[00:00] <ali1234> indicators are special stand-alone processes
[00:00] <ali1234> what apps do, is talk to those processes
[00:00] <ali1234> if the app wants to just make a generic icon, like a old-style notification icon, then it has to talk to indicator-application
[00:01] <ali1234> if the app is a music player, then it has to go through indicator-sound, which present information in a different way
[00:01] <ali1234> and same for messaging apps
[00:02] <ali1234> there's nothing to prevent an app from creating both a notification icon and an indicator-application icon
[00:02] <ali1234> in fact pidgin does this, if you tell it to
[00:03] <ali1234> so the way that indicators used to work is that the panel plugin loads the .so file, and the .so file goes and asks for the indicator service on dbus, then dbus will run the backend process (which is just a normal executable)
[00:03] <elfy> ok
[00:03] <elfy> tha makes sense to me 
[00:03] <ali1234> that is changed in trusty - now the panel/desktop environment is responsible for launching the backend. it won't happen automatically
[00:04] <ali1234> there are two problems with this
[00:04] <elfy> and this is the tedg things ?
[00:04] <ali1234> one is that stuff like lightdm-gtk-greeter doesn't even know it has to do this
[00:04] <ali1234> the other is related to dbus activation
[00:04] <ali1234> previously, dbus would run the backed when something asked for the service
[00:05] <ali1234> so as soon as the backend started up, something was instantly ready to use the provided dbus service
[00:05] <ali1234> then the backend would quit when there were no more watchers of the service
[00:06] <ali1234> under the new model there is no guarantee that there will be something waiting to use the dbus service when the backend starts, so the backends often just instaquit
[00:06] <elfy> ok - that's logical 
[00:06] <elfy> right
[00:06] <ali1234> yes, this is the tedg thing
[00:07] <elfy> ok - so I wonder with clementine, I tell it to use desktop notification - I assume then I would just get left with the one in ind-panle
[00:07] <ali1234> "desktop notification" means the notification area
[00:07] <elfy> apprently not
[00:08] <ali1234> it might also mean notifications as in the pop up bubbles
[00:08] <elfy> yea
[00:08] <ali1234> those have nothing at all to do with either the notification area or indicators, it's just a naming collision
[00:08] <elfy> ok - bit more with it now
[00:08] <elfy> thanks ali1234 
[00:08] <elfy> yep - got that collision
[00:09] <elfy> good job I've not been called the tech lead I guess LOL
[00:10] <elfy> my 'biggest' problem is I know what you are all talking in general terms, but it doesn't actually mean much specifically 
[00:11] <ali1234> the whole indicator thing is pretty confusing really - there are many parts to it, plus it uses dbus heavily
[00:11] <elfy> though in truth the biggest problem is most people just test it and carry on without telling us :)
[00:12] <ali1234> but the worst thing is since this is a race condition, it doesn't affect everyone
[00:12] <ali1234> it just depends how quickly your computer boots up
[00:12] <elfy> yep - I can understand that 
[00:12] <elfy> not 'so' fast here 
[00:12] <ali1234> there's about 20 different bugs about it on LP and nobody knows which are duplicates or what
[00:12] <elfy> just faster than the last one :)
[00:13] <ali1234> becuse there was a bug on saucy that had a similar effect (missing indicators) but it got fixed
[00:13] <elfy> ali1234: LOL - there are about 100k bugs like that :)
[00:13] <ali1234> this one is particularly bad because it seems like me and tedg are the only two people who actually understand it :)
[00:13] <elfy> :)
[00:14] <ali1234> so there's a lot of people complaining about this, but spread over many bugs
[00:14] <elfy> all I can say is I'm glad you're about :)
[00:14] <ali1234> anyway, i sent a patch to fix this, but tedg is still reworking the whole codebase, so it's all still up in the air
[00:14] <elfy> ali1234: yea - I see that, understand that from a user position- but most either come into here from a different desktop or from windows
[00:15] <ali1234> until he finishes with that we can't do much
[00:15] <elfy> and those last lot just assume a reinstall fixes it 
[00:15] <ali1234> haha yeah
[00:15] <elfy> yep
[00:15] <ali1234> that never works on linux, ever
[00:15] <elfy> we're in the position of really being bottom heavy here 
[00:15] <ali1234> bottom heavy?
