sidi-valencia | may i suggest the codebase should be made tougher rather than the duplication removed? :) | 00:00 |
---|---|---|
bluesabre | that's the plan | 00:03 |
bluesabre | part of it is the different menu implementations | 00:03 |
bluesabre | which is why alacarte does not work with xfce any more | 00:03 |
bluesabre | menulibre adapts to the different oddities | 00:04 |
bluesabre | for example: lxde-settings = "DesktopSettings", gnome-settings = "Preferences", xfce-settings = "Settings" | 00:06 |
bluesabre | it seems each gnome-based thing (gnome, unity, cinnamon, possibly pantheon) use "Preferences" in this case | 00:06 |
bluesabre | or how "Utility" is always "Accessories" | 00:08 |
bluesabre | silly to have a specification that everybody follows differently | 00:08 |
bluesabre | I'm fairly confident that I have the fix now knome | 00:26 |
knome | just a moment, just a moment | 00:28 |
knome | so you want me to take the menu files in the bug | 00:28 |
knome | and edit them accordingly | 00:28 |
bluesabre | sent you an email | 00:32 |
knome | heh, ta | 00:32 |
bluesabre | with the details | 00:32 |
bluesabre | but yeah, try that too (first) | 00:32 |
* knome meh's at some code | 00:34 | |
knome | give me a few mins more. | 00:34 |
bluesabre | take your time | 00:34 |
* bluesabre found a bug before it was reported | 00:35 | |
bluesabre | :D | 00:35 |
* amigamagic finally created its first xfce theme graybird based | 00:38 | |
ochosi | hehe, congrats on that, bluesabre | 00:58 |
ochosi | and night everyone | 00:58 |
amigamagic | ochosi, night | 00:59 |
amigamagic | tomorrow I will post a screenshot of my modded greybird theme | 00:59 |
knome | bluesabre, actually, i think i need to look at the bug tomorrow. sorry | 01:03 |
knome | 4am... | 01:03 |
knome | bluesabre, besides, how am i supposed to try to reproduce? | 01:03 |
knome | i couldn't do that the first time either ;) | 01:03 |
bluesabre | oh | 01:04 |
bluesabre | not sure | 01:04 |
bluesabre | you pulled it off somehow | 01:04 |
bluesabre | :) | 01:04 |
knome | hah | 01:04 |
knome | did you notice that bug was not mine? | 01:04 |
bluesabre | yes | 01:04 |
bluesabre | but I think it is related | 01:04 |
bluesabre | since the file you sent me has the same issue | 01:05 |
knome | aha | 01:05 |
bluesabre | in a way, that bug is the explanation of the bug you reported ;) | 01:05 |
knome | aha, so a dupe? :P | 01:05 |
knome | but yeah, the hard part is that i *really* don't know how to reproduce... | 01:06 |
knome | and thus, in a way, no way to verify the bug is gone... | 01:07 |
knome | it probably has something to do with me trying to get a launcher show in "other" under settings manager and not somewhere else | 01:08 |
deddokatana | OK | 01:18 |
deddokatana | us there a dev in? | 01:18 |
sidi-valencia | Hello? | 01:18 |
deddokatana | where can i file a beta test report for xubuntu desktop utopic | 01:19 |
knome | you can't. | 01:19 |
* sidi-valencia was gonna ping knome | 01:19 | |
knome | sidi-valencia, for what? | 01:19 |
deddokatana | :D ive successfully debootstraped it from the ubuntu repo | 01:19 |
sidi-valencia | ^ | 01:19 |
deddokatana | just a webcam issue, (i think something was forgotten from the xubuntu-desktop metapackage) | 01:20 |
knome | deddokatana, be helpful and run tests later in the cycle when we ask you to | 01:20 |
Logan_ | Unit193: would you consider forwarding that to the Debian maintainer? | 01:21 |
deddokatana | i would, but the proper package is unknown, but i suggest uvccapture | 01:23 |
deddokatana | maybe is should be in the reccomended for cheese | 01:23 |
deddokatana | knome, my twitter is @deddokatana, please contact me to book a vnc / shh / script test session | 01:24 |
deddokatana | i look forward to learning a few tricks | 01:25 |
knome | deddokatana, just join the ~xubuntu-testers launchpad team and you'll get the calls for testing to your inbox | 01:26 |
Logan_ | Unit193: also, looks like we can sync irker now :) | 01:41 |
Unit193 | Yep, sure can. | 01:41 |
Unit193 | (To the second.) | 01:41 |
Unit193 | Logan_: I'll, uh, try and think about forwarding the darkice update back. Just contacted the alpine guy. | 01:48 |
Logan_ | ok | 01:51 |
deddokatana | where would i file an underpowered usb hub to? possible kernel module bug. there was only one device plugged into the hub. | 01:56 |
knome | don't file any bugs against utopic | 01:56 |
deddokatana | cause right now it appears to be a duplicate of trusty? | 01:57 |
knome | appears or not, it's not trusty | 01:58 |
deddokatana | that just boosted my bragging rights amongst my freinds.. | 01:59 |
deddokatana | ok, im going to see if anyone needs help in community support | 02:00 |
pleia2 | ok, updated http://xubuntu.org/press/ | 02:18 |
knome | goodie | 02:19 |
