[01:14] <sethj> andyrock, I was able to reproduce the lock screen/keyboard layout bug. It's not 100% consistent, but I did get it to trigger two of out 5 times.
[01:33] <sethj> I take it the EOL scripts haven't been run as I'm finding a lot of really old bugs..
[05:14] <hikiko> hello
[05:30] <pitti> Good morning
[06:14] <didrocks> hello!
[06:23] <pitti> bonjour didrocks !
[06:23] <didrocks> ça va pitti ?
[06:24] <pitti> didrocks: ça va bien, et toi ? Dieu merci c'est vendredi :-)
[06:24] <didrocks> pitti: en effet, c'est vendredi ! Moi, je suis encore malade
[06:24] <didrocks> pitti: j'ai le nez bien bouché et je me sens comme dans du coton
[06:25] <didrocks> (oreilles bouchées, cerveau moyennement actif…)
[06:25] <pitti> didrocks: hm -- replace work chair with a hot bath?
[06:26] <didrocks> pitti: would love to, however, yesterday evening, I was given some snappy stuff to do before this afternoon…
[06:26] <didrocks> pitti: manager's preparation for next week…
[06:26] <pitti> oh noes
[06:27] <seb128> good morning desktopers
[06:27] <pitti> bonjour seb128 !
[06:27] <seb128> hey didrocks pitti
[06:27] <seb128> pitti, wie gehts?
[06:27] <didrocks> re seb128!
[06:30] <pitti> seb128: prima, danke!
[06:30] <pitti> seb128: as-tu encore reçu le stollen ?
[06:30] <seb128> pitti, non, pas encore
[06:33] <TheMuso> Hey desktoppers. :) Heading off for the weekend, and indeed for a week off. You see you folks in a week or so. :)
[06:33]  * TheMuso -> EOD
[06:33] <seb128> hey TheMuso, enjoy your w.e!
[06:34] <didrocks> enjoy your week off TheMuso!
[06:34] <seb128> and your holidays as well
[06:34] <larsu> good morning!
[06:34] <larsu> bye TheMuso :)
[06:35] <didrocks> great morning larsu
[06:36] <larsu> salut didrocks
[06:36] <larsu> ça va?
[06:36] <seb128> hey larsu
[06:36] <didrocks> larsu: bof, toujours malade…
[06:36] <larsu> didrocks: ugh, sorry to hear :/
[06:37] <larsu> seb128: hola! que tal?
[06:37] <seb128> larsu, prima, danke!
[06:38] <FJKong> good afternoon
[06:40] <seb128> hey FJKong
[06:40] <larsu> hi FJKong
[06:40] <seb128> just as a note, I'm going to be offline in the afternoon
[06:40] <larsu> Friday \o/
[06:40] <seb128> travelling, so working a bit offline and going to catch up a bit after arriving
[06:40] <FJKong> seb128: larsu hey~
[06:41] <didrocks> afternoon FJKong
[06:41] <didrocks> seb128: safe driving!
[06:41] <seb128> didrocks, thanks :-)
[06:41] <FJKong> it seems that I need to learn more about golang if we work on snappy.
[06:41] <FJKong> right/
[06:42] <FJKong> s/\/?
[06:44] <seb128> FJKong, not everybody is going to work on snappy
[06:44] <seb128> but you can have a look to go if you want, doesn't hurt
[06:45] <FJKong> just look at some demo apps, I find most of them are written with golang
[07:52] <willcooke> good morning you fine fellows
[07:53] <hikiko> hello willcooke
[07:53] <willcooke> morning hikiko.  That bug from yesterday, I can't reproduce it, so I dont know why I marked it as rls-w-incoming
[07:53] <willcooke> you can close if you want
[07:54] <hikiko> thanks willcooke :)
[07:54] <seb128> hey willcooke hikiko
[07:55] <willcooke> morning seb128
[07:55] <larsu> morning all
[07:55] <willcooke> g'day larsu
[07:56] <larsu> this is n²...
[07:56] <larsu> at least it comes in waves :)
[07:58] <willcooke> I prefer to think of it as The Waltons effect
[07:58] <larsu> heh
[08:00] <willcooke> http://www.ubuntu.com/usn/usn-2743-3/
[08:00] <willcooke> seb128, did you see that ^^ ?
[08:01] <hikiko> hi seb128 larsu et al
[08:01] <seb128> willcooke, saw now, "fun"
[08:03] <Laney> yo
[08:03] <seb128> hey Laney
[08:04] <didrocks> good morning willcooke, hikiko, Laney!
[08:05] <didrocks> 3 in a row \o/
[08:05] <hikiko> hi didrocks :)
[08:07] <seb128> Laney, seems like we got adwaita-icon-theme-full on the iso, unsure if that's wanted?
[08:07] <seb128> (I guess not, otherwise the split becomes useless?)
[08:07] <Laney> yes
[08:08] <Laney> you looking at it?
[08:11] <seb128> not yeah, I don't have it installed on my desktop, so probably a recommends somewhere
[08:11] <seb128> need to download a daily iso I guess ;-)
[08:11] <larsu> hi Laney :)
[08:11] <larsu> pesky recommend!
[08:12] <larsu> *recommends
[08:12] <Laney> hey larsu and didrocks and seb128 btw :)
[08:12] <Laney> seb128: I would usually check http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.wily/all first
[08:12] <Laney> (probably don't open that in your browser)
[08:12] <Laney> (firefox...)
[08:13] <didrocks> btw, if anyone wants to debug the media keys not working, I'm again in a session with that state
[08:13] <didrocks> as yesterday
[08:13] <didrocks> as the day before…
[08:13] <didrocks> and such :p
[08:13] <willcooke> hey didrocks, how are you feeling today?
