-)
[07:40] <Laney> hello!
[07:41] <didrocks> hey Laney!
[07:43] <Laney> hey didrocks
[07:43] <seb128> hey Laney, wie gehts? you are on IRC early today!
[07:44] <Laney> yeah want to send an email first ;-)
[07:45] <Laney> doing good thanks!
[07:46] <Laney> chillaxed at the weekend https://goo.gl/photos/MAbjkKbN2ZkU7uSD6
[07:46] <Laney> how's it going?
[07:53] <seb128> good!
[07:54] <seb128> nice picture ;-)
[08:06] <Laney> we taught him a fun trick
[08:06] <Laney> when he put both of his arms up in the air everyone around copied it and cheered
[08:07] <Laney> he wouldn't stop doing it for the rest of the day
[08:08] <seb128> haha
[08:08] <seb128> sounds like you had some familly fun :-)
[08:11] <seb128> hum, mterry is online, I've a feeling some people have a work week in Europe ;-)
[08:11] <mterry> seb128, heh  :)  I'm in Isle of Man, yeah
[10:59] <seb128> didrocks, pitti_, if a systemd service A Requires/After=B.service and B.service is not installed, should A start or not?
[11:00] <seb128> bregma, hey, is there any unity7 landing scheduled for wily?
[11:00] <seb128> I see that some fixes are approved since may and still not landing
[11:43] <didrocks> seb128: A wouldn't start
[11:43] <didrocks> Requires is a strong dependency
[11:43] <seb128> didrocks, ok, I don't get why the job is in a failed state then...
[11:43] <didrocks> so A will never start if B couldn't get activated
[11:43] <didrocks> what's the message?
[11:43] <seb128> would it start if it owns a dbus name and something try do dbus active it?
[11:43] <seb128> what message?
[11:44] <didrocks> seb128: sudo sytemctl status <service_which_fails>
[11:44] <didrocks> systemctl*
[11:45] <seb128> oh sorry, I got confused
[11:45] <didrocks> not sure what you are asking for, as it seems that B isn't installde in your case?
[11:45] <seb128> I don't get why it tries to start
[11:45] <didrocks> installed*
[11:45] <seb128> it fails because it timeouts
[11:45] <didrocks> ah
[11:45] <seb128> which is expected, since it shouldn't be active
[11:45] <seb128> yeah
[11:45] <seb128> that's powerd.service
[11:45] <didrocks> so powerd.service tries to start and you don't know why, right?
[11:45] <seb128> which Depends on lxc-android-start
[11:45] <seb128> which is not installed
[11:45] <didrocks> and nothing Requires/Wants it?
[11:45] <seb128> yes
[11:46] <seb128> unsure
[11:46] <seb128> how do I tell? ;-)
[11:46] <didrocks> grep -r Requires /lib/systemd/system/*
[11:46] <didrocks> and same for Wants :p
[11:46] <seb128> user friendly :p
[11:46] <didrocks> yeah, I don't know of a reverse depends command
[11:47] <didrocks> (also, do you mind pastebining the service file?)
[11:47] <seb128> let me see what I can do
[11:47] <seb128> it's on the snappy personal image in a vm
[11:47] <seb128> let's see if I can get internet working to pastebin things
[11:48] <didrocks> heh, I bet! :
[11:48]  * didrocks gets some water meanwhile now that I'm back from running
[11:57] <seb128> didrocks, http://paste.ubuntu.com/11756343/
[11:58] <seb128> it's not really clear
[11:58] <didrocks> WantedBy=multi-user.target
[11:58] <seb128> but first 10 lines are systemctl --failed
[11:58] <didrocks> are you sure it's not activated?
