[00:32] <TheMuso> jbicha: yes thats the plan, thanks, I need to file an MIR.
[00:43] <jbicha> TheMuso: it should be an easy mir since it's just switching one for the other
[00:52] <TheMuso> jbicha: I know, I have been working in Debian to get things set up such that this would be easy.
[02:47] <duflu> robert_ancell: Is is correct to assume graphical login bugs are unity-greeter and never lightdm?
[02:47] <robert_ancell> duflu, broadly speaking, yes
[02:47] <duflu> OK, ta
[02:47] <robert_ancell> duflu, since lightdm doesn't do anything except launch the X process, i.e. it doesn't attempt to configure anything X related
[02:48] <duflu> robert_ancell: So is that also you, or a designer person?
[02:48] <duflu> Cimi?
[02:48] <robert_ancell> duflu, u-g is fairly unmaintained, but most officially it's owned by ~ubuntu-desktop
[02:49] <duflu> Hurray. More projects than maintainers agao
[02:49] <duflu> again
[02:51] <robert_ancell> duflu, yeah :( When U8 greeter comes along it will fade away though
[05:57] <hikiko>  hi
[05:57] <TheMuso> Hey hikiko.
[05:58] <hikiko> hi TheMuso
[07:46] <happyaron> morning EU guys
[07:46] <flocculant> I hope that includes UK people :)
[07:47] <duflu> happyaron: Asia says good afternoon
[07:47] <happyaron> ah yea :)
[07:47] <happyaron> flocculant: sure, lol
[07:48] <flocculant> \o/
[07:48] <flocculant> :)
[08:50] <willcooke> good morning all
[08:57] <Sweet5hark> moin
[09:00] <seb128> good morning desktopers
[09:00] <seb128> hey Sweet5hark willcooke
[09:00] <seb128> had a good w.e?
[09:00] <flexiondotorg> Morning seb128 willcooke happyaron
[09:00] <seb128> hey flexiondotorg
[09:00] <flexiondotorg> Yes good weekend here seb128. Took my daughter to the Winchester Science Centre :-)
[09:01] <flexiondotorg> Which is absolutely ace.
[09:01] <Laney> morning
[09:02] <didrocks> hey flexiondotorg, Laney! re seb128 :)
[09:02] <flexiondotorg> Morning didrocks Laney
[09:02] <flexiondotorg> didrocks Sleeping much? :-)
[09:02] <didrocks> flexiondotorg: my side, yeah as I'm back home at night, my wife, at the clinic less :)
[09:03] <didrocks> (working from the clinic today)
[09:03] <flexiondotorg> didrocks When will everyone be home?
[09:03] <seb128> hey Laney
[09:03] <seb128> flexiondotorg, that sounds nice :-)
[09:03] <didrocks> flexiondotorg: unsure yet, maybe tomorrow, depending on how much weight/fast he will take
[09:04] <willcooke> didrocks, working?!
[09:04] <didrocks> willcooke: yep, that way, I keep my official 3 days off when we'll all be at home
[09:04] <Laney> hey didrocks flexiondotorg seb128 willcooke
[09:04] <seb128> Laney, had a good w.e?
[09:04] <didrocks> which will be the most difficult part it seems :)
[09:04] <willcooke> 3 days, is that all you get?
[09:04] <didrocks> willcooke: and 4G, working from clinic, FTW!
[09:04] <didrocks> yeah
[09:04] <willcooke> wow
[09:04] <willcooke> that sucks
[09:04] <didrocks> I have saved 5 days from my 2015 holidays
[09:05] <didrocks> so, we'll add that up
[09:05] <didrocks> but yeah, not that good for father in France
[09:05] <seb128> you can take 15 days out of work as well no?
[09:05] <didrocks> sure, unpaid though
[09:05] <seb128> which are paid by the state and less that your normal pay though...
[09:05] <didrocks> or way less
[09:05] <seb128> right
[09:06] <didrocks> I don't remember, we looked at it, as working from home, I guess that's going to be manageable
[09:06] <didrocks> between 2 deep sleep :)
[09:06] <seb128> :-)
[09:07] <Laney> seb128: not bad - family visit (cousin), hanging out @ library, pub with friends, baking bread and making soup!
[09:07] <Laney> that's even the right order
[09:07] <Laney> no climbing because I did something weird to my finger nail on thursday
[09:07] <Laney> think it's better now though
[09:07] <Laney> you?