[08:13] <didrocks> Trevinho: once you are here, I thought you were looking into this race? ^
[08:14] <didrocks> willcooke: worse than yesterday, speaking heavily from the nose…
[08:14] <willcooke> :(
[08:14] <didrocks> (add that to a French accent :p)
[08:14] <willcooke> ;)
[08:14] <larsu> lol
[08:14]  * larsu hugs didrocks 
[08:14]  * didrocks hugs larsu back (IRC germ firewall FTW \o/)
[08:16] <seb128> larsu, Laney, do we want gnome-themes-standard on the iso?
[08:16] <seb128> it's what pulls in adwaita-icon-theme-full
[08:17] <Laney> why is that there?
[08:17] <larsu> seb128: I thought this was not needed anymore?!
[08:18] <larsu> maybe it has the gtk2 theme?
[08:18] <Laney> calm
[08:47] <Trevinho> hi
[08:47] <Trevinho> didrocks: media keys...
[08:47] <Trevinho> mh
[08:48] <didrocks> Trevinho: media and shorcuts
[08:48] <didrocks> like ctrl+alt+t, and such
[08:48] <Trevinho> yeah, the code is the same
[08:48] <didrocks> remember, we talked about a race that started to happen last cycle if I'm right?
[08:48] <didrocks> yeah
[08:48] <Trevinho> I don't remember we did... But might be :)
[08:48] <didrocks> so, I'm in that state currently (happened quite more recently than before)
[08:49] <Trevinho> last cycle the API changed, but I never got issues because of tat
[08:49] <Trevinho> that*
[08:49] <didrocks> like once every 4 login
[08:49] <didrocks> willcooke experienced it for sure
[08:49]  * willcooke me toos
[08:49] <Trevinho> didrocks: when it happens have you tried to give dbus-monitor a check?
[08:49] <didrocks> Trevinho: what do you want me to monitor in particular?
[08:49] <willcooke> heh, dbus monitor is broken, right larsu :)
[08:50] <Trevinho> well, see what's dispatched and how
[08:50] <Trevinho> well, mh it's on interface org.gnome.Shell
[08:51] <Trevinho> signal AcceleratorActivated
[08:51]  * didrocks stops the workrave
[08:51] <didrocks> ok, let's see
[08:51] <didrocks> because workrave is spamming dbus
[08:51] <Trevinho> but... well, that might not work if it has not been registered
[08:51] <didrocks> Trevinho: ok, signal is sent
[08:51] <Trevinho> didrocks: ahhh... actually better, maybe trying to initctl restart unity-settings-daemon?
[08:51] <didrocks> http://paste.ubuntu.com/12552555/
[08:51] <didrocks> for instance
[08:51] <Trevinho> no, dono't do that then
[08:52] <didrocks> Trevinho: killing unity-settings-daemon doesn't work normally
[08:52] <Trevinho> So.... registration of keybindings on startup is done, so unity is aware and it does the right thing...
[08:52] <Trevinho> now, I guess it's still u-s-s that should receive the AcceleratorActivated signal and do what expected
[08:52] <Trevinho> attente: is this right, correct?
[08:53] <seb128> he's probably sleeping
[08:53] <didrocks> I don't remember what it spawing those, I think you are right, it's u-s-s
[08:53] <didrocks> u-s-d*
[08:54] <Trevinho> yaeah, ...-d :P
[08:55] <didrocks> I can see the MediaKeys interface on the bus
[08:55] <didrocks> org.gnome.SettingsDaemon.Keyboard
[08:55] <larsu> yeah people, don't use dbus-monitor with a destination match
[08:55] <larsu> it's broken
[08:56] <larsu> `busctl --match destination=...` works
[08:56] <larsu> or dbus-monitor and ... err .... grep
[08:56] <larsu> (bug is filed)
[08:56] <didrocks> destination=(null destination)
[08:56] <didrocks> is that supposed to be the case?
[08:56] <didrocks> (it's what we have for a broadcast, I guess?)
[08:58] <larsu> didrocks: sorry, I meant when you give a specific destination
[08:58] <larsu> like, dbus-monitor destination=org.freedesktop.Notifications
[08:58] <larsu> you will *not* see bus traffic to notification daemon
[08:58] <didrocks> larsu: but those media keys don't have any destination in the normal case, right (do you mind checking by giving it a try?)
[08:58] <larsu> it works if you leave off the destination match, or use busctl
[08:59] <larsu> didrocks: in a 1:1 right now. Is there a bug I can get the summary from?
[08:59] <didrocks> as you are in a working state, and I hope to keep my broken state to debug it
[08:59] <didrocks> willcooke opened one, let me check if I can find it
[09:01] <didrocks> bug #1482181
[09:01] <larsu> ah *this* problem
[09:01] <larsu> I've had this
[09:01] <larsu> didrocks: please STAY in the broken state. I'll be with you in a few minutes
[09:02] <didrocks> thx!
[09:02]  * didrocks commented
[09:02] <didrocks> and stay broken
[09:02] <didrocks> (broken and sick, nice Friday! :p)
[09:04] <larsu> didrocks: :1.28 is u-s-d? (busctl status --user :1.28)
[09:05] <didrocks> larsu: no, it's compiz
[09:06] <didrocks> it's the compiz gnome plugin IIRC which handles the keys
[09:06] <didrocks> (as we have some special build process for that)
[09:06] <didrocks> so compiz gets the press event, and broadcast it correctly
[09:07] <didrocks> then, the other side (which should listen on that iface and launch the corresponding app) doesn't work
[09:07] <didrocks> I wonder if that couldn't be a side effect of the "listen to a signal, and then read the key"
[09:07] <didrocks> (in gsettings)
[09:07] <didrocks> on the receiver side
[09:08] <larsu> gsettings shouldn't be involved here
[09:08] <didrocks> sorry, I meant the glib signal change
[09:09] <didrocks> but yeah
[09:09] <didrocks> sorry ECONFUSED ;)
[09:09] <larsu> so we don't use u-s-d media keys plugin?