[00:15] <elfy> ali1234: agreed with that ;)
[00:16] <elfy> yep - lots of people at the bottom pushing up - but not so many at the top that can take the lid of the kettle to let the pressure out
[00:16] <ali1234> well, if you ask me, the problem is that debian packaging is too damn hard
[00:17]  * elfy is a physical engineeer - work with wire and stuff - will wire you a jumbo jet
[00:17] <ali1234> i've tried to learn it countless times, i don't think i will ever understand it
[00:17] <elfy> ali1234: it's all dutch to me :)
[00:17] <ali1234> but, as a user, apt-get is the best
[00:18] <ali1234> it's a tricky one
[00:18] <elfy> I never created games with commodores and the like
[00:18] <elfy> had kids
[00:18] <elfy> apt-get is best 
[00:18] <elfy> but
[00:18] <elfy> I'd love it with the rpm delta thing
[00:18] <elfy> which I got used to between unity in 11.04(ish) and coming to xubuntu
[00:19] <elfy> that was just great thing
[00:19] <ali1234> from a "making packages" PoV, gentoo ebuilds are the easiest to create i think
[00:19] <elfy> I know there is an apt one - but it never really worked for me 
[00:19] <ali1234> but emerge just sucks for the users
[00:20]  * elfy has no idea - I want a system I can install, link music to, reboot - job done
[00:20] <drc> ah...linux from scratch.
[00:20] <elfy> though I have done it - just really, no thanks :)
[00:21]  * elfy prefers building planes from that ... 
[00:21] <elfy> ali1234: thanks for all that - sorted it out a bit for me :)
[00:23] <elfy> the current question really is - how to get people that have some interest involved in the sponsor queue
[00:23] <elfy> which 'could' be a managiing situation rather than a coding one - in which case I'd be interested
[00:24] <elfy> /no idea though if it is or isn't
[00:24] <ali1234> if i were the boss of xubuntu i'd just say "screw it, xubuntu is now a ppa, we ignore ubuntu release schedule"
[00:25] <ali1234> and just put out ISOs with whatever we feel like
[10:10] <Unit193> So, deluge indicator works fine in Ubuntu (so I'm shown and told), but looks like http://i.imgur.com/UqdLbnJ.png in Xubuntu (Predating saucy, saucy, and trusty+gtk3)...
[10:12] <Unit193> http://i.imgur.com/jOuuf2h.png is what it should look like.
[10:25] <knome> Unit193, heh.
[10:26] <Unit193> Finally went to deluge about it, and he shows me it working in Ubuntu.  Sooo, no idea.
[10:32] <brainwash> Unit193: looks like it falls back to notification icon
[10:32] <brainwash> did you apply the workaround?
[10:32] <Unit193> Workaround?  And no, that's an indicator, trayicon works fine.
[10:33] <brainwash> https://wiki.ubuntu.com/Xubuntu/Roadmap/Specifications/Trusty/Gtk3Indicators
[10:33] <brainwash> "Indicators may fail to load"
[10:35] <Unit193> No, haven't needed to as they've been working fine... notification icon == tray icon?
[10:35] <brainwash> yes
[10:36] <brainwash> indicators are special
[10:36] <Unit193> Then indicators are working fine, I know the difference. :P  (Also, it'd work much better if it were falling back, and the problem happens here on saucy.)
[10:36] <brainwash> I would like to test it myself... but something broke wrapper-2.0 (xfce4-indicator-plugin) over night
[10:37] <brainwash> maybe caused by the final move to gnome 3.10
[10:38] <Unit193> My indicator-application is the one from saucy though, in the trusty vm.
[10:38] <brainwash> or by one of my debug additions
[11:04] <Unit193> brainwash: But you'll be able to take a look later on perhaps?
[11:19] <brainwash> Unit193: yes, the menu looks different in unity and xubuntu
[14:01] <damiank> hi
[16:02] <ali1234> Unit193: the tray menu and indicator both load exactly the same glade file
[16:06] <brainwash> but the strings are different and some icons are hidden too in unity/gnome
[16:09] <ali1234> never the less, it is the same menu object, from the same glade file
[16:10] <brainwash> yes, you said that already :)
[16:11] <brainwash> did the strings got translated?
[16:20] <ali1234> no
[16:20] <ali1234> i think i see the problem
[16:20] <ali1234> however i don't understand why it works in unity
[16:21] <ali1234> this could be a bug in the panel plugin
[16:22] <brainwash> very likely, the normal try icon menu looks ok
[16:22] <brainwash> tray
[16:23] <ali1234> right, fixed it
[16:24] <brainwash> yet another bug which you were able to fix :)
[16:25] <ali1234> are you really sure this works in unity?
[16:25] <ali1234> i don't see how that is even possible
[16:27] <brainwash> it looked ok in unity and gnome
[16:38] <michel_> just tried the trusty daily build. why is iBus used for keyboard layouts instead of xkb?