[11:58] <seb128> then it's cat powerd
[11:58] <seb128> then status lxc-android-config
[11:58] <seb128> well, powerd gets activated
[11:59] <seb128> which is the issue
[11:59] <seb128> it shouldn't
[11:59] <seb128> there is no android side there
[11:59] <ogra_> powerd should be activated .... but fixed for non android use cases ;)
[11:59] <didrocks> systemctl is-enabled powerd
[11:59] <didrocks> seb128: ^
[12:01] <didrocks> I think I know what you mean now… I bet powerd is enabled on your unit
[12:01] <didrocks> so, if A Requires/After B
[12:02] <didrocks> and is A enabled (through a target, or as a dependency)
[12:02] <didrocks> A will try to be activated
[12:02] <didrocks> if B isn't there, A will fail to active
[12:02] <didrocks> (as you have here)
[12:02] <didrocks> systemd doesn't say "only active A if all requirements are installed"
[12:03] <didrocks> it will activate A, see that it needs B, try to start B, B fails (as not installed), and so A fails because of dependency missing
[12:04] <seb128> didrocks, is-enabled -> enabled
[12:04] <seb128> (sorry, had to restart the vm, took a bit)
[12:04] <didrocks> no worry
[12:04] <didrocks> but yeah, so it's what I meant here ^
[12:05] <didrocks> (back in 5)
[12:05] <bregma> seb128, we have a bunch of Unity 7 fixes in a silo for wily but we haven't yet figured out why there are suddenly so many AP test failures
[12:05] <seb128> bregma, k
[12:09] <Guest84030> seb128: no, in that case A sohuld not start; it would if it was a Wants= instead of a Requires=
[12:09] <seb128> pitti_, hey
[12:10] <seb128> bregma, hum, k, good luck with that
[12:10] <pitti_> meh, what's up with freenode?
[12:10] <pitti_> I can't be "pitti"
[12:10] <didrocks> pitti_: but it will fail as seb128 saw, right? (due to missing Requirements)
[12:10] <seb128> there can be only one pitti!
[12:10] <seb128> didrocks, well, it doesn't fail
[12:10] <seb128> it starts
[12:10] <seb128> then get killed by timeout because powerd doesn't work without android
[12:10] <pitti_> ca va seb128 ! (brb)
[12:10] <didrocks> ah, I misread then
[12:11] <seb128> so it's in failed state
[12:11] <didrocks> that's indeed very weird
[12:11] <seb128> what if the job owns a dbus name and it's dbus activated?
[12:12] <Laney> pitti_: try /msg nickserv release pitti
[12:12] <seb128> it's->is
[12:12] <Laney> if you're authed first
[12:15] <pitti_> Laney: still says "nick pitti is temporarily unavailable"; I already tried ghosting too
[12:15] <pitti_> didrocks: yes, it will fail
[12:16] <seb128> pitti_, you are in IoM this week?
[12:16] <pitti_> seb128: no, just entered the train from Dresden to Augsburg
[12:17] <seb128> oh ok
[12:17] <pitti_> seb128: we were visiting family/friends again, and today I had a long appointment here
[12:17] <seb128> I see
[12:17] <seb128> seems like there is a snappy week in IoM I though you might be there ;-)
[12:19] <pitti_> seb128: nah, I'm not in the snappy business :)
[12:19] <seb128> ;-)
[12:19] <seb128> k, so I don't understand why powerd try to start on that snappy personal issue
[12:19] <seb128> lxc-android-config is not installed and is a Requires
[12:21] <pitti_> seb128: those are separate issues
[12:22] <pitti_> seb128: starting A or not is unrelated to A's dependencies
[12:22] <pitti_> the question is if some target or service wants/requires A
[12:22] <seb128> pitti_, powerd is WantedBy=multi-user.target
[12:22] <pitti_> if powerd has [Install]WantedBy=multi-user.target or so, and gets enabled, it will try to starr
[12:23] <pitti_> seb128: perhaps it's easier to explain what you want to do?
[12:23] <seb128> pitti_, I want to have a clean boot on the snappy personal iso
[12:23] <seb128> atm powerd is listed as failed because it starts, can't work, timeout and creates a fail
[12:23] <pitti_> ok, so it wil always try to start (unless you disable it)
[12:23] <seb128> it can't work because it's a desktop amd64 install and there is no android side
[12:24] <seb128> so it shouldn't try to start if lxc-android-config is missing
[12:24] <seb128> not sure how to state that though
[12:24] <pitti_> seb128: so, what you can do:
[12:24] <pitti_> - disable it
[12:24] <seb128> locally?