[09:10] <davmor2> Morning all
[09:11] <seb128> w.e was mostly non eventfull, I was supposed to have a tennis afternoon yesterday but the courts were frozen so we couldn't play, otherwise did some shopping and did some house cleaning (got the holidays tree out) and we had dinner with friends yesterday
[09:12] <davmor2> seb128: but you could of put skates on and combine the tennis and skating what's wrong with you ;)
[09:14] <Laney> probably likes his legs unbroken :P
[09:14] <Sweet5hark> my weekend was fine, but mostly boring. Some window shopping on Saturday, some biking on Sunday (twas cold but dry and sunny). I dont know much about american football but watched Green Bay vs. Dallas Cowboys yesterday -- dammit, if those games are always such nailbiters, I had missed a lot so far ... ;)
[09:16] <flexiondotorg> Morning Sweet5hark davmor2
[10:18] <flexiondotorg> Laney I have a question about update excuses
[10:18] <flexiondotorg> I uploaded MATE 1.17 dev snapshots over the weekend after first testing in a PPA>
[10:19] <flexiondotorg> caja and eom are in update excuses
[10:19] <flexiondotorg> Depends: caja exempi (not considered)
[10:19] <flexiondotorg> exempi is not a version B-D.
[10:19] <flexiondotorg> So I don't understand why caja and eom are not be promoted out of proposed.
[10:20] <flexiondotorg> exempi has update excuses but the version in zesty universe is suitable.
[10:28] <Laney> flexiondotorg: https://launchpadlibrarian.net/302363945/buildlog_ubuntu-zesty-amd64.caja_1.17.2-0ubuntu1_BUILDING.txt.gz
[10:28] <Laney> Depends: [...] libexempi3 (>= 2.4.0) [...]
[10:31] <flexiondotorg> debian/control:               libexempi-dev,
[10:31] <flexiondotorg> That is the same in caja and eom.
[10:31] <Laney> Doesn't matter
[10:31] <Laney> You got a versioned runtime dependency
[10:31] <Laney> That's what matters
[10:31] <flexiondotorg> Is that implied?
[10:32] <flexiondotorg> Because exempi is not versioned in the debian/control.
[10:32] <Laney> It's calculated using shlibs / symbols
[10:32] <Laney> laney@nightingale> cat libexempi3.shlibs                                                                                                                                   ~/temp/exempi-2.4.0/debian
[10:32] <Laney> libexempi 3 libexempi3 (>= 2.4.0)
[10:33] <flexiondotorg> But is exempi 2.4.0 has not been promoted yet, how to caja and eom build against it?
[10:33] <flexiondotorg> *how did
[10:33] <Laney> Builds always happen against all of proposed
[10:33] <flexiondotorg> Aha.
[10:33] <flexiondotorg> OK, thanks.
[10:33] <flexiondotorg> TIL.
[10:34] <Laney> Looks like maybe those tests really do want fixing too
[10:34] <Laney> i.e. real failures
[10:46] <hikiko> hey
[10:46] <hikiko> Trevinho,
[10:47] <hikiko> http://pastebin.com/61jdjiTw
[10:47] <Trevinho> hey hikiko
[10:47] <hikiko> apparently I was asking in the wrong channel :/
[10:47] <hikiko> pastebin with questions for you
[10:47] <Trevinho> np, let me check
[10:47] <hikiko> btw almost fixed the U7 part
[10:48] <hikiko> but I think the design for ucc wont switch to lowgfx instantly
[10:49] <Trevinho> hikiko: well, in the video I see, there's no actual instant change... Only unity settings are updated instantly (and that would happen anyway) like the dash blurring. But for example shadows and other compiz settings doesn't look like they're updated instantly with the change you suggested, but would rather will be used at next reboot anyway
[10:49] <hikiko> no, they are
[10:49] <hikiko> checked that
[10:50] <hikiko> every time you modify the Default.ini
[10:50] <hikiko> u7 is reloaded with the new settings
[10:50] <hikiko> that's why I copy the settings from lowgfx.ini
[10:50] <hikiko> you can run it and see yourself
[10:50] <hikiko> but you have a good point about the tool
[10:50] <hikiko> so I am going to do that as well
[10:51] <hikiko> but I think we should keep the copy
[10:51] <Trevinho> hikiko: wasn't you successifully by using ccsSetProfile() instead?
[10:52] <hikiko> ccsSetProfile => dependency on compizconfig
[10:52] <Trevinho> hikiko: mh, I'm fine to do that if for you it's the only way to update these settings... .But the in the startup script, delete that default.ini if any...
[10:52] <Trevinho> hikiko: we already have it and it's fine to add it.
[10:52] <hikiko> if you delete it
[10:52] <hikiko> compiz crashes :)
[10:52] <Trevinho> hikiko: before runing unity.
[10:52] <Trevinho> hikiko: in the startup script..
[10:52] <hikiko> the python script?