[16:54] <slickymaster> afternoon all
[19:13] <ali1234> brainwash: Unit193: finally tested in unity. same bug, as i expected
[19:13] <ali1234> so this really is a bug in deluge
[19:16] <ali1234> http://imagebin.org/287402
[19:18] <ali1234> the reason this happens is because of the lines in the glade file like this:
[19:18] <ali1234> <property name="stock">gtk-media-pause</property>
[19:18] <ali1234> That is on the "Pause All" button
[19:18] <ali1234>  /menu item
[19:18] <ali1234> when that property is set, the label gets changed to "Pause"
[19:19] <ali1234> the fix is simply to remove all those "stock" properties
[19:20] <ali1234> i suppose it might be consideed a bug in indicator-application or somewhere else in that stack
[19:24] <Unit193> sudo sed '/name="stock"/d' tray_menu.glade -i fixed it, weird.
[19:26] <ali1234> i don't understand the purpose of those lines in the glade, nor do i understand how deluge is setting the custom icons on the upload/download limit menus
[19:42] <Unit193> Perhaps glade/torrent_menu.glade:215 and systemtray.py:146?
[19:49] <ali1234> no
[19:50] <ali1234> it's just 146
[19:51] <Unit193> Bah, copied the wrong one, meant the GtkImage id in tray_menu, but I don't know glade.
[19:52] <Unit193> But OK, anywho it's a deluge bug.
[20:03] <ali1234> still could be indicators
[20:10] <ali1234> here's a test case demonstrating the bug: http://paste.ubuntu.com/6776134/
[20:14] <ali1234> i'm beginning to think that it's a bug indicators really
[20:14] <ali1234> i mean it seems reasonable to expect the same glade to look the same in both places
[20:14] <Unit193> And the glade has been the same for "years"
[20:14] <ali1234> of course
[20:21] <ali1234> hmm there's a changelog item "Fix custom labels on stock items."
[20:21] <ali1234> bug 520048
[20:24] <ali1234> the file that patch applies to doesn't even exist any more
[20:24] <ali1234> i'd say this is a regression
[20:38] <ali1234> ok, reproduced it with no glade weirdness
[20:44] <ali1234> bug 1270486
[20:45] <Unit193> Wow, thanks very much!
[20:46] <brainwash> how does unity work around this bug?
[20:46] <ali1234> it doesn't
[20:46] <ali1234> deluge is broken there too
[20:46] <ali1234> don't even try to tell me it isn't
[20:46] <ali1234> i just spent two hours installing and upgrading VMs to test it
[20:46] <brainwash> the menu looked ok
[20:47] <brainwash> no play button and correct label "resume all"
[20:47] <brainwash> for example
[20:48] <holstein> i usually try as a different user. removing my config from the equation
[20:48] <brainwash> here is the screenshot linked before in this channel
[20:48] <brainwash> http://i.imgur.com/jOuuf2h.png
[20:48] <brainwash> it looks like this in unity/gnome
[20:48] <ali1234> maybe, it might be an old version
[20:48] <ali1234> it's broken in trusty and saucy
[20:48] <brainwash> trusty
[20:48] <holstein> i would also try something without forced 3d
[20:49] <ali1234> brainwash: try the testcase if you don't believe me...
[20:49] <brainwash> ok, but let me try it
[20:49] <holstein> sorry.. wrong channel.. though, it happens to read so relevantly ;)
[20:51] <Unit193> brainwash, ali1234: That was precise.
[20:51] <ali1234> the screenshot? yeah, this bug has been fixed before...
[21:01] <brainwash> so testcase indicators shows "hello"
[21:01] <brainwash> indicator
[21:02] <brainwash> ali1234: unity
[21:02] <ali1234> ಠ_ಠ
[21:03] <brainwash> so something wrong with xfce4-indicator-plugin?
[21:06] <ali1234> no
[21:06] <ali1234> when you see "hello" you don't see the pause icon right?
[21:06] <brainwash> right
[21:07] <ali1234> thus, you haven't really broken the testcase
[21:07] <ali1234> te idea is to see the pause icon and have the label read "hello" at the same time
[21:07] <ali1234> i'm not sure why the icon isn't there
[21:07] <brainwash> now I get it
[21:08] <brainwash> we want both.. icon and correct label :)
[21:11] <ali1234> yeah
[21:11] <ali1234> so you need gnome-icon-theme-full installed to have that icon installed
[21:11] <ali1234> if not installed, gtk will ignore it -> so appindicator bug won't be triggered
[21:11] <ali1234> this is useful tho, i'll do another test case i suppose, with something installed by default
[21:13] <brainwash> gnome-icon-theme-full is installed
[21:14] <brainwash> even selected
[21:14] <brainwash> but still no icon
[21:16] <Unit193> I get the icon, but not the text "hello", I do not have -full but have humanity and gnome-icon-theme.