[12:25] <pitti_> - ship an override unit or drop-in which disables it
[12:25] <pitti_> - add a Condition* to check if lxc/android is available
[12:25] <seb128> is there a standard Condition stating "if that job is started"?
[12:25] <pitti_> like, ConditionPathIsMountPoint=/srv/lxc/android or whatever
[12:26] <pitti_> or
[12:26] <pitti_> - adjust the postinst to only enable the service in ubuntu touch / with android available
[12:26] <pitti_> no, and that wouldn't help you
[12:27] <ogra_> in any case file a bug so it learns to handle ACPI stuff
[12:27] <seb128> why wouldn't it?
[12:28] <seb128> ogra_, right
[12:28] <pitti_> seb128: because chances are high that it's not yet started at the time when powerd.service is considered
[12:28] <ogra_> (it is what the UI talks to about all brightness stuff ... and if available aboout proximity sensor covering etc)
[12:28] <pitti_> (in fact it most probably won't be)
[12:28] <didrocks> (that's what I told: it will try to start without looking at the dependencies…)
[12:29] <didrocks> seems like this didn't get to the wire though :p
[12:29] <pitti_> seb128: if you want to express that, then say it directly:
[12:29] <seb128> didrocks, I guess it's just me not understand the systemd model well enough :p
[12:29] <seb128> upstart \o/
[12:29] <pitti_> WantedBy=lxc-android.service
[12:29] <pitti_> instead of multi-user.target
[12:29] <ogra_> nobody understants  it ... :P
[12:29] <ogra_> that is why pitti needs to write plea-for-help mails :)
[12:29] <didrocks> this is simple: basically there is no magic
[12:30] <didrocks> (for this)
[12:30] <pitti_> seb128: perhaps that would be cleanest, if we only ever want to start it with android?
[12:30] <pitti_> well, you say "please always start powerd" and then complain that it does :)
[12:30] <seb128> well, "wait for that job to start to start" is not magic
[12:30] <didrocks> yeah, it's explicit, as you told it in Requires/After
[12:31] <didrocks> explicitely
[12:31] <seb128> well, requires/after is not enough to state that
[12:31] <pitti_> seb128: yes, that's After=
[12:31] <seb128> since it starts despite the other one not being started
[12:31] <seb128> pitti_, powed has Requires/After lxc
[12:31] <didrocks> that's the part I don't understand
[12:31] <didrocks> are you sure the daemon starts?
[12:31] <seb128> yes
[12:31] <seb128> status is clear
[12:31] <didrocks> or it's just trying to activate the service
[12:31] <seb128> it starts and hit the timeout and get killed
[12:31] <didrocks> you didn't get us the status output
[12:32] <seb128> need to restart the vm
[12:32] <seb128> but it stats the daemon
[12:32] <pitti_> what it says now is:
[12:32] <seb128> but the daemon hangs because no android side
[12:32] <seb128> and gets killed by timeout
[12:32] <seb128> because it never daemonize
[12:32] <pitti_>  - always start powerd (WantedBy=multi-user.target)
[12:32] <pitti_> and
[12:33] <didrocks> seb128: I guess the service starts, but not the daemon
[12:33] <didrocks> and the service waits for lxc-android-config to start
[12:33] <seb128> http://bazaar.launchpad.net/~phablet-team/powerd/trunk/view/head:/debian/powerd.service
[12:33] <didrocks> and fails after a timeout
[12:33] <pitti_>  - I need lxc-android.service (Requires=) and start after it (After=)
[12:33] <pitti_> apparently that's not what we want then?
[12:33] <pitti_> (sorry, I really don't know much about powerd, and whether it should be started in any other use case)
[12:33] <seb128> pitti_, btw you added that service according to the blame :p
[12:33] <pitti_> my guess is that WantedBy=lxc-android.service and After=lxc-android.service should be what you want?
[12:34] <seb128> would WantedBy state that lxc-android wants powerd?