[09:10] <didrocks> not for listening in the unity settings, no
[09:10] <didrocks> (we never had since unity)
[09:10] <didrocks> it's the compiz-gnome plugin
[09:11] <larsu> media keys plugin is active though..
[09:11] <didrocks> yeah, unsure why…
[09:12] <larsu> maybe these two step on each other's feet?
[09:12]  * willcooke fires up his test laptop
[09:12] <larsu> let me check the bus traffic
[09:12] <larsu> (working for me right now
[09:12] <larsu> )
[09:12] <didrocks> yeah
[09:13] <didrocks> so, what is the listening interface on the other side to receive the signal and spawn a process?
[09:13] <didrocks> sounds like compiz, not that the interface is GnomeShell?
[09:13] <larsu> wow, 10 messages for one key
[09:13] <larsu> oh, with returns
[09:13] <didrocks> I only have one?
[09:14] <didrocks> yeah, just one here
[09:14] <didrocks> and no returns :p
[09:14] <larsu> didrocks: the gnome shell interface is exposed by unity
[09:14] <didrocks> so it's back on Trevinho's plate? :)
[09:14] <didrocks> larsu: yeah, that's what I'm seeing from the .service file
[09:14] <larsu> didrocks: you should at least also get messages to notify-osd and "backlight-changed" or similar
[09:15] <didrocks> larsu: no, nothing, (and of course, no notify-osd)
[09:15] <larsu> this is because it's broken, right?
[09:15] <didrocks> yep
[09:16] <larsu> holy shit whatever is sending the notify thing is asking the daemon about it's capabilities all the time
[09:17] <larsu> *sigh*
[09:17] <didrocks> larsu: just in case, "did you change yet?" :p
[09:17] <larsu> haha
[09:18] <willcooke> Laney, wallpaper judging has begun....
[09:18] <didrocks> so, if I use a Method like UngrabAccelerator(0), I get a return value
[09:18] <Laney> good!
[09:18] <didrocks> there is clearly something listening on the other side
[09:18] <didrocks> but not at signals on that iface?
[09:18] <willcooke> desktoppers:  Take a look at the wallpaper entries... https://www.flickr.com/groups/ubuntu-fcs-1510/pool/
[09:19] <didrocks> larsu: so, seems like to be in Unity code?
[09:19] <larsu> didrocks: you see the signal though, right?
[09:19] <larsu> just not the "result"
[09:19] <larsu> why would unity send itself a signal via dbus, I wonder
[09:19] <didrocks> larsu: yeah, I have the signal, but no response
[09:20] <didrocks> larsu: because it's 2 different plugins handling it? :p
[09:20] <larsu> hah
[09:21] <larsu> hm, unity-settings-daemon is talking to itself
[09:21] <Laney> seb128: did you look at it or should I?
[09:22] <didrocks> it seems that GnomeGrabber::Impl::activateAction isn't called for whatever reason
[09:22] <Laney> seb128: seems either metacity or gnome-themes-standard or both could fix this
[09:22] <didrocks> larsu: I guess it's where we are blocking on Trevinho to have a look now
[09:24] <larsu> sorry, brb
[09:26] <seb128> Laney, I didn't, sorry, busy with something else and travelling this afternoon, feel free if you want
[09:26] <Laney> thx
[09:26] <seb128> otherwise it's likely going to be on monday for me
[09:26] <Laney> np
[09:27] <seb128> thanks
[09:29] <seb128> larsu, didrocks, just for info https://code.launchpad.net/~attente/unity/gnome-key-grabber was the changeset to use those interfaces
[09:29] <seb128> it might make easier to see the code impacted from the diff of the feature landing
[09:30] <seb128> but basically unity should do the grabbing and dbus message u-s-d which does the actions
[09:36] <hikiko> https://code.launchpad.net/~hikiko/compiz/compiz.fix-811591/+merge/272359 Trevinho could you approve this? (it's a bug from ages sam proposed that solution that works for me but it was never added to ubuntu)
[09:36] <hikiko> or andyrock ^
[09:36]  * hikiko gets rid of bugs :p
[09:37]  * Trevinho on it
[09:38] <Trevinho> mh hikiko don't comment code, or if you do add some explaination please. However, I'd just go by dropping those lines. In case bzr annotate or log will help to understand what changed
[09:38] <hikiko> ok :)
[09:39] <hikiko> sec
[09:39] <larsu> seb128: thanks!
[09:40] <larsu> didrocks: you're getting the signal, so I'm guessing the problem is in settings-daemon
[09:40] <larsu> which actually has two plugins talking to themselves on the bus :/
[09:41] <hikiko> Trevinho, fixed
[09:41] <Laney> hi u-s-d, how are you?
[09:42] <Laney> find u-s-d, and you?
[09:42] <Laney> fine*
[09:42] <Laney> great thanks u-s-d!
[09:42] <seb128> didrocks, larsu, isn't media-key what has the handlers for those dbus calls?
[09:42] <seb128> what is the second plugin?
[09:43] <larsu> seb128: the media key plugin calls the org.gnome.SettingsDaemon.Power.Screen.GetPercentage
[09:43] <larsu> and then .StepUp
[09:43] <larsu> but all of this doesn't seem to be happening for didrocks
[09:44] <seb128> why does it call that?