deddokatana | buttery smooth but one mainline kernel may be responsible for 2 of my nuked hard disk.. that or stirring up shit about banks. | 02:21 |
deddokatana | ok heading to off topic | 02:21 |
Unit193 | Logan_: Woah, the alpine maintainer was amazingly receptive. | 06:17 |
jhenke | good morning | 06:18 |
Unit193 | Howdy. | 06:18 |
ochosi | morning | 06:23 |
elfy | morning ochosi | 06:23 |
elfy | ochosi: was reading scrollback in -offtopic - you said to someone - only install xfpm from the PPA - when I tried to do that it wouldn't let me - had to upgrade lightlocker as well | 06:24 |
elfy | is that something that shouldn't have happened? | 06:24 |
elfy | might be why it didn't work for me :) | 06:26 |
elfy | s/install/upgrade | 06:26 |
ochosi | well, at least it was unncessary | 06:30 |
ochosi | why could you not only upgrade xfpm? | 06:30 |
ochosi | after adding the PPA, run "sudo apt-get update && sudo aÃpt-get install xfce4-power-manager" | 06:31 |
elfy | well I didn't install - I upgraded ... | 06:31 |
ochosi | yeah | 06:31 |
elfy | I'll have another go later with installing it | 06:31 |
ochosi | but the install command upgrades too | 06:31 |
ochosi | if there's a newer version | 06:31 |
elfy | yea | 06:31 |
elfy | however, if it does work with that xpfm - then there's something dodgy going on with that lightlocker I guess :p | 06:32 |
elfy | shall report back this afternoon | 06:32 |
ochosi | well the hidden cursor issue is likely unrelated to light-locker | 06:32 |
ochosi | so forget about that package for now | 06:32 |
elfy | I don't get hidden cursor | 06:41 |
elfy | or did once | 06:41 |
ochosi | never got that one myself | 06:42 |
elfy | by once I actually mean 1 time | 06:42 |
Unit193 | Oh, so the terrible oh-no,-screen-won't-come-back-from-suspend seems to only be on lid close, I use the other suspend and it's always fine. | 06:43 |
Unit193 | Oh duh, nevermind. >_< | 06:43 |
elfy | lol - catch up :p | 06:43 |
ochosi | Unit193: any idea where the package glade-xfce went? | 06:45 |
Unit193 | (No, I forgot I didn't count, I use systemd) | 06:45 |
ochosi | used to be around in raring, but not anymore | 06:45 |
Unit193 | !info libxfcegui4-4 | 06:46 |
ubottu | libxfcegui4-4 (source: libxfcegui4): Basic GUI C functions for Xfce4. In component universe, is optional. Version 4.10.0-2 (trusty), package size 225 kB, installed size 1147 kB | 06:47 |
Unit193 | drop glade-xfce package, now provided by libxfce4ui. | 06:47 |
ochosi | right, i read that | 06:48 |
ochosi | but somehow glade-gtk2 still complains about not finding the catalog | 06:48 |
ochosi | so maybe something is messed up there | 06:48 |
Unit193 | Whoops, missed one important one: * debian/libxfcegui4-4.install: | 06:49 |
Unit193 | - stop installing glade files since they are gone. | 06:49 |
ochosi | meh | 06:50 |
ochosi | they're still useful though | 06:50 |
ochosi | just installed the raring package for testing purposes, but now glade-gtk2 segfaults \o/ | 06:50 |
Unit193 | Checking logs. | 06:53 |
Unit193 | ochosi: How badly do you need them and for what? | 06:56 |
ochosi | for being able to modify xfce dialogs with glade | 06:57 |
ochosi | those that use xfcetitleddialog | 06:57 |
Unit193 | ochosi: You can try out the glade-xfce package I have here, but not sure if it'll work. | 09:25 |
ochosi | i'm already on the road to finishing the dialog in mousepad | 09:28 |
ochosi | it's tedious, but at least mousepad doesn't randomely crash ;) | 09:28 |
ochosi | i'll happily try it when i've finished this here though | 09:28 |
Unit193 | Cool, grab the deb from ppa:unit193/staging. | 09:45 |
bluesabre | I usually just edit the glade to make it GtkDialog | 09:46 |
bluesabre | do what I need to do, then change it back | 09:47 |
knome | went ahead and added the community council checkup to the team calendar | 09:55 |
knome | and the next meeting | 09:55 |
ochosi | bluesabre: works fine with gtk3, but with gtk2 i just get segfaults | 09:58 |
bluesabre | probably something not linked correctly | 09:59 |
bluesabre | I think libxfce4-ui is now libxfce4-ui-1 | 09:59 |
bluesabre | right? | 09:59 |
bluesabre | libxfce4ui is now libxfce4ui-1 | 09:59 |
bluesabre | (just a guess, probably wrong) | 10:00 |
Unit193 | 1 is gtk2, 2 is gtk3. | 10:04 |
ochosi | yeah | 10:05 |
ochosi | but that's only the problem of using xfcetitleddialog | 10:05 |
ochosi | i meant: glade-gtk2 just crashes when i create an empty file and try to add a gtkdialog as topÃlevel widget | 10:05 |
bluesabre | ah | 10:06 |
bluesabre | yeah, that version of glade is crashy as hell | 10:06 |
bluesabre | (as are most versions) | 10:06 |
ochosi | well, the gtk3 version is *much* better ;) | 10:06 |
ochosi | anyway, meanwhile finished revamping the whole xfpm dialog | 10:07 |