[10:54] <Trevinho> like... you do as you do right now (which, I personally don't love, as I'd prefer to avoid such duplication of .ini files), but... If it's the fastest way to get the settings changed is fine (although, that won't change profile... Then if an user changes settings in ccsm in that moment, they won't be applied to the "lowgfx" profile but to the "unity" one.
[10:54] <Trevinho> But assuming, you're doing that... At startup in the compiz-profile-checker script, you firstly check what proflie to use, set the env variable and then remove the default.ini..
[10:54] <Trevinho> then compiz is started (with the proper env)
[10:55] <hikiko> right
[10:55] <hikiko> I'll check this out
[10:55] <Trevinho> Like.... I think this scenario won't work with the current setup:
[10:56] <Trevinho> 1) Enable launcher autohide
[10:56] <Trevinho> 2) switch to lowgfx
[10:56] <Trevinho> 3) reload the user session
[10:56] <Trevinho> 4) the launcher won't be in auto hide...
[10:56] <Trevinho> 1) and 2) can be in different order
[10:56] <hikiko> I could backup the default.ini every time
[10:57] <hikiko> and then when you switch back overwrite it with the backup
[10:57] <hikiko> like:
[10:57] <hikiko> 1) Enable autohide
[10:57] <hikiko> switch to lowgfx
[10:57] <Trevinho> no, no... get rid of it. It can be fine as a temporary file to be there, but it has not to be the reference for our settings
[10:57] <hikiko> so if I delete it
[10:57] <Trevinho> we've gsettings for that, and we can't rely on that as a primary source
[10:57] <hikiko> 1-4 will work?
[10:58] <hikiko> I didn't understand that step :|
[10:58] <Trevinho> ohhhhh... maybe my scenario is wrong... since I'm covering both the profiles in unity-control-settings, but... if you use ccsm for something else, then it will be that.
[10:59] <hikiko> yep
[10:59] <hikiko> I asked willcooke about that
[10:59] <hikiko> I had 3 suggestions:
[10:59] <Trevinho> hikiko: what I mean is that if you then run unity using that default.ini as source, then all the profiles settings won't work as expected, since unity will still be in the unity profile, while you try to change the lowgfx... or the other way around
[10:59] <hikiko> I kno
[10:59] <hikiko> you ll have the defaults always
[11:02] <hikiko> we could either 1- backup the settings, 2- ignore ccsm and suppose that the user that uses ccsm will know how to export a profile and load it again if he wants to experiment with lowgfx, 3- warn the user he might lose custom settings 4- add a 3rd option Full Effects () Limited Effects () Previously Used Effects () (with better wording) and load a backup if the 3rd option is selected
[11:02] <hikiko> willcooke, said 2 is fine
[11:02] <hikiko> 3 is also very easy to implement
[11:04] <Trevinho> No, I'm fine with ignoring ccsm, I just don't want to use a default.ini file to be there... If adding that temporary and then cleaning it up works... I can accept it, but since we've a CCSContextSetProfile that does exactly what we need, I don't see a reason why not to use it...
[11:04] <Trevinho> hikiko: if you don't want to add a new dependency to ucc, I don't see the reason, but in case unity depends on it thus you can just add a tool to unity package and call it from ucc...
[11:05] <hikiko> because when we last discussed that we agreed that it's not a good idea to add a compizconfig dependency
[11:05] <hikiko> I am fine to add a dependency
[11:05] <Trevinho> We probably said that in previous cycle since we were in freezes
[11:05] <hikiko> but you had your obkections :)
[11:05] <Trevinho> but we've still some days before FF to do it
[11:06] <Trevinho> So, go for that instead of messing with files... It's just the solution to use. More maintalable and that will work properly
[11:06] <hikiko> alright I can do this today, but we have to review that until wednesday because I have to start working on mir-something afterwards
[11:06] <hikiko> is this fine by you?
[11:07] <Trevinho> As you prefer, doing it should be matter of few hours.... MP it as soon as you can
[11:07] <hikiko> yeah
[11:07] <hikiko> so, I'll ping you later :)
[11:10] <Trevinho> hikiko: in case, if the default.ini file doesn't exist when compiz starts, and you add it... Does compiz will read from that on creation?
[11:13] <hikiko> if Default.ini doesn't exist compiz creates it
[11:14] <hikiko> and puts there the default plugins
[11:14] <hikiko> with the default settings
[11:14] <hikiko> in our case it has the unity.ini contents
[11:14] <hikiko> Trevinho, ^
[11:15] <Trevinho> hikiko: no, I mean... Without our tools. By default there's never a default.ini
[11:15] <Trevinho> without the *new tools*
[11:16] <hikiko> mmm Trevinho I don't know
[11:16] <hikiko> does xenial have the new tools?