[09:44] <larsu> to chnage brightness (the media key I pressed)
[09:44] <larsu> for volume, I guess it calls pulse directly?
[09:45] <seb128> oh, I see
[09:46] <seb128> ok, good desktop people I need to drop offline, going for lunch and then travelling
[09:46] <seb128> I'm going to work a bit offline and should be back betwee 17-18
[09:46] <larsu> hehe enjoy!
[09:46] <seb128> see you later or have a good w.e for those who call it a day by then
[09:47] <seb128> thanks!
[09:47] <larsu> you too!
[09:47] <willcooke> cya seb128
[09:47] <larsu> didrocks: can you paste the output of `busctl --user call org.freedesktop.DBus /org/freedesktop/DBus org.freedesktop.DBus.Debug.Stats GetAllMatchRules` please
[09:53] <didrocks> larsu: sorry, in a HO
[09:53] <didrocks> just read last line but: http://paste.ubuntu.com/12552889/
[09:56] <larsu> didrocks: no worries, that was the only interesting one
[10:05] <didrocks> larsu: ok, can focus more on this, and opened to any experience that doesn't include physical danger :p
[10:09] <larsu> didrocks: ha ok
[10:09] <larsu> so, you have the match rule
[10:10] <larsu> currently going through u-s-d code to see what could go wrong
[10:10] <didrocks> larsu: but it's not u-s-d which spawn the process, it's only handling the notification communication?
[10:10] <didrocks> meaning, we would have:
[10:10] <didrocks> compiz-gnome -> send signal
[10:10] <didrocks> 1. u-s-d grab it and send notification
[10:11] <didrocks> 2. unity grabs it and launch the corresponding action/process
[10:11] <didrocks> ?
[10:11] <didrocks> (like for ctrl+alt+t, launch the terminal)
[10:11] <larsu> right, unity grabs some keys
[10:11] <larsu> but volume and brightness are handled directly
[10:11] <larsu> they all don't work for you, right?
[10:11] <didrocks> yeah, all
[10:12] <didrocks> so media keys are per see, and other command shorcuts
[10:12] <didrocks> let me try to see if I see different traffic for ctrl+alt+t
[10:12] <didrocks> argh,and of course, I tried ctrl+alt+t to have a terminal to monitor the traffic…
[10:13] <larsu> try volume
[10:13] <didrocks> ok, same with ctrl+alt+t, just getting the broadcast signal
[10:13] <larsu> or brightness
[10:13] <didrocks> volume is what I tried in my previous paste
[10:13] <larsu> ok, then unity is not involved at all
[10:13] <didrocks> so, same, just getting broadcast signal in both case
[10:13] <didrocks> yeah, can be dbus not delivering to any of them?
[10:14] <larsu> hm?
[10:14] <larsu> someone has the match rule, so it should arrive
[10:14] <larsu> and that someone is most likely u-s-d
[10:14] <didrocks> shouldn't we have 2 then?
[10:14] <didrocks> (matching)
[10:14] <didrocks> u-s-d for media keys
[10:14] <didrocks> unity for shorcuts
[10:15] <larsu> does unity also listen on the gnome shell interface?
[10:15] <didrocks> and then, they filter and only care about some part of the answer
[10:15] <larsu> unity *is* the gnome shell interface
[10:15] <larsu> I hope they simply handle Ctrl+Alt+T internally?!
[10:15] <didrocks> well, if this was internally handled, it would work, right?
[10:15] <didrocks> seems like it's using the exact same pipe
[10:15] <larsu> indeed
[10:17] <larsu> maybe u-s-d handles all of those?
[10:18] <didrocks> larsu: maybe (as it's working for you), set some new env variable to u-s-d
[10:18] <didrocks> then ctrl+alt+t
[10:18] <didrocks> and see if the terminal gets them?
[10:18] <didrocks> that would tell us if it's unity or u-s-d spawning the app?
[10:20] <didrocks> Trevinho: maybe you know this workflow, what is supposed to launch a terminal for instance on ctrl+alt+t, is it unity?
[10:21] <didrocks> or does compiz-gnome just grab the keys, and forward them to the gnomeshell iface?
[10:21] <Trevinho> didrocks: unity only grabs the keys, the it's up to u-s-d to launch
[10:21] <larsu> right
[10:21] <didrocks> ok, so at least, none of them working makes sense
[10:22] <didrocks> so, the issue is in u-s-d
[10:22] <didrocks> thanks Trevinho
[10:23] <larsu> didrocks: gsettings reset org.gnome.settings-daemon.plugins.media-keys terminal
[10:23] <larsu> does this make the terminal key work?
[10:24]  * didrocks tries
[10:24] <didrocks> larsu: nope
[10:25] <didrocks> the value didn't change (I did a get before), and still the default: '<Primary><Alt>t'
[10:25] <larsu> ah
[10:25] <larsu> try changing it to something
[10:25] <didrocks> ok
[10:25] <didrocks> larsu: yeah, after changing, it works
[10:25] <didrocks> resetting though, doesn't
[10:26] <larsu> FUCK
[10:26] <larsu> ok
[10:26] <larsu> but resetting after changing keeps it working, right?
[10:26] <didrocks> nope
[10:26] <larsu> oh?
[10:26] <didrocks> I resetted it (and so '<Primary><Alt>t' again)
[10:26] <didrocks> and it doesn't work
[10:26] <larsu> but if yo set it to something else it does?
[10:27] <Trevinho> Weird
[10:27] <didrocks> yep, I tried Alt + t
[10:27] <didrocks> want me to try with the volume keys?