ochosi | so... | 10:07 |
bluesabre | yay! | 10:07 |
bluesabre | once knome reports back about that bug being fixed, I can probably do a menulibre release | 10:08 |
ochosi | cool | 10:08 |
bluesabre | and minor releases on the other projects are coming up too | 10:09 |
bluesabre | then I'd like to experiment some more with the display settings and try to get a fancy dialog working | 10:09 |
ochosi | sounds cool | 10:09 |
ochosi | only thing is that i've only ever seen a fancy dialog in python so far | 10:10 |
ochosi | or in c, but >2000 lines heavy | 10:10 |
Unit193 | Meh, too many deps for glade-gtk2 or I'd try it. | 10:12 |
knome | bbl | 13:10 |
elfy | !team | 15:12 |
ubottu | bluesabre, elfy, GridCube, jjfrv8, knome, lderan, micahg, mr_pouit, Noskcaj, ochosi, pleia2, skellat, slickymaster, Unit193 | 15:12 |
elfy | please check http://pad.ubuntu.com/5opi9dQG5e | 15:12 |
elfy | that is how I'm planning to call for testing this cycle - dates are there - leaves all the milestone dates free | 15:13 |
elfy | leaves us a big gap at the end that we can call for specific things as required | 15:13 |
elfy | I'll leave the pad now - please use that for comments if any - I'll be sorting this all out early next week. | 15:14 |
elfy | thanks :) | 15:14 |
elfy | at that point I will be building the new testsuites and won't be wanting to fiddle again | 15:15 |
knome | elfy, that looks good to me | 18:42 |
elfy | knome: yep - did me | 19:31 |
ali1234 | elfy: do you still have that VM where you installed ubuntu-desktop^ over xubuntu? | 19:55 |
elfy | ali1234: not any more - but I'm happy enough to set one up and keep it hanging about | 19:55 |
ali1234 | my multimedia keys stopped working after i did that | 19:55 |
elfy | oh - I'm not sure I'd be any help with that - no media keys | 19:56 |
ali1234 | bug 1314782 | 19:58 |
ubottu | bug 1314782 in xfce4-volumed (Ubuntu) "multimedia keys don't work when xfce4-volumed is run in daemon mode" [Undecided,New] https://launchpad.net/bugs/1314782 | 19:58 |
elfy | only media keys I've got are vol+/- and mute | 19:59 |
ali1234 | that will do | 19:59 |
elfy | k - I'll set up the xubuntu one now | 19:59 |
ali1234 | installing ubuntu-sdk seems to have a lot of really nasty side-effects for us | 20:00 |
ali1234 | also, has anyone tested if those lock screen bypass bugs affect us? | 20:01 |
elfy | no idea - which bug #'s | 20:01 |
ali1234 | just checking | 20:01 |
ali1234 | bug 1308572 | 20:01 |
ubottu | bug 1308572 in unity (Ubuntu) "Ubuntu 14.04: security problem in the lock screen" [Critical,Fix released] https://launchpad.net/bugs/1308572 | 20:01 |
ali1234 | that one probably doesn't but there was a new one that involves indicators | 20:02 |
elfy | not seen any references to that one | 20:02 |
elfy | ali1234: so what do you want me to do exactly here - install xubuntu then ubuntu-sdk? | 20:02 |
ali1234 | bug 1313885 | 20:03 |
ubottu | bug 1313885 in Unity "lock screen bypass" [Critical,In progress] https://launchpad.net/bugs/1313885 | 20:03 |
elfy | not seen that either | 20:03 |
ali1234 | elfy: install ubuntu-desktop^ then reboot and test multimedia keys. then if they still work, install ubuntu-sdk and test again | 20:04 |
elfy | ok | 20:04 |
ali1234 | if they still work after that then i'm stumped | 20:04 |
elfy | ok | 20:04 |
elfy | and the vol key doesn't work in vm anyway ... | 20:05 |
ali1234 | hmm | 20:06 |
ali1234 | oh well :/ | 20:06 |
brainwash | did the sdk install any gtk modules? | 20:07 |
brainwash | bug 1239014 seems to be similar | 20:08 |
ubottu | bug 1239014 in xfce4-settings (Ubuntu) "xfsettingsd unable to daemonize properly when overlay scrollbars are activated" [Undecided,Confirmed] https://launchpad.net/bugs/1239014 | 20:08 |
brainwash | and our lock screen is not affected by those bugs, because it's running in a different vt | 20:09 |
ali1234 | still? i thought that changed? (i don't ever lock my computer) | 20:11 |
brainwash | changed? | 20:11 |
brainwash | ah, it's planned to be changed | 20:11 |
elfy | ali1234: seems that multimedia keys are not passed to vbox guest | 20:15 |
sidi-valencia | can you check if gnome-settings-daemon is running? | 20:15 |
sidi-valencia | or gnome-*-daemon for the matter | 20:15 |
ali1234 | maybe try while holding the right ctrl (or whatever the passthrough key is) | 20:16 |
elfy | ali1234: tried that :( | 20:16 |
ali1234 | http://paste.ubuntu.com/7368694/ | 20:16 |
ali1234 | ps waxf | grep gnome ^ | 20:17 |
ali1234 | sidi-valencia: i don't understand why daemon mode makes a difference... really odd | 20:22 |
sidi-valencia | ali1234, will check the code in a minute | 20:24 |
brainwash | other Xfce components do the same (time out) | 20:24 |
ali1234 | ah... i see | 20:24 |
brainwash | appfinder, xfsettingd | 20:24 |