[21:18] <brainwash> Unit193: can you build your xfdesktop package with this patch applied? https://bugzilla.xfce.org/show_bug.cgi?id=10605
[21:19] <brainwash> we need to test it, elfy is affected too
[21:20] <Unit193> It's by eric, does it really need any testing? ;)
[21:21] <Unit193> Sure, should apply.
[21:21] <brainwash> thanks
[21:23] <ali1234> i am looking for an icon where this works out the box...
[21:28] <ali1234> hmm
[21:28] <ali1234> so this really does bevahe differently on trusty
[21:28] <ali1234> but the glade testcase is still broken
[21:30] <ali1234> brainwash try this: https://bugs.launchpad.net/ubuntu/+source/indicator-application/+bug/1270486/+attachment/3951740/+files/testcase2.py
[21:32] <ali1234> it might be a gtk2/3 thing
[21:34] <brainwash> testcase 2: pause icon + "pause" label
[21:34] <ali1234> right, as expected
[21:51] <Unit193> Dang, uploaded to wrong ppa. >_<
[22:18] <Unit193> So how do you even see if it's gone?  It didn't happen every time, and restarting xfdesktop won't help as it "fixed" the problem before. :P
[22:20] <ali1234> eric said he knew what was causing it
[22:21] <Unit193> brainwash: update.
[22:30] <brainwash> Unit193: thank you :)
[22:30] <brainwash> it usually happens after you log in
[22:31] <Unit193> Yeah, had it on this desktop, restarted lightdm resterday (or, restarted dbus which crashed lightdm but left X hanging.)
[22:31] <brainwash> before updating I'll try to delay xfsettingsd to force this bug
[22:31] <brainwash> so I'll have a 100% testcase for the patch
[22:34] <Unit193> Wow, xmir actually works on the one desktop now, quite amazing.  (After a slight xorg crash and fall back to lightdm)
[22:41] <brainwash> it might work almost perfectly, but without a hardware cursor and dedicated support by the mir guys I feel like we are going nowhere
[22:42] <Unit193> (This should have gone to -offtopic, whoops) No, I mean it was real bad before, http://unit193.net/mir/DSCF3866.AVI
[22:43] <brainwash> xmir talk is always welcome :D
[22:43] <Unit193> (I don't like -devel, so tend to migrate to -offtopic a lot)
[22:49] <Unit193> Hah, xonotic took it down. :P
[22:55] <Unit193> brainwash: Patch didn't work for me.
[23:05] <brainwash> same here
[23:06] <Unit193> Can poke eric here, in #xfce-dev or bug. :P
[23:24] <brainwash> eric_the_idiot: the patch does not fix the xfdesktop issue (10605)
[23:25] <brainwash> xfsettingsd is like the first thing to start on session login
[23:26] <brainwash> or isn't it?
[23:30] <brainwash> xfce4-panel simply refreshes its appearance after xfsettingsd is finally up and running
[23:30] <brainwash> xfdesktop does not
[23:31] <brainwash> added a 5sec delay
[23:33] <ali1234> yes, normally things do refresh when the settings daemon starts/stops, which is odd
[23:33] <ali1234> maybe it is something different about the way it is initiallly launched
[23:35] <brainwash> now I got it, so xfdesktop does actually wait for xfwm4 and now additionally for xfsettingsd (patch)
[23:35] <damiank> hi brainwash
[23:35] <brainwash> but there is a 5sec timeout
[23:36] <brainwash> then xfdesktop will start
[23:36] <brainwash> despite the status of xfwm4/xfsettingd
[23:37] <brainwash> hey damiank 
[23:37] <ali1234> yeah, and it always times out
[23:37] <damiank> i have a lot of spare time this weekend, so i'm downloading today daily to test
[23:38] <brainwash> ali1234: so we got two issues here.. the 5sec timeout and the missing theme refresh
[23:39] <ali1234> i think what might be happening is that xfdesktop is waiting for xfwm, but xfwm is waiting for xfdesktop
[23:39] <brainwash> woot
[23:39] <brainwash> :D
[23:40] <ali1234> this would be a good candidate for some syslog debugging, to log all the times that everything gets started
[23:40] <ali1234> i just so happen to have a shim that will do this without patching everything
[23:40] <ali1234> http://paste.ubuntu.com/6777124/
[23:41] <ali1234> rename xfdesktop to xfdesktop.real
[23:41] <ali1234> copy the built shim to where xfdesktop was
[23:41] <brainwash> ah nice
[23:41] <ali1234> do the same for xfwm and xfsettingsd
[23:41] <ali1234> it automatically knows what to run
[23:41] <brainwash> ok
[23:41] <ali1234> and it will log everything
[23:43] <brainwash> I'll test it tomorrow
[23:43] <brainwash> here's the bug report + upstream report in case you want to add some findings
[23:44] <brainwash> bug 1270261
[23:44] <brainwash> good night