[10:27] <didrocks> to confirm this
[10:28] <larsu> yes please
[10:28] <didrocks> same
[10:28] <didrocks> so setting it to Alt +t, I can increase the volume
[10:28] <didrocks> gsettings reset it
[10:28] <larsu> should be the same changed signal
[10:28] <didrocks> and then, the media key doesn't work
[10:28] <larsu> this is awesome :)
[10:28] <didrocks> isn't it? :p
[10:28] <Trevinho> lovely
[10:28] <didrocks> let me dbus-monitor this
[10:29] <didrocks> to ensure that it's still unity grabbing the key in the second case
[10:29] <didrocks> (when setting it to something else)
[10:31] <didrocks> so: http://paste.ubuntu.com/12553084/
[10:31] <didrocks> and then, getting all the dbus spam with notification, panel service…
[10:31]  * didrocks looks for the send id
[10:32] <didrocks> interesting btw
[10:32] <didrocks> the old key still send a signal
[10:32] <Trevinho> using dbus-monitor --session "interface='org.gnome.Shell'"  does help in filtering?
[10:33] <didrocks> can be, but I see enough here
[10:33] <didrocks> sender=:1.28 in both case
[10:33] <didrocks> still weird that the previous old volume up key send the signal
[10:33] <didrocks> so something is wrong unity-side for sure already
[10:33] <didrocks> oh
[10:33] <didrocks> the first arg has different values
[10:34] <larsu> are you sure? You always get ths signal, regardless of whether the key works, right?
[10:34] <didrocks> yeah
[10:34] <Trevinho> action id maybe is not well defined
[10:34] <didrocks> so, I changed to Alt+t, right?
[10:34] <didrocks> for volume up
[10:34] <didrocks> and then:
[10:35] <didrocks> http://paste.ubuntu.com/12553099/
[10:35] <didrocks> the first value of that signal is the keycode pressed (the values are different) right? it's not a "capability index" like volume up or such?
[10:35] <larsu> no, it's the accel id
[10:36] <didrocks> so, it should be the same between the 2 calls?
[10:36] <didrocks> or at least, it's unique for "volume up"?
[10:36] <larsu> no, u-s-d gets a new one from GrabAccelerator() every time it does a grab
[10:36] <didrocks> ok
[10:36] <larsu> if I understand this correctly
[10:37] <larsu> basically it works like this: u-s-d asks unity to grab "Alt+T" and gets an id
[10:37] <larsu> and the unity sends signals with that id - so that u-s-d knows which accel was hit
[10:37] <didrocks> that's why you thought reset would work, right? like reasking to regrab "volume up"
[10:37] <larsu> yes
[10:37] <didrocks> and the bug would have been u-s-d starting before unity
[10:37] <larsu> indeed
[10:38] <larsu> but since it stops working again when you regrab the old key, this doesn't seem to be the case
[10:38] <didrocks> yeah…
[10:38] <larsu> but we're onto something
[10:38] <larsu> try this:
[10:38] <larsu> dbus-monitor interface=org.gnome.Shell
[10:39] <larsu> gsettings set org.gnome.settings-daemon.plugins.media-keys terminal "<Alt>t"
[10:39] <larsu> gsettings reset org.gnome.settings-daemon.plugins.media-keys terminal
[10:39] <larsu> gsettings set org.gnome.settings-daemon.plugins.media-keys terminal "<Alt>t"
[10:39] <larsu> in between, hit Alt+T and Ctrl+Alt+T
[10:39] <larsu> and send the log of the monitor
[10:39] <didrocks> k
[10:39] <larsu> thanks for being so patient!
[10:40] <didrocks> no worry, I'm eager to see it fixed, this waited for too long :)
[10:43] <didrocks> larsu: here we go, with annotation: http://paste.ubuntu.com/12553139/
[10:43] <larsu> you're the best!
[10:43]  * didrocks hugs larsu
[10:43] <didrocks> so, the weird thing
[10:44] <didrocks> UngrabAccelerator
[10:44] <didrocks> the id isn't the one for Primary alt T
[10:44] <larsu> yeah
[10:44] <larsu> looks like it's getting confused?!
[10:44] <didrocks> yeah
[10:44] <didrocks> if I keep pressing it
[10:45] <didrocks> resetting*
[10:45] <didrocks> I always have a GrabAccelerator -> id 3
[10:45] <didrocks> but an ungrab id incrementing
[10:45] <didrocks>    uint32 736
[10:45] <didrocks>    uint32 737
[10:45] <didrocks> …
[10:45] <didrocks> (and I just gsettings reset in loop, didn't set it anything else in between)
[10:46] <larsu> but also Activated() doesn't send the right accel id
[10:46] <didrocks> yep
[10:46] <larsu> so that's *clearly* wrong
[10:47] <didrocks> indeed
[10:47] <didrocks> and that explains why unity doesn't ungrab
[10:47] <larsu> unless...... there's an old grab that u-s-d forgot to ungrab
[10:47] <larsu> and unity only sends signals for one grab
[10:47] <larsu> this interface is the biggest shit I've ever seen
[10:47] <larsu> so many problems
[10:47] <didrocks> larsu: I'm sure you can always find worse! :)
[10:47] <larsu> lol
[10:48] <larsu> it should work like this: hey unity, I'm interested in these 10 keys!
[10:48] <didrocks> ah, remapping rather than incremental?