ali1234 | in daemon mode, it never "starts up"? | 20:24 |
ali1234 | cos it hangs somewhere in the init, any idea why/where? | 20:25 |
brainwash | so far it has been caused by the overlay scrollbar gtk module | 20:25 |
ali1234 | yeah, i remember that bug from last year | 20:26 |
ali1234 | but did we ever figure out why or fix it? | 20:26 |
brainwash | appfinder's daemonize routine has been rewritten upstream to use gdbus instead | 20:26 |
ali1234 | i will test the workaround from your bug | 20:26 |
brainwash | this fixed the problem | 20:26 |
sidi-valencia | lookup main.c if curious | 20:26 |
sidi-valencia | it forks | 20:26 |
sidi-valencia | cleans up its std* | 20:26 |
sidi-valencia | nothing special at all | 20:26 |
sidi-valencia | so there might be something going on when it forks, maybe another daemon has a time window to steal the keys? | 20:27 |
ali1234 | brainwash: setting the env var doesn't help. also i don't have overlay scrollbars anywhere. no doubt it is something similar though | 20:27 |
ali1234 | i'll try stracing it i guess | 20:27 |
brainwash | maybe strace it | 20:27 |
sidi-valencia | you could maybe write some sort of test with a delayed SIGALRM and only fork when you receive that, see if you can use the keys before the fork | 20:27 |
sidi-valencia | but im just going to call your system broken and say you should use one volume daemon at a time ;) | 20:27 |
brainwash | xD | 20:28 |
ali1234 | i am though | 20:28 |
ali1234 | i killed all others | 20:28 |
sidi-valencia | ps aux | grep gnome | 20:28 |
sidi-valencia | ps aux | grep pulse | 20:28 |
sidi-valencia | ps aux | grep -v sidi | 20:28 |
sidi-valencia | :D | 20:28 |
ali1234 | yes, i run pulseaudio | 20:28 |
ali1234 | nothing else shows up on that search | 20:28 |
sidi-valencia | (it has nothing to do with pulse obviously ;) ) | 20:28 |
sidi-valencia | hm interesting | 20:29 |
ali1234 | "volume" only shows things related to gvfs, ie disk volumes | 20:29 |
ali1234 | i guess i could start removing ubuntu packages until it works again... not sure that's the reason it is broken though | 20:30 |
ali1234 | right, with strace, the forked child just hangs | 20:31 |
ali1234 | attached gdb... it's still inside keybinder_init | 20:32 |
sidi-valencia | ali1234, can you open main.c and comment out the setsid part? | 20:32 |
sidi-valencia | oh | 20:33 |
sidi-valencia | hm | 20:33 |
ali1234 | it is trying to get a lock on something inside glib | 20:33 |
ali1234 | http://paste.ubuntu.com/7368788/ | 20:34 |
sidi-valencia | bad news | 20:34 |
sidi-valencia | this is not code from me | 20:34 |
ali1234 | well yeah | 20:34 |
ali1234 | this is an ubuntu bug | 20:35 |
sidi-valencia | can you build it with another lib? | 20:35 |
sidi-valencia | or a compiled version? | 20:35 |
ali1234 | what do you mean? | 20:35 |
sidi-valencia | its probably a file being locked by a faulty package | 20:35 |
sidi-valencia | compiled version -> debugging version | 20:35 |
sidi-valencia | sorry me speak bad english | 20:36 |
sidi-valencia | :D | 20:36 |
ali1234 | you mean, build with debug symbols? | 20:36 |
sidi-valencia | keybinder is built with glic i think | 20:36 |
sidi-valencia | unless its a glib thing | 20:37 |
sidi-valencia | trying to figure this out now | 20:37 |
sidi-valencia | waaaha. the amount of dependencies this package pulls in | 20:38 |
sidi-valencia | ali1234, im surprised you *only* have xvd not working | 20:41 |
ali1234 | oh plenty of other stuff was broken too | 20:41 |
sidi-valencia | stuff hanging? | 20:41 |
sidi-valencia | crashing? | 20:41 |
ali1234 | yes, or crashing with security exceptions | 20:41 |
sidi-valencia | and what did you install that could've changed your glib? | 20:41 |
ali1234 | ubuntu-desktop^ and ubuntu-sdk | 20:42 |
ali1234 | but nothing could have changed glib | 20:42 |
sidi-valencia | what is this ^? | 20:42 |
ali1234 | it means a package group, not a package called ubuntu-desktop | 20:43 |
ali1234 | it's about 400mb of packages | 20:43 |
ali1234 | sdk is about another 200mb | 20:43 |
sidi-valencia | ok | 20:45 |
sidi-valencia | nothing replaces the glib, 100% sure? | 20:45 |
ali1234 | of course | 20:45 |
brainwash | got a list of files which were installed? | 20:49 |
sidi-valencia | the kernel? | 20:49 |
sidi-valencia | pthreads? | 20:49 |
sidi-valencia | what hangs is a call to pthread_rwlock_rdlock | 20:49 |
ali1234 | this isn't arch, we don't screw around with different kernels and pthreads implementations :) | 20:49 |
ali1234 | there is only one of each for all ubuntu | 20:50 |
sidi-valencia | ohohohoh | 20:50 |
sidi-valencia | though note that it could be a change to cairo, pango, atk.... | 20:50 |
ali1234 | maybe | 20:50 |
sidi-valencia | any lib launched by xvd, that has been modified by your packages, could cause this if this lib uses glib | 20:50 |