[11:16] <Trevinho> "by new tools" I mean the ones you're adding...
[11:16] <hikiko> lol
[11:17] <hikiko> before I add the tools I checked this manually
[11:18] <Trevinho> mh, i see
[11:18] <hikiko> compiz crashed if I was deleting Default.ini and when I forced a restart it used to recreate it. Then if I overwrite it
[11:18] <hikiko> it instantly loads the new settings
[11:18] <Trevinho> ok
[11:24] <Trevinho> hikiko: otherwise I've just checked that modifying by hand .config/compiz-1/compizconfig/config is enough to change the profile... If you don't want to use the ccs function...
[11:24] <Trevinho> it's just about adding/removing a
[11:24] <Trevinho> [general_ubuntu]
[11:24] <Trevinho> profile = unity-lowgfx
[11:24] <Trevinho> part
[11:25] <hikiko> that sounds better Trevinho :)
[11:25] <hikiko> thanks
[11:25] <hikiko> I am going to try this
[11:26] <Trevinho> hikiko: you should use https://developer.gnome.org/glib/stable/glib-Key-value-file-parser.html to edit it...
[11:26] <hikiko> I am finishing the u7 changes now, I ll ping you in a few hrs when I ll have both ready for review, thanks
[11:26] <hikiko> !
[11:27] <hikiko> so, there's a parser already
[11:27] <hikiko> cooooooool :)
[11:44] <willcooke> sorry, was in a meeting - hikiko|ln, Trevinho  do I need to do anything?
[11:44] <Trevinho> willcooke: no, no... We were just discussing how to implement that lowgfx switch
[11:45] <willcooke> I'll read scrollback in a bit
[11:46] <Trevinho> sure
[12:48] <andyrock> hey guys
[12:55] <willcooke> hey andyrock
[13:10] <ksamak> hikiko: hi, i have a trick question. that might also need Trevinho's opinion
[13:11] <hikiko> ksamak, go on
[13:11] <ksamak> as you might remember, we've made a focus tracking for ezoom module in compiz, that uses AT-SPI for backend.
[13:12] <ksamak> we've just finished a version, and i'd like to push whatever is possible in compiz upstream (all would be great)
[13:12] <ksamak> previously, ezoom used mousepoll to poll the mouse and know it's position changes
[13:12] <ksamak> i've added the logic to mousepoll, so it manages caret (keyboard) focus as well
[13:13] <ksamak> i've tried to make another copycat of mousepoll, calling it focuspoll, but it freeze the whole thing, so i did it whithin mousepoll.
[13:14] <ksamak> i'll push the modif soon, and you can give me your opinions on that
[13:14] <ksamak> secondly
[13:15] <ksamak> there's a slight modification that has been made to the zoom centering algorythm, to include the possibility of legacy mouse behaviour, or an "always centered mouse" as requested by visually disabled people
[13:16] <ksamak> Trevinho: so this option breaks the algorythm for the showmouse plugin, introducing the need to communicate between ezoom and showmouse. would that be acceptable?
[13:17] <hikiko> lol
[13:17] <ksamak> btw, you can test our repo at debian.hypra.fr/debian
[13:17] <hikiko> ksamak, the process goes like that:
[13:17] <hikiko> you create an lp account
[13:18] <hikiko> you upload your branch
[13:18] <ksamak> i've already pushed a couple minor things into compiz
[13:18] <hikiko> and you propose for merge
[13:18] <hikiko> then we ll review it
[13:18] <ksamak> but it's just about asking your opinion
[13:18] <hikiko> if it breaks stuff
[13:18] <hikiko> you ll have to fix it
[13:18] <hikiko> or
[13:18] <hikiko> add some code that doesnt allow you to enable ezoom + showmouse together
[13:18] <hikiko> that is not desirable i think
[13:19] <hikiko> willcooke, is the desktop team manager btw
[13:19] <ksamak> some people actually need both
[13:19] <hikiko> so Trevinho and I can only advise you for the code
[13:19] <hikiko> when it comes to the design
[13:19] <hikiko> you better ask willcooke
[13:19] <ksamak> ok
[13:33] <andyrock> ksamak: do you have some code already? for focuspoll
[13:34] <andyrock> we can take a look and see if it crash
[13:34] <andyrock> otherwise post the diff of mousepoll and I can tell you if it's ok or not
[13:35] <ksamak> andyrock: i have an old branch that uses a focuspoll plugin
[13:36] <ksamak> andyrock: i'll do that push, first i'll try to push an option for switcher.