[10:48] <larsu> ok :1.xyz, I'll send signals for these keys until you die
[10:48] <larsu> or tell me a new set of keys
[10:48] <didrocks> yeah
[10:49] <larsu> it needs to check lifetime anyway, in case it crashes
[10:49] <larsu> so ungrab() is completely useless
[10:49] <didrocks> indeed
[10:49] <larsu> and apparently causing the problems here
[10:49] <didrocks> I don't have any crash of u-s-d, in case this would have been the issue…
[10:50] <didrocks> like asked for grabbing, crashing and using weird id
[10:50] <didrocks> that's… weird
[10:50] <didrocks> so: here is another try:
[10:51] <larsu> no I think it doesn't crash but gets the ungrabbing wrong
[10:51] <didrocks> http://paste.ubuntu.com/12553213/
[10:52] <didrocks> so, the id for alt + T which is grabbed when pressed is different from the one set? (always 3)
[10:52] <larsu> oh wait, 3 is input, not return
[10:53] <didrocks> ah yeah, it's a flag
[10:54] <larsu> didrocks: sorry I have a lunch date with systemd hackers now. I think I have enough information to figure this out now
[10:54] <larsu> didrocks: thanks a lot again. I might have more questions in an hour or two
[10:55] <didrocks> larsu: no worry, keeping it broken for the day. Thanks for looking at it!
[10:55] <didrocks> enjoy your lunch :)
[10:56] <larsu> man, this totally works for me thougj
[10:56] <larsu> I bet it gets confused at startup for you
[10:56] <larsu> and then stays in some weird state
[10:56] <larsu> didrocks: thanks! Eating bulgarian
[10:56] <didrocks> sounds nice! :)
[10:56] <didrocks> larsu: yeah, I think there is something at startup and that's why it's lost then *forever*
[10:57] <larsu> indeed
[11:15] <willcooke> desktoppers:  https://goo.gl/jqRgRC
[11:16] <didrocks> cute!
[11:16] <Trevinho> nice
[11:17] <Trevinho> Dual landing, then! :(
[11:17] <Trevinho> :) ^
[11:17] <didrocks> ahah
[11:21] <willcooke> :D
[11:28] <attente> good morning
[11:28] <didrocks> hey attente!
[11:29] <attente> hi didrocks!
[11:34] <willcooke> o/ attente
[11:35] <attente> o/
[11:48] <andyrock> good morning!
[11:51] <willcooke> hey andyrock
[12:58] <Trevinho> didrocks: so, I was looking further... And I noticed that the gnomecompat plugin in compiz is still enabled (actually we should remove it from our list nowadays), and that it tries to grab Ctrl+Alt+t... So maybe disabling that plugin (or keybinding) should explain why the activated signal is emitted anyway when chaning setting
[13:00] <didrocks> Trevinho: can be, yeah
[13:00] <Trevinho> Oh, well no... Actually it shouldn't be there /usr/share/session-migration/scripts/00_remove_gnomecompat_in_unity_session.py
[13:00] <Trevinho> (as per this, but maybe it doesn't work?)
[13:00]  * Trevinho tries a clean session
[13:00] <didrocks> it's not enable here
[13:00] <didrocks> enabled*
[13:00] <Trevinho> ah, ok fine
[13:01] <Trevinho> so no troubles from that... It's all in unity
[13:01] <Trevinho> but I don't see unity behave badly... I mean, it should do what expected.
[13:01] <didrocks> yeah, it seems to be more on the u-s-d side maybe
[13:01] <didrocks> or unity doesn't ungrab the first key
[13:03] <Trevinho> didrocks: if you've installed libxpathselect, you might be able to try to run these AP test autopilot/unity/tests/test_gnome_key_grabber.py
[13:03] <Trevinho> they play with the interface...
[13:04] <Trevinho> So you might check if ungrabbing happens also
[13:05] <andyrock> willcooke: congrats!! :D
[13:05] <willcooke> thanks andyrock :)
[13:22] <kgunn> tjaalton: ping
[13:22] <Trevinho> didrocks: oh... Actually I was looking at the ungrab code... And well it's probably not working well
[13:23] <Trevinho> didrocks: did you try to manually call the UngrabAccelerator command with a valid id and see what's returning?
[13:23] <Trevinho> if it's false, as I believe it might do, then... we're into problems
[13:28] <qengho> good morning
[13:29] <willcooke> hey qengho
[13:35] <larsu> Trevinho: well, unity does some things wrong
[13:36] <larsu> for example, it sends the AccelaratorActivated signal to everyone
[13:36] <larsu> even though only one process can know about the id
[13:36]  * larsu wonders if gnome shell gets this right
[13:38] <larsu> ah, seems to
[13:38] <Trevinho> larsu: well, that's right... But since the ID is generated with methods, others shouldn't listen that
[13:38] <Trevinho> however, i was mostly concerned by this didn't work, but it actualyl does http://pastebin.ubuntu.com/12554355/
[13:39] <larsu> well, others get woken up
[13:39] <larsu> and then ignore the message because it doesn't have an id they know about
[13:40] <larsu> could set up a match rule with arg0 set to the id
[13:40] <larsu> ... for every key
[13:40] <larsu> man this interface is dumb
[13:41] <Trevinho> we should emit the id to the register only, true... To do that we should change our glib::DBusObject too, but feasible
[13:54] <didrocks> sorry guys, still on a HO
[14:01] <didrocks> larsu: Trevinho: do you want still me to try to manually call the ungrabAccel (I'm unsure about the first key id which was set though)
[14:04] <Trevinho> didrocks: well, in theory it's the same id you get when you use the key and unity emits the signal
[14:04] <didrocks> Trevinho: oh sure, let me try
[14:05] <didrocks> ok, called ungrabbed
[14:05] <didrocks> returned true
[14:05] <didrocks> pressing and working now
[14:05] <didrocks> with another id
[14:05] <didrocks> so, yeah, the key is registered twice
[14:05] <didrocks> and unity only send the first one
[14:06] <didrocks> so, this starts to make sense
[14:06] <didrocks> larsu: Trevinho ^
[14:06] <didrocks> to sum up:
[14:06] <didrocks> -> key id was 33
[14:06] <didrocks> I ungrabbed(33)
[14:06] <didrocks> press again
[14:06] <didrocks> and now, it's 720
[14:06] <didrocks> and the notification popups and such
[14:06] <Trevinho> Ok, makes sense... So we need to ensure that an action is just registered once
[14:06] <didrocks> yep
[14:07] <didrocks> as it will stop at the first match I guess
[14:07] <Trevinho> and no register again if someone tries to do that
[14:07] <didrocks> but then, it means there is one case where it registers twice
[14:07] <didrocks> without ungrabbing?