sidi-valencia | or pthread | 20:50 |
ali1234 | i can't figure out how we are building it without gstreamer support | 20:50 |
ali1234 | yes, but the point is that no low level libraries can have been modified, because there is only ever one version of each in ubuntu | 20:51 |
sidi-valencia | and excuse-me but here on arch there are only 2 kernel, a single pthread implementation and no weird patches coming out of nowhere ;) | 20:51 |
ali1234 | we have multiple kernels - they are in different packages, so one can never replace another | 20:52 |
ali1234 | same for everything else really | 20:52 |
sidi-valencia | anyway if you tell me theres only one pango, cairo, etc then i've got no reason to not believe you :) | 20:52 |
sidi-valencia | it would be weird anyway | 20:52 |
ali1234 | let me find the package log... | 20:52 |
sidi-valencia | yet those packages you pulled apparently cause one of these libs to be affected | 20:52 |
sidi-valencia | could we trace what is called in keybinder depending on whether xvd is daemonized? | 20:53 |
ali1234 | if i can figure out how to rebuild it | 20:53 |
sidi-valencia | (also pls try to run it daemonize with setsid() commented out in main.c, i'm a bit curious about this) | 20:53 |
sidi-valencia | do you use systemd? | 20:53 |
ali1234 | no | 20:53 |
ali1234 | http://paste.ubuntu.com/7368880/ <- line 88 is the last time it worked for sure | 20:55 |
ali1234 | note the long lines | 20:55 |
sidi-valencia | you installed ubuntu-desktop and got some systemd and wayland stuff pulled in?? :D | 20:58 |
ali1234 | i already had wayland stuff installed | 20:58 |
* sidi-valencia was trolling but it failed | 20:59 | |
ali1234 | i like systemd | 20:59 |
sidi-valencia | ok so im reading up about setsid stuff, maybe maybe changing the process group and session of the process causes the glib to be unhappy? (/me doesnt know what to explore) | 21:00 |
sidi-valencia | me too! | 21:00 |
ali1234 | yeah, that is a distinct possibility | 21:00 |
ali1234 | would also explain the permissions stuff | 21:00 |
sidi-valencia | can you comment out the setsid bit? | 21:00 |
ali1234 | if i can figure out how to rebuild this thing | 21:00 |
ali1234 | it keeps complaining about missing libraries | 21:00 |
sidi-valencia | well ali1234 i dont really remember why i put that but now that there's cgroups and everything in Linux setsid might have changed meaning a bit.. | 21:01 |
ali1234 | gstreamer stuff that i don't have | 21:01 |
sidi-valencia | oh | 21:01 |
sidi-valencia | send me messages pls | 21:01 |
sidi-valencia | i need to update the build chain | 21:01 |
sidi-valencia | it just builds too easily on my system | 21:01 |
ali1234 | i don't think it is a problem per se | 21:01 |
sidi-valencia | (also pls tell me which ubuntu patches you have if any) | 21:01 |
ali1234 | the ubuntu package obviously uses some configure options to disable that stuff | 21:01 |
ali1234 | there's no distro patches in ubuntu afaik | 21:01 |
ali1234 | the thing is though... why does it only break with ubuntu stuff installed? they must be applying more security restrictions in some config somewhere... | 21:02 |
sidi-valencia | i dont think its security-related | 21:03 |
ali1234 | commenting setsid didn't help | 21:04 |
sidi-valencia | oki | 21:05 |
sidi-valencia | can you also try with freopen, chdir and fork? | 21:05 |
sidi-valencia | (with fork also comment setsid or it might be unhappy) | 21:05 |
ali1234 | the version in ubuntu is a little different | 21:07 |
sidi-valencia | can you pls show me what they changed? | 21:07 |
ali1234 | http://paste.ubuntu.com/7368945/ | 21:07 |
ali1234 | i assume it is just an old version | 21:07 |
ali1234 | well, i'll figure this out the hard way i guess | 21:10 |
ali1234 | let me find my syslog debugging patches... | 21:10 |
ali1234 | i notice that in newer versions, gtk_init has been moved to after the fork | 21:12 |
sidi-valencia | really? | 21:12 |
ali1234 | yeah, compare the paste with HEAD | 21:12 |
ali1234 | HAH that fixes it | 21:13 |
ali1234 | i wonder why we have this old version anyway? | 21:13 |
ali1234 | the version in ubuntu is marked as 0.2.0 | 21:15 |
ali1234 | xfce git seems to be on 0.1.13 | 21:16 |
ali1234 | i don't understand :( | 21:16 |
sidi-valencia | oh you guys probably got a version bump from mr_pouit's patch | 21:21 |
sidi-valencia | is it the -pulse version? it would explain the different version number | 21:22 |
ali1234 | yes | 21:22 |
sidi-valencia | then i guess its normal | 21:22 |
ali1234 | where does that come from? | 21:22 |
sidi-valencia | so the problem was gtk_init before fork? | 21:22 |
ali1234 | yeah | 21:22 |
ali1234 | "a" problem, anyway | 21:22 |
ali1234 | but it still doesn't exlain why it worked before, and then stopped | 21:22 |