[13:37] <ksamak> andyrock: if you want to have a look already => https://git.hypra.fr/hypra/compiz
[13:38] <andyrock> where should I look? :D
[13:38] <andyrock> ah found it
[13:39] <andyrock> btw better give me a diff
[13:43] <ksamak> andyrock: yeah, i'm working on that
[13:43] <ksamak> andyrock: it's ok, i'll make a push, then i guess you'll see all i did wrong XD
[13:46] <andyrock> ok i see you don't touch mousePos
[13:46] <andyrock> should be ok
[13:46] <andyrock> ksamak: ^^^
[13:46] <andyrock> so for me (and I guess for Trevinho as well) it should be ok to do that inside the mousePoll plugin
[13:48] <andyrock> maybe just add a11ywatcher->check_and_process_queue(); inside the if
[13:49] <andyrock> so we don't wase cpu cycles
[13:58] <ksamak> andyrock: ok, kewl
[14:11] <ksamak> andyrock: u know if there's any way of checking another module's gsettings, when the other's not loaded in memory?
[14:12] <ksamak> eg: showmouse checking some gsetting from ezoom, when ezoom's not active?
[14:12] <andyrock> just use gsettings directly but it's a bit tricky
[14:12] <andyrock> mmm not a good design btw
[14:13] <ksamak> yeah i know, but it would prevent putting a dependancy to ezoom in showmouse
[14:13] <ksamak> although i wonder if it's a problem at all
[14:13] <ksamak> since these modules are kinda linked to a11y
[14:14] <andyrock> if showmouse requires ezoom is a good idea to do that
[14:14] <andyrock> what's the problem you're trying to solve
[14:16] <ksamak> the mouse centering algorythms of ezoom and showmouse, which have to be in sync
[14:16] <ksamak> and since we introduce a new one in ezoom..
[15:04] <andyrock> mmm not sure tbh
[15:11] <hikiko> Trevinho,
[15:11] <hikiko> editing .config/compiz-1/compizconfig/config just crashes compiz in my case
[15:12] <hikiko> it doesn't make any real-time changes
[15:12] <seb128> seems like a bug worth debugging/fixing?
[15:12] <hikiko> the only thing that works ok so far is overwriting the default.ini
[15:15] <hikiko> mmm this time it didn't crash but it didn't make any change at all
[15:16] <hikiko> Trevinho, it only works after reboot
[15:17] <hikiko> restart*
[15:17] <hikiko> the only thing that works for me is overwriting the Default.ini :/
[15:18] <seb128> hikiko, did you look if your segfault issue is reported?
[15:19] <hikiko> which segfault?
[15:20] <hikiko> seb128, did you run u-c-c without the unity branch?
[15:20] <seb128> hikiko, <hikiko>  editing .config/compiz-1/compizconfig/config just crashes compiz in my case
[15:21] <seb128> compiz shouldn't crash because you edit a config
[15:21] <seb128> seems worth debugging
[15:21] <hikiko> seb128, it just crashed once, now it's completely ignored unless I restart unity
[15:22] <hikiko> seb128, maybe the first crash was just a result of a meshed up desktop, in a clean system editing the config file is just ignored
[15:23] <hikiko> messed*
[15:23] <seb128> k
[15:23] <seb128> and you didn't try to get a backtrace or an apport report when you had it?
[15:26] <hikiko> no :/
[15:27] <hikiko> it wasn't running on gdb to get a bt
[15:33] <hikiko> seb128, here we are:
[15:33] <hikiko> http://pastebin.com/Um9DxYy0
[15:33] <hikiko> I got another crash
[15:33] <hikiko> but it seems totally irrelevant to lowgfx
[15:33] <hikiko> I received it while I was recording my desktop
[15:34] <hikiko> something at session GnomeManager
[15:34] <hikiko> :S
[15:34] <seb128> seems you don't have the glib debug symbols and maybe not the libunity-core ones either?
[15:38] <hikiko> (gdb) f 5
[15:38] <hikiko> #5  unity::session::GnomeManager::Impl::SetupLogin1Proxy (this=0x56222b0e8a90,
[15:38] <hikiko>     session_path="/org/freedesktop/login1/session/c87")
[15:38] <hikiko>     at ./UnityCore/GnomeSessionManager.cpp:235
[15:38] <hikiko> 235     ./UnityCore/GnomeSessionManager.cpp: No such file or directory.
[15:38] <hikiko> wtf
[15:38] <hikiko> seb128, I installed them, repeated the process
[15:38] <hikiko> and got this
[15:38] <hikiko> but I can't make sense
[15:38] <seb128> Trevinho or andyrock should be able to help you there I guess
[15:39] <hikiko> yeah
[15:39] <hikiko> or someone who works on gnome?