[14:07] <Trevinho> unless there's an ungrab (best if from the same who asked to grab, but I wouldn't do that for avoiding trobules :P)
[14:07] <didrocks> (for all keys, with some racy start condition?)
[14:07] <didrocks> yeah
[14:07] <didrocks> ungrab is always called before grabbin?
[14:08] <didrocks> grabbing*
[14:08]  * didrocks looks
[14:08] <didrocks> at least, seems so from the logs
[14:08] <didrocks> so it's supposed to work
[14:08] <Trevinho> larsu: do you want me also to emit the signal only to the one who registered the grab? Although, if u-s-s crashes, we need to re-register every keybinding... it seems too much noise, considering that security isn't there ayway
[14:08] <didrocks> (your strategy to not reregister)
[14:09] <didrocks> good call on the crash case, you will need to track and reregister
[14:09] <didrocks> where broadcast only enable you to "not care"
[14:10]  * Trevinho takes the occasion to semplify this stuff
[14:11] <didrocks> Trevinho: ok, confirmed with another key where I didn't do any reset/set
[14:11] <didrocks> and of course, the id is way smaller (89)
[14:11] <didrocks> doesn't fix the real bug where is this something registering twice at startup though
[14:12] <didrocks> but weird, especially with no u-s-d crash
[14:13] <willcooke> didrocks, so we have got to the bottom of the issue then?
[14:15] <didrocks> willcooke: well, we know how it happens, Trevinho is working on a fix (I guess?), we don't know exactly why it goes to that state sometimes though
[14:16] <willcooke> oki, progress though, thanks all
[14:16] <didrocks> yeah, at least, this should be fixed from an user perspective :)
[14:16] <Trevinho> yeah, working on that
[14:16] <larsu> Trevinho: hm? You need to watch the service anyway to unregister the ids (in case it crashes)
[14:17] <didrocks> would be good to understand exactly why this happens, as this kinds of things are always stricking back in a worse case
[14:18] <Trevinho> larsu: yes, that's true... But I don't think it's something done right now... I was doing that, anyways...
[14:18] <Trevinho> larsu: i guess that's the reason why we didn't replace actions before
[14:20] <larsu> actions?
[14:21] <Trevinho> I mean keybindings... The problem was that we were just adding new keys ignoring the old ones
[14:21] <larsu> right, that's the problem didrocks is seeing?
[14:21] <Trevinho> so, when restarting usd, we were just adding the same keybindings twice
[14:22] <Trevinho> yeah... To reproduce just do something like
[14:22] <Trevinho> gdbus call --session --dest org.gnome.Shell --object-path /org/gnome/Shell --method org.gnome.Shell.GrabAccelerator "<Shift><Control>x" 0
[14:22] <Trevinho> twice or more...
[14:22] <Trevinho> We don't return false
[14:22] <larsu> indeed
[14:23] <larsu> so this is the issue?
[14:23] <larsu> didrocks' u-s-d crashes
[14:23] <larsu> once on startup?
[14:23]  * larsu checks
[14:23] <Trevinho> no, it adds twice the same shortcuts..
[14:23] <larsu> well, it adds them again when restarting
[14:24] <Trevinho> so only the first id is used by ucc, but unity sends the last registered one
[14:24] <larsu> and unity doesn't delete the old ones
[14:24] <larsu> oh
[14:24] <Trevinho> Well, that wouldn't be an issue, as restarting adds new ones...
[14:24] <larsu> it only registers them once for me
[14:24] <larsu> Trevinho: except for the obvious memory leak ;)
[14:25] <Trevinho> here's what I get http://pastebin.ubuntu.com/12554875/
[14:25] <didrocks> larsu: I don't have anything in /var/crash
[14:25] <didrocks> as told
[14:25] <Trevinho> yeah, I just think that for some reason you get the method called twice
[14:25] <didrocks> yep
[14:25] <didrocks> and without ungrab in the middle
[14:26] <larsu> Trevinho: I don't see it twice in didrocks' logs
[14:26] <Trevinho> didrocks: for double checking, you might set UNITY_TEST_LOG_SEVERITY="<root>error;unity.key.gnome.grabber=debug" for your session and see if you get anything relevant
[14:26] <Trevinho> larsu: I guess it happens on startup..