[12:34] <seb128> which is not the case, lxc-android doesn't care about if powerd is there or no
[12:34] <seb128> not
[12:35] <pitti_> seb128: yeah :) (cf. "I don't know much when it shold be started)
[12:35] <seb128> my understand is that it should be started on systems with an android side
[12:35] <seb128> because that's what it talks to
[12:35] <didrocks> I guess the ConditionPathExists= would be the best bet…
[12:36] <seb128> that feels hackish
[12:36] <seb128> like I've no clue is the path is stable or bound to change
[12:36] <pitti_> it states that you want to start it when lxc-android.service starts
[12:36] <seb128> "it"?
[12:36] <pitti_> and it sounded like that's what you wanted
[12:36] <pitti_> (ICBW)
[12:36] <seb128> it does
[12:37] <seb128> the issue atm is that it starts on systems where lxc-android.service is not installed
[12:37] <pitti_> start on started dbus and android
[12:37] <pitti_> respawn
[12:37] <seb128> that's upstart syntax?
[12:37] <pitti_> that's the upstart job, which is fairly equivalent to that
[12:37] <pitti_> (dbus is implied, you don't need to declare that)
[12:38] <seb128> well, the upstart job start only on "android"
[12:38] <pitti_> yes, that's from debian/upstart
[12:38] <seb128> which is what we want
[12:38] <seb128> the systemd job stats atm despite having no android afaik
[12:38] <seb128> starts*
[12:38] <seb128> http://bazaar.launchpad.net/~phablet-team/powerd/trunk/view/head:/debian/powerd.service
[12:39] <seb128> would deleting the WantedBy work?
[12:39] <seb128> if that's what makes it start atm
[12:39] <didrocks> well, it will not start then on android system
[12:39] <pitti_> seb128: right, so changing WantedBy to lxc-android.service (please check the correct name) should be fine
[12:39] <seb128> k
[12:39] <pitti_> seb128: just deleting it will then not make it start at all, unless something else requires/wants powerd
[12:39] <seb128> but lxc-android doesn't want it
[12:39] <seb128> so it would work
[12:40] <seb128> but it feels like a wrong workaround
[12:40] <pitti_> (which might be the case)
[12:40] <seb128> so there is no way in systemd to state "start if that job is started"?
[12:40] <seb128> or is that WantedBy?
[12:40] <seb128> the naming is confusing if that's the case
[12:40] <didrocks> this is the WantedBy (conterpart of Wants)
[12:40] <seb128> it suggests the other job wants yours
[12:40] <didrocks> counterpart*
[12:40] <seb128> when the other job doesn't care
[12:41] <didrocks> well, it's the only way to achieve what you want, that or Condition
[12:41] <pitti_> seb128: well, we don't have a target for android support, so that's the closest thing
[12:41] <seb128> k
[12:41] <seb128> didrocks, pitti_, thanks
[12:41] <seb128> that wfm
[12:41] <seb128> the naming is just counter intuitive
[12:42] <pitti_> seb128: that's WantedBy=, more or less (it's usually being used with targets, where reading it makes more sense)
[12:42] <seb128> they should have called if "IfJobStarted" or something
[12:42] <pitti_> in a dependency model you can't do "start me if that other thing starts", that's the event-based upstart model
[12:43] <seb128> pitti_, well you can condition on the existence of a while, which can be created by another job
[12:43] <seb128> so I don't get why you couldn't condition on the status of another job
[12:43] <seb128> it's just another random condition
[12:43] <seb128> of a file*
[12:43] <pitti_> seb128: yes, you can also use conditions
[12:44] <seb128> but there is no builtin condition "is that job active"
[12:44] <seb128> ?
[12:44] <pitti_> not quite
[12:44] <seb128> k
[12:44] <pitti_> no, there isn't
[12:45] <pitti_> it wouldn't do what you think it does
[12:45] <pitti_> or you had to start everything serially
[12:45] <seb128> well, you would have to block in the case where you depends from something else to be ready
[12:46] <seb128> but if that's what you want/need...