ali1234 | but certainly a bug that needs fixing | 21:22 |
sidi-valencia | it doesnt explain anything | 21:23 |
sidi-valencia | at all | 21:23 |
ali1234 | i know, but meh | 21:23 |
sidi-valencia | it shouldnt make a difference | 21:23 |
ali1234 | https://launchpad.net/xfce4-volumed-pulse <- so this is in fact a fork | 21:23 |
knome | sidi-valencia, nice to see you back and working on things! | 21:24 |
sidi-valencia | knome, not really | 21:24 |
sidi-valencia | please | 21:24 |
sidi-valencia | dont say that | 21:24 |
sidi-valencia | :D | 21:25 |
sidi-valencia | it pressures me | 21:25 |
sidi-valencia | :D | 21:25 |
knome | sidi-valencia, yess... | 21:25 |
sidi-valencia | ali1234, kind of | 21:25 |
sidi-valencia | xfce4-volumed is old crappy unmaintained software | 21:25 |
* knome wants to make sidi empathise | 21:25 | |
knome | well at least you're around, and talking | 21:26 |
sidi-valencia | haha | 21:27 |
sidi-valencia | :) | 21:27 |
ali1234 | there's no "kind of" - it says it's a fork right on the LP page :) | 21:27 |
sidi-valencia | ali1234, i dont know what to say. this story is very very weird | 21:27 |
knome | the first step to contributions ;) | 21:27 |
sidi-valencia | knome, im writing SandboxUtils and im appreciative of any sanity tests on the build process | 21:27 |
* knome knows a bit too little about that | 21:28 | |
sidi-valencia | its just some random app | 21:28 |
sidi-valencia | build it and type commands in the README | 21:28 |
sidi-valencia | it does *nothing* anyway | 21:28 |
sidi-valencia | it doesnt even sandbox anything | 21:28 |
knome | lol | 21:28 |
knome | i guess i can look at that later | 21:28 |
ali1234 | brainwash: would you like to check if xfsettingsd also forks after gtk_init()? | 21:30 |
ali1234 | so... daemon mode got moved to after gtk_init to support using gtk to parse command line | 21:32 |
sidi-valencia | bbl | 21:33 |
ali1234 | although i don't see why tbh | 21:33 |
ali1234 | it's only using glib | 21:33 |
ali1234 | hmm | 21:34 |
ali1234 | why does it even use fork() to go daemon? | 21:37 |
ali1234 | "It seems that gtk_init connect to dbus session. Currently, it is called before fork, i.e. parent process. Parent process quit after the checking gui enable. So dbus session become disconnected." | 21:39 |
ali1234 | i bet this is the problem | 21:39 |
sidi-valencia | ali1234, anyway command line will be exactly the same before and after the fork?? | 21:40 |
sidi-valencia | i use fork because i want to detach from the terminal | 21:41 |
sidi-valencia | and you shouldnt setsid without fork in case you're already detached | 21:41 |
sidi-valencia | cant setsid if youre a process group leader | 21:41 |
sidi-valencia | whatever that means | 21:41 |
sidi-valencia | posix obscurities | 21:41 |
sidi-valencia | anyways, you have to fork+setsid because, erm, that's life :D | 21:41 |
ali1234 | hmm, as i suspected, xfsettingsd also calls gtk_init (and also opens a dbus connection) before forking | 21:42 |
ali1234 | i will look at what gnome-settings-daemon does i guess | 21:43 |
ali1234 | gnome-settings-daemon doesn't fork, anywhere | 21:48 |
ali1234 | or if it does, it uses some glib function to do it | 21:49 |
brainwash | ali1234: http://git.xfce.org/xfce/xfce4-appfinder/commit/?id=4a065a10945c72c985e254ff1ef13df188f3e11e | 21:50 |
ali1234 | daemon mode was removed from gnome-settings-daemon because "it confuses dbus" | 21:51 |
sidi-valencia | google "fork before gtk_init"? | 21:51 |
ali1234 | brainwash: what is that? | 21:52 |
ali1234 | okay, i think i understand what is happening | 21:54 |
ali1234 | normally gtk_init() does not connect to dbus | 21:54 |
ali1234 | however, overlay scrollbar is implemented as a preload library | 21:54 |
ali1234 | *it* almost certainly connects to dbus | 21:54 |
ali1234 | so if enabled, gtk_init() now will connect to dbus earlier than normal | 21:55 |
ali1234 | then when you fork, dbus gets "confused" and bad things happen | 21:55 |
ali1234 | it might not be overlay scrollbar, itmight be some other ubuntu hack | 21:55 |
ali1234 | like appmenus for example | 21:55 |
ali1234 | yes | 21:56 |
ali1234 | it is UBUNTU_MENUPROXY | 21:56 |
ali1234 | unsetting it, and then running the volumed fixes the problem | 21:57 |
ali1234 | so this is in fact the same old xfsettingsd bug, with a slightly different way to get to the same problem | 21:57 |
brainwash | aha | 21:58 |
ali1234 | so now we can fix both these problems properly :) | 21:58 |
brainwash | the linked commit fixed the problem for the appfinder | 21:58 |
ali1234 | that probably causes the child process to open a new dbus connection rather than reusing the gtk one | 21:58 |
ali1234 | so fix was a side effect | 21:59 |
brainwash | yes, it's a side effect | 21:59 |
ali1234 | so fixing bug 1307657 would fix this | 22:02 |