[15:39] <hikiko> or it's not related?
[15:39] <hikiko> Trevinho, ping :p
[15:39] <Trevinho> let me check
[15:39] <Trevinho> I was about to push something on Qt's gerrit and that made me mad :-D
[15:40] <hikiko> Trevinho, also check what I wrote about editing the config
[15:40] <hikiko> it doesn't switch to lowgfx instantly
[15:40] <hikiko> only after restart
[15:40] <hikiko> :/
[15:40] <Trevinho> hikiko: well, it does if the file's here in a manual check
[15:40] <hikiko> only way to do it so far is either like I did it or by adding a dependency to compizconfig
[15:41] <Trevinho> hikiko: but maybe if the file wasn't there it won't wor
[15:41] <Trevinho> hikiko: but maybe if the file wasn't there it won't work*
[15:41] <hikiko> I had the file
[15:41] <hikiko> :s
[15:41] <Trevinho> ~/.config/compiz-1/compizconfig/config ?
[15:41] <hikiko> yes
[15:41] <Trevinho> let me try in a guest
[15:41] <hikiko> and I edited it to be profile = unity-lowgfx
[15:42] <hikiko> it works only after reboot
[15:42] <hikiko> restart
[15:42] <hikiko> 10th time I say reboot instead of restart in a day :p
[15:43] <hikiko> Trevinho, so far only overwriting it works ok for me
[15:43] <hikiko> s:it/default
[15:53] <Trevinho> So...... hikiko about the crashes, if switching profile causes something like that, I really think we should fix it properly... So gdb is your friend in that case. Crashes are probably due to the fact that all the plugins are unloaded/reloaded. And probably one of them (very likely unity) has something that needs to be fixed.... Now... We without stack
[15:53] <Trevinho> traces we can't fix them. So make sure you're using also a compiled verison of unity
[15:53] <Trevinho> install debug symbols
[15:57] <Trevinho> as for the problem itself, just try to use ccsSetProfile then...
[15:58] <Trevinho> I also had some crashes when try to change the profile with ccsm, so first debug those issues frist
[16:06] <hikiko>  Trevinho seb128 andyrock http://pastebin.com/qfzyX5Nj
[16:06] <hikiko> it's clearly non-related to lowgfx
[16:06] <hikiko> and to plugins loading
[16:07] <hikiko> the bug is somewhere in gnome session manager
[16:07] <hikiko> but that's not a code I am familiar to...
[16:07] <hikiko> do you understand what might be wrong here? or do you know who knows that part better?
[16:08] <Trevinho> hikiko gnome session manager is part of unity...
[16:08] <andyrock> hikiko: likely something like that
[16:09] <andyrock> https://www.irccloud.com/pastebin/UvUHrnVT/
[16:09] <andyrock> should fix the issue
[16:10] <andyrock> not shure why that part of code is being called at unity unload
[16:10] <hikiko> andyrock, sorry :p I dont understand the issue
[16:11] <andyrock> in GnomeSessionManager
[16:11] <hikiko> what does this code do?
[16:11] <andyrock> https://www.irccloud.com/pastebin/x6qhKoYX/
[16:11] <hikiko> basically, what's the gnome session manager?
[16:11] <andyrock> just trying adding the if
[16:11] <Trevinho> hikiko: it's the unity module to manage session using gnome..
[16:11] <andyrock> it's a "wrapper" between unity and the session manager
[16:11] <hikiko> oh
[16:12] <hikiko> so the paste was the fix
[16:12] <Trevinho> might be
[16:12] <hikiko> so andyrock the problem is that the login_proxy doesnt exist
[16:12] <Trevinho> anyyway, it seems that the crash happens when a gdbus return value is sent...
[16:12] <hikiko> and we are querying its properties?
[16:12] <Trevinho> it could be that it gets deleted in different order....
[16:12] <hikiko> yup
[16:13] <hikiko> let's see if that fixes it
[16:14] <Trevinho> I'm not entirely sure, but it could be
[16:14] <hikiko> yeah
[16:14] <hikiko> looks probable
[16:15] <Trevinho> mh, not sure, since the return actually happens...
[16:16] <Trevinho> it could be like a race... the lambda we passed as callback, has been actually destroyed...