[14:26] <Trevinho> err, sorry wrong env name
[14:27] <didrocks> Trevinho: well, it's not a one 100% startup case, but yeah, I can do that for future logins
[14:27] <Trevinho> didrocks: it's UNITY_LOG_SEVERITY="<root>error;unity.key.gnome.grabber=debug" for your session and see if you get anything relevant
[14:27] <Trevinho> didrocks: yeah, I know, but you might try to relogin tons of times :)
[14:28] <didrocks> Trevinho: unsure about now, I have quite some stuff opened and working on, but yeah, adding for later :)
[14:28] <Trevinho> didrocks: sure, no worries
[14:28]  * Trevinho has an uptime of weeks generally, so... I also never leave my session :)
[14:30] <didrocks> larsu: so, let's say I press volume down, I see the "32" key id being fired
[14:30] <didrocks> larsu: then, I ungrab(32)
[14:30] <didrocks> larsu: I press again, and I get another event with key id 89, everything now works
[14:30] <didrocks> larsu: so, there are been at some point, 2 grab() calls without ungrab
[14:31] <didrocks> and unity only send the first match
[14:31] <didrocks> where u-s-d expects the second
[14:31] <larsu> ya
[14:31] <didrocks> (which is, I guess for it, the only valid one)
[14:31] <larsu> crashing u-s-d is enough to trigger this for me
[14:31] <larsu> I guess for the same reason
[14:31] <didrocks> the weird part is u-s-d didn't crash for me
[14:31] <larsu> Trevinho: you need to watch the name and remove all grabs when it disappears
[14:32] <larsu> s/you/unity :P
[14:32] <larsu> didrocks: thank!
[14:32] <didrocks> so, there is a case where 2 grabs are called, without ungrab in between
[14:32] <didrocks> yw!
[14:32] <larsu> right
[14:32] <larsu> which is the case when it crashes
[14:32] <didrocks> larsu: I do expect Trevinho to look at every laptop and watch for the grabs :)
[14:32] <didrocks> yep
[14:32] <larsu> haha
[14:33] <larsu> Trevinho: do you want to fix it or should we make a better API?
[14:33]  * larsu wonders what else is using this
[14:35] <Trevinho> larsu: well, as you prefer... I was about to use this API as it was the same gnome uses, but we can also make a new one
[14:35] <Trevinho> larsu: what would you prefer? I'd say the new one, right?
[14:35] <larsu> Trevinho: do you know if anyone else is using this interface?
[14:36] <Trevinho> larsu: I was about asking you the same :)
[14:36] <larsu> because if you need to implement it anyways, we might as well keep it
[14:36] <Trevinho> larsu: well... let me see... We use it for menus I believe, but not sure if it's unity module or we do in unity (/me checks)
[14:36] <didrocks> well… we need something for wily though
[14:37] <didrocks> and I don't think you want to introduce the new API now :p
[14:37] <willcooke> yes!
[14:37] <Trevinho> no, we do that internally for menus, so I don't think anyone would use
[14:37] <Trevinho> well... private stuff... :P
[14:37] <willcooke> didrocks, yes, as in, no - no you dont
[14:37] <Trevinho> Ok, I'm reusing the old one....
[14:37] <larsu> good point
[14:38] <didrocks> still, behavior change and such, you can still do a bandaid for now
[14:38] <didrocks> and nothing prevents to work on the new API now :)
[14:38] <didrocks> but first bandaid I would say ;)
[14:39]  * Trevinho hates bandaids...
[14:39] <Trevinho> but yeah...
[14:39] <didrocks> well, as everything, it's a risk/benefit assement :)
[14:40] <Trevinho> larsu: let's decide it now though... as actually I'd avoid to do things twice and probably it's not the best thing to rewrite this in u-s-d and unity. Considering we can rewrite just one side
[14:40] <larsu> ya
[14:40] <Trevinho> (keeping u-s-d even closer to upstream)
[14:43] <larsu> ok, so nothing needed from me?
[14:44] <didrocks> larsu: don't you think we should investigate why the grab can happen twice at some startup?
[14:44] <didrocks> (and without an ungrab in between)
[14:44] <larsu> ah right you said it's not because it crashed
[14:44] <didrocks> yeah…
[14:44] <didrocks> that worries me
[14:44] <didrocks> maybe we just get lucky most of the time and the last grab is what unity sees
[14:44] <didrocks> but maybe sometimes…
[14:45] <didrocks> and so, we would double grab for everyone at every login?
[14:46] <larsu> sigh
[14:47] <didrocks> larsu: so, I just looks at /proc/pid/maps
[14:47] <didrocks> larsu: it doesn't load (phew) the g-s-d media keys plugin
[14:47] <didrocks> I wonder though why I have every plugin listed 4 times?
[14:47] <larsu> haha good idea
[14:47] <larsu> hm? in maps?
[14:48] <didrocks> yep
[14:48] <didrocks> even libwacom and such
[14:48] <didrocks> that might be "normal", I would have expected to see it only once though
[14:48] <larsu> that's all libs though, not only the plugins
[14:48] <larsu> at least for me
[14:48] <didrocks> yeah
[14:48] <didrocks> sale
[14:49] <didrocks> same*
[14:49] <didrocks> ok, so "normal"
[14:49] <didrocks> I guess it spawns multiple memory chunks
[14:49] <didrocks> so, it's not the g-s-d plugin interfering, so, in the code itself…
[14:49] <tjaalton> kgunn: pong, EOW already, but if you have a bug just assign it to me :)
[14:50] <kgunn> tjaalton: no worries, i'll just email you...oh, i'm assuming you're the person to talk to about mesa uploads ?
[14:51] <tjaalton> kgunn: yes, did 11.0.0 break something?
[14:52] <kgunn> tjaalton: yep :) so just wanted to establish some kinda of communication/agreement between us
[14:52] <tjaalton> first point release will arrive today, i'll push it next week
[14:52] <tjaalton> ah
[14:52] <kgunn> busted our mir ci jobs for wily
[14:53] <tjaalton> well the FFE was filed a month ago and raof tested it iirc
[14:53] <tjaalton> but i hear you
[16:02] <didrocks> have a good week-end guys!
[17:11] <Laney> bye!
[17:12] <willcooke> gnight Laney
[17:12] <willcooke> gnight all
[18:11] <Trevinho> Good weekend folks!
[21:49] <attente> bye!