ubottu | bug 1307657 in xfce4-indicator-plugin (Ubuntu) "UBUNTU_MENUPROXY should not be set in Xfce" [Undecided,Confirmed] https://launchpad.net/bugs/1307657 | 22:02 |
sidi-valencia | so the fix is sudo aptitude purge ubuntu*? | 22:07 |
sidi-valencia | :D | 22:07 |
ali1234 | i can almost guarantee that xvd is hanging on a dbus call to get the keyboard layout from x server | 22:07 |
ali1234 | and yes | 22:07 |
ali1234 | (but we already knew that, just didn't know why) | 22:07 |
ali1234 | so i think we should go through and fix all the gtk_init() before fork() in all xfce software | 22:08 |
ali1234 | but i'm not sure of the best way to do it | 22:09 |
sidi-valencia | now i still have a question | 22:11 |
sidi-valencia | why on earth do i use gtk_init?? | 22:11 |
ali1234 | well -pulse uses it to parse command line | 22:11 |
sidi-valencia | it was here before | 22:11 |
sidi-valencia | i dont remember why i use it | 22:11 |
sidi-valencia | i think its just out of convenience | 22:11 |
ali1234 | yeah but after the fork() which is fine | 22:11 |
sidi-valencia | why did they move it? | 22:12 |
sidi-valencia | what was the rationale? | 22:12 |
ali1234 | so that gtk_init(argv, argc); would work | 22:12 |
sidi-valencia | and how is vanilla XFCE? | 22:12 |
ali1234 | i can show you the commit | 22:12 |
sidi-valencia | but fork does not clear argc/argv, does it?? | 22:12 |
ali1234 | no, but they added a --no-daemon command line | 22:12 |
ali1234 | not much point parsing that after you alread went daemon | 22:12 |
ali1234 | http://bazaar.launchpad.net/~mrpouit/xfce4-volumed-pulse/master/revision/53#src/main.c | 22:13 |
ali1234 | mr_pouit: you broke it again :) | 22:13 |
ali1234 | there are ways to parse the command line without calling gtk_init though | 22:14 |
ali1234 | we should probably do that | 22:14 |
sidi-valencia | oh | 22:14 |
ali1234 | wait | 22:14 |
sidi-valencia | really | 22:14 |
ali1234 | actually | 22:14 |
ali1234 | i think it's already doing that with glib | 22:14 |
ali1234 | so i dunno why gtk_init() got moved | 22:14 |
sidi-valencia | but thats not a reason. i have a --no-daemon option they can just change the name of my app instead | 22:15 |
ali1234 | maybe so that gtk gets gtk specific command lines earlier | 22:15 |
sidi-valencia | of my app's command line | 22:15 |
ali1234 | yeah, i think you;re right | 22:15 |
ali1234 | i don't know what's going on here | 22:15 |
ali1234 | maybe it was just a silly mistake; it happens | 22:15 |
sidi-valencia | it is just a silly mistake | 22:15 |
sidi-valencia | in fact the original change | 22:16 |
sidi-valencia | did not anticipate at all the changes made to gtk_init by Canonical | 22:16 |
sidi-valencia | but i think now that there is a reason why it happens more often with the ubuntu codebase than other distros ;P | 22:16 |
ali1234 | absolutely, it's because of the weird gtk plugins they made that aren't thread safe | 22:17 |
ali1234 | or rather, fork() safe | 22:17 |
sidi-valencia | because when you do undocumented hack-ish changes here and there only testing with your own packages, but having dependencies based on other hackers then at some point some of them play poorly. I must say kudos to you ali1234 because i would never have identified the origin of this bug as quickly (in fact im a bit jealous ;) ) | 22:17 |
sidi-valencia | i know xfce4-volumed is terribly dirty software... yet it somehow gets used | 22:17 |
ali1234 | we have this problem a lot | 22:18 |
sidi-valencia | i think we should make sure to identify exactly whats wrong with it and ruin the life of the original author (aka me) until he pushes the *proper* fix rather than ubuntu patches | 22:18 |
sidi-valencia | im very very grateful to mr_pouit for his pulse patch. but there should be no other ubuntu/xubuntu change to vanilla xfce software than that | 22:18 |
ali1234 | well the xfce4-volumed is fine, it's the -pulse version that is broken :) | 22:18 |
sidi-valencia | because then i get bug reports that i cant make sense of ;) | 22:18 |
sidi-valencia | yeah but its not the -pulse code that breaks | 22:18 |
sidi-valencia | its the *other* changes | 22:18 |
ali1234 | yeah | 22:19 |
sidi-valencia | which dont seem justified, which i dont know about and which could be done more neatly | 22:19 |
ali1234 | still, this is why i reported the bug on lp, not xfce bugs | 22:19 |
ali1234 | i knew all along it was something ubuntu did | 22:19 |
sidi-valencia | (for instance i could upstream that daemon() syscall if asked why i should do it, and i could also use --no-daemon rather than --no-daemonize or whatever i have) | 22:19 |
ali1234 | because it worked before | 22:19 |
sidi-valencia | yeah well im only subscribed to LP actually ;p | 22:19 |
ali1234 | i didnt even know there were two versions of this software | 22:20 |