[16:16] <Trevinho> so when we get the reply
[16:16] <Trevinho> the call in glibdbusproxy ...(*callback)(value);
[16:16] <Trevinho> fails
[16:18] <Trevinho> hikiko: please put some void DBusProxy::GetProperty to understand what property is fetching... and thus to see what's failing to get back
[16:18] <andyrock> from the bt the problem happens inside SetGetterFunction not inside the callback
[16:20] <hikiko> I still get segfault
[16:21] <hikiko> give me a few minutes to see if I can find something more relevamt
[16:21] <Trevinho> andyrock: the problem is in the call at proxy->GetProperty("Display", [this, proxy] (GVariant *variant) {
[16:21] <andyrock> yep
[16:22] <Trevinho> so, the problem is that probably when we get the reply, the session manager has already been destroyed
[16:23] <andyrock> but would be nice to add a GCancellable flag
[16:23] <andyrock> to GerProperty
[16:23] <andyrock> and block the call in the dtor
[16:23] <Trevinho> andyrock: that's the thing I was going to look at
[16:24] <Trevinho> andyrock: but we already have a gcancellable there, and that should be invalidated since also the proxy should  be destroyed at that levle
[16:24] <Trevinho> level*
[16:25] <andyrock> that's true too
[16:26] <Trevinho> in fact g_cancellable_is_cancelled(canc) probably returns true, and it that case we can just return
[16:26] <Trevinho> but..... in that case error should be set as cancelled, that's why I'm not understanding why it's failing this way
[16:29] <Trevinho> in fact... as per docs `If cancellable is canceled, the operation will fail with G_IO_ERROR_CANCELLED.`
[16:30] <Trevinho> buttttttttttttttttt..... Ok, that's clear
[16:31] <Trevinho> proxy isn't deleted since it waits for the reply (being passed as a copy to the lambda), thus there's no cancellation
[16:31] <Trevinho> andyrock: ^
[16:31] <andyrock> yep
[16:31] <andyrock> so proxy should be a member of the class
[16:32] <Trevinho> or... we pass a cancellable
[16:32] <andyrock> and we need to change a lot of stuff inside DbusProxy
[16:32] <andyrock> not a lot but still
[16:34] <andyrock> ok it's few lines of code
[16:34] <andyrock> I can prepare a branch
[16:34] <andyrock> Trevinho: hikiko ^^^
[16:35] <Trevinho> andyrock: I think we could just instead use sigc::track_obj
[16:35] <Trevinho> since it's a trackable, it will do all for free
[16:36] <andyrock> hikiko: can you try it?
[16:36] <andyrock> https://www.irccloud.com/pastebin/MKOaAu8N/
[16:36] <andyrock> here just wrap the lambda in sigc::track_obj
[16:36] <andyrock> something like
[16:37] <andyrock> https://www.irccloud.com/pastebin/HZi54O8t/
[16:37] <seb128> k, I'm going for a tennis match now, will read backlog after sport&dinner
[16:38] <Trevinho> mhmhmh, it doens't compile here
[16:38] <Trevinho> seb128: enjoy!
[16:38] <seb128> Trevinho, thanks!
[16:40] <Trevinho> andyrock: ah.... we both forgot  the *this thing :-=)
[16:41] <Trevinho> hikiko: http://paste.ubuntu.com/23811089/
[16:42] <hikiko> Trevinho, sorry I had swich DE
[16:42] <hikiko> I am trying
[16:42] <hikiko> thank you
[16:55] <hikiko> Trevinho, andyrock unfortunately it's not that but I have some ideas, I'll check it
[16:55] <hikiko> thank you :)
[16:55] <Trevinho> hikiko: still same backtrace?
[16:55] <hikiko> it must be something related
[16:55] <andyrock> hikiko: still the same crash?
[16:55] <hikiko> yep
[16:56] <Trevinho> hikiko: so..... try with this other possibility
[16:56] <hikiko> I am checking other possible invalid memory access
[16:56]  * Trevinho codes...
[16:56] <Trevinho> naaaa, I think it's probably not that far
[17:01] <Trevinho> hikiko: http://paste.ubuntu.com/23811174/
[17:01] <Trevinho> try that
[17:03] <hikiko> Trevinho, give me a minute
[17:07] <Trevinho> ksamak: sorry I read things later..
[17:07] <Trevinho> ksamak: instead of mouse polling, also using xinput2 might help... as it allows to monitor the position in any case... Also when the ptr is grabbed
[17:07] <Trevinho> ksamak: something like I did in https://code.launchpad.net/~3v1n0/unity/xi2-input-monitor/+merge/303122
[17:08] <Trevinho> there is some overcomplication there to keep compatibility with XInput, but the simplest method is quite easy... Check xinput (tool) code in case
[17:26] <ksamak> Trevinho: ok. but the question was more about general architecture than for the actual mouse polling. i pretty much didn't change the method that was used for that.