[12:46] <pitti_> well, it could be done with "running or will be run in the current transaction"
[12:46] <pitti_> but anyway, it's rather counterintutitive in a dependency based model
[12:46] <seb128> yeah, I guess I just didn't adapt to the the dependency model yet
[12:47] <seb128> that feels less natural to me atm
[12:47] <seb128> I might eventually get used to it ;-)
[12:47] <seb128> didrocks, pitti_, thanks
[12:48] <pitti_> seb128: thanks to you too
[12:50] <pitti_> sorry for having misunderstood powerd when I wrote that service
[12:50] <seb128> no worry
[12:50] <seb128> ideally powerd should work on linux !android systems as well
[12:50] <pitti_> ok, back to autopkgtest-cloud :)
[12:51] <seb128> but it doesn't atm
[14:35] <Sweet5hark> desrt, larsu, seb128: http://pranavk.github.io/open-source/initial-preview-of-libreoffice-integration-with-gnome-documents/ <- libreoffice running in a gnome-documents container ...
[14:36] <Sweet5hark> (even with basic selections/editing)
[14:36] <seb128> cool I guess
[14:36] <seb128> you are doing something similar for our touch viewer, right? ;-)
[14:38] <Sweet5hark> seb128: I throwing a hint over the fence and hope someone who already knows about QML and foo will pick it up from there, yeah ;)
[14:38] <seb128> right ;-)
[14:38] <seb128> let's see how that goes
[14:38] <Sweet5hark> s/hint/proof of concept/
[14:38] <seb128> Sweet5hark, so you gave the example to dpm&co?
[14:39] <Sweet5hark> seb128: working on it right now.
[14:39] <seb128> great
[14:39] <seb128> k, need to restart machine to test some iso, back in a bit
[15:21] <dpm> excellent, thanks Sweet5hark!
[15:38] <Sweet5hark> dpm: FYI, I get it to load a document and to report that the document is a writer document (so it should already be done with the parsing etc.) for that ... however I get a segfault after that, dunno why yet.
[15:38]  * Sweet5hark installs libreoffice-dbg.
[15:59] <dpm> thanks for following up Sweet5hark
[16:01] <Laney> Sweet5hark: looks like we got new wps from an autosync - https://jenkins.qa.ubuntu.com/view/Wily/view/AutoPkgTest/job/wily-adt-libreoffice/91/ARCH=i386,label=adt/console now fails
[16:02] <Laney> can we take http://anonscm.debian.org/cgit/pkg-openoffice/libreoffice.git/commit/?id=99bd1e8ec3736c450b32f09388787a70e2dbbf95 or similar or are you going to upload a new version?
[16:32] <Sweet5hark> dpm, seb128: *squeee*
[16:32]  * Sweet5hark just rendered a document into a png
[16:33] <seb128> nice
[16:34] <Sweet5hark> ca. 20 lines of C++11 to render the image into the buffer. ca. 60 lines of C to tell libpng to write it to a file.
[16:38]  * didrocks1 waves good evening
[16:41] <larsu> Sweet5hark: you're comparing a language from a couple of years ago with a 20-year-old library...
[16:41] <larsu> (which has a really bad API)
[16:50] <Sweet5hark> larsu: it wasnt meant as a comparison between C/C++ really. Rather just to show that the 'LibreOffice-specific' stuff to use this is just some 20 lines, the rest being boilerplate for the demo.
[16:56] <larsu> fair enough :)
[17:17] <dpm> Sweet5hark, great. Let me or popey know once you've got the final example
[17:31] <Sweet5hark> dpm: yep, will tindy it up a bit and add some documentation, you will have it tomorrow.
[17:43] <dpm> thanks Sweet5hark!
[18:38] <dupingping> awesome notes editor: http://sourceforge.net/projects/ubuntusticky/files/ubuntu-sticky-trial_1.0.0-0ubuntu1_i386.deb/download
[18:39] <GunnarHj> darkxst: Hi Tim, did you notice my comments at bug #1440275?
[18:47] <Sweet5hark> dpm: you've got mail ;)
[22:55] <robert_ancell> Enabled Inbox for work email today. So. Much. Easier. :)
[23:29] <robert_ancell_> rsalveti, I think I'm set as the bug supervisor for libhybris in Launchpad. Can you remove me from that? https://launchpad.net/libhybris/+configure-bugtracker