[17:42] <Trevinho> ksamak: in general what I read is fine... I'm looking at the implementation
[17:44] <hikiko> Trevinho, I was almost sure your fix was the one this time
[17:45] <hikiko> but after several repeated switches
[17:45] <hikiko> it crashed again :/
[17:45] <hikiko> oh no wait
[17:45] <hikiko> that's a different crash
[17:46] <Trevinho> hikiko: it has to be different :-P
[17:46] <hikiko> Trevinho, I think it works :)
[17:46] <hikiko> well,
[17:46] <hikiko> now I have to fix the other
[17:46] <hikiko> GRRR :)
[17:46] <hikiko> thank you!
[17:46] <hikiko> Trevinho,
[17:46] <hikiko> will you submit the fix?
[17:47] <hikiko> or you want me to do add it in my branch and give you credits
[17:47] <hikiko> to save time?
[17:47] <hikiko> :s/do//
[17:49] <hikiko> btw if the final decision is that we have to add the compizconfig dependency and try to change the profile on the fly
[17:49] <hikiko> maybe I shouldn't spend time to fix phantom bugs that occur when you switch from one mode to the other 50 times
[17:50] <hikiko> but are we sure this is going to work?
[17:50] <hikiko> Trevinho, you think I should go for the compizconfig solution? currently the only thing that works ok on the fly is what I've already done with the copy
[18:03] <Laney> night!
[18:07] <hikiko> Trevinho, ?
[18:16] <Trevinho> hikiko: I'm going to propose the script...
[18:17] <Trevinho> hikiko: however when you encounter bugs... It's better if you try to fix them or nobody else will do it. A part from us, but... There's also something else to do. So...
[18:18] <Trevinho> hikiko: changing profile with compizconfig lib should work on the fly
[18:18] <Trevinho> using default.ini to override stuff, is something I wouldn't do for various reasons as I said
[18:22] <hikiko> yes Trevinho I agree but the deadline is tomorrow :)
[18:23] <hikiko> ok
[18:23] <hikiko> I am gonna do the compizconfig dependency
[18:23] <hikiko> final decision
[18:23] <Trevinho> hikiko: ok, then fix bugs later... finish the impl first
[18:28] <attente> jbicha: hi, since https://bugs.launchpad.net/ubuntu/+source/bubblewrap/+bug/1643734 is fix released in yakkety, could you also help with sru'ing it to xenial?
[19:14] <jbicha> yes
[20:17] <willcooke> hey robert_ancell
[20:20] <robert_ancell> willcooke, hi
[20:36] <flocculant> robert_ancell: hi there (and you'll now be facepalming I guess :D ) to cut a long story short - we appear to have lost lock/unlock from desktop and suspend too - spent a bit of time today trying to see when it last worked, best I can find is a change in lightdm in November - if I grab an old package - lock works as expected - reported on bug 1656399
[20:36] <flocculant> not sure if you're the person to talk to (still) about that
[20:37] <flocculant> hi willcooke too :)
[20:38] <willcooke> hi flocculant
[20:39] <robert_ancell> flocculant, hey, will have a look. This is through light-locker right?
[20:39] <flocculant> robert_ancell: yup that's right
[20:39] <flocculant> thanks :)
[20:40] <flocculant> at least it's not the month before release this time - hopefully we'll catch these things sooner in future - added specific lock testing to one of our iso.tracker suites
[20:40] <robert_ancell> flocculant, do you have a lightdm.log after a failed lock?
[20:41] <flocculant> I can certainly do the lock locally and add one to the bug - will do that shortly
[20:48] <flocculant> robert_ancell: ok - attached lightdm.log to bug
[20:48] <robert_ancell> flocculant, thanks
[20:49] <flocculant> appear to have got a xorg crash too now after forcing a reboot
[20:49] <robert_ancell> flocculant, we have regression tests for the locking functionality, hopefully whatever is broken can be added to those to stop it regressing
[20:49] <flocculant> which is a duplicate
[20:49] <flocculant> robert_ancell: awesome :)
[20:50] <flocculant> though we will still test it as part of our post installation tests
[20:50] <robert_ancell> the more tests the better!
[20:50] <flocculant> unfortunately - locking isn't something that *I* do normally
[20:50] <flocculant> yea for sure :)
[20:50] <willcooke> night all
[20:50] <flocculant> night wi
[20:50] <flocculant> llcooke
[20:51] <flocculant> he goes quickly :p
[20:51] <robert_ancell> I reckon he has a script that writes "night all" when he quits
[20:51] <flocculant> :)
[20:54] <flocculant> robert_ancell: thanks for looking - if you need anything ask me or bluesabre or ochosi
[20:54] <robert_ancell> will do
[23:21] <flexiondotorg> robert_ancell Evening :-)
[23:31] <robert_ancell> flexiondotorg, hiya