sidi-valencia | this app uses launchpad for bug reports, as far as i know | 22:20 |
sidi-valencia | yeah its about time i merge mr_pouit 's code | 22:20 |
ali1234 | -pulse does, but i reported against the ubuntu package, not the upstream software (which i didn't know existed) | 22:20 |
sidi-valencia | but guys please always report on the XFCE bugzilla or devel ML when you patch Xfce | 22:20 |
sidi-valencia | whatever it is, unless it's literally just libubuntusomething | 22:20 |
sidi-valencia | ali1234, i just meant that im subscribed everywhere anyway. im not blaming you at all for reporting the bug haha | 22:21 |
ali1234 | yeah, kudos for actually paying attention to distro bugs. a lot of devs would just say "report it on upstream" (and then close it for being distro specific) | 22:22 |
ali1234 | and especially given you don't even use ubuntu... | 22:22 |
sidi-valencia | do you have pointers to the ubuntu code that causes DBus calls in gtk init? | 22:22 |
sidi-valencia | ali1234, you guys write amazing software for me | 22:23 |
sidi-valencia | so i want my software to work for you | 22:23 |
sidi-valencia | i hate canonical but i love xubuntu | 22:23 |
ali1234 | something in here: http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/utopic/unity-gtk-module/utopic/view/head:/src/main.c | 22:24 |
ali1234 | is what cuases it | 22:24 |
ali1234 | unity-gtk-module is loaded if UBUNTU_MENUPROXY=1 and then it "hijacks" menus and exports them on dbus | 22:25 |
ali1234 | it's how they implement global menus | 22:25 |
sidi-valencia | https://gitorious.org/gnome-settings-daemon/gnome-settings-daemon/commit/c0400bb599e6a57a46274a4ea3b550365577c93c | 22:27 |
sidi-valencia | very interesting | 22:27 |
sidi-valencia | so anyone using gtk_init should fork first or expect trouble | 22:28 |
bluesabre | knome: ping | 22:28 |
sidi-valencia | i asked on #gtk+ for some wisdom about this | 22:28 |
ali1234 | yeah, there are loads of references to such bugs. basically gtk_init() before fork() is a bad idea, even if it usually works | 22:28 |
ali1234 | okay simply moving the call to gtk_init() fixes it, and no side effects with arguments or anything | 22:32 |
sidi-valencia | great | 22:34 |
sidi-valencia | pls change the --no-daemonize if need be | 22:35 |
sidi-valencia | ali1234, thanks again. you did awesome work. | 22:41 |
sidi-valencia | How did you identify the origin of the bug actually? | 22:41 |
ali1234 | what brainwash said made something click | 22:41 |
sidi-valencia | hmm | 22:41 |
sidi-valencia | brainwash, how did you come up with what you said? :D | 22:41 |
ali1234 | actually cos i noticed that the old version has the fork after init | 22:42 |
ali1234 | so i tried swapping them just at random to see what would happen | 22:42 |
ali1234 | we've had this problem before but never really figured out why it happened.. it just happened at random | 22:44 |
ali1234 | and then randomly got fixed | 22:44 |
ali1234 | and i have quite a lot of pointless knowledge about how ubuntu hacks work | 22:44 |
ali1234 | because they tend to break everything | 22:45 |
ochosi | holy crap, so much backlog... | 22:47 |
ali1234 | tl;dr every xfce app that calls fork() after gtk_init() needs fixing | 22:48 |
ali1234 | this includes xfsettingsd and xfce4-volumed-pulse | 22:48 |
ali1234 | i just put in a MR for the latter | 22:48 |
sidi-valencia | yeah | 23:05 |
sidi-valencia | great | 23:05 |
sidi-valencia | good job | 23:05 |
sidi-valencia | i'd love so badly to put up a Xubuntu-like team on another distro... | 23:06 |
sidi-valencia | if you guys werent losing time over ubuntu upkeep who knows what you'd be up to? :D | 23:06 |
bluesabre | losing time over gnome upkeep | 23:09 |
bluesabre | :) | 23:09 |
bluesabre | and/or gtk | 23:09 |
ali1234 | http://askubuntu.com/questions/458238/how-to-revert-from-gtk3-to-gtk2-indicators-in-xfce4-panel | 23:10 |
ochosi | hope we'll be able to replace the indicators with xfce panel plugins sometime in the not so far future | 23:10 |
ali1234 | anyone want to add anything to that? | 23:10 |
sidi-valencia | bluesabre, this is unity upkeep | 23:13 |
sidi-valencia | gtk upkeep is costly enough but what does unity have to do with us? ;) | 23:13 |
sidi-valencia | correct me if im wrong but Xfce still uses only GTK2 right? | 23:13 |
ochosi | it does, but gtk3 dropped lots of stuff lately and introduced some new things that will make some of our core apps a bit weird/borked | 23:14 |
bluesabre | parole is full gtk3, some other components compile as gtk2 or gtk3 | 23:14 |
sidi-valencia | oh i see | 23:15 |
ali1234 | sidi-valencia: do you have indicators in arch? | 23:16 |
bluesabre | brainwash: poke | 23:25 |
bluesabre | https://code.launchpad.net/~thad-fisch/lightdm-gtk-greeter/zoomed | 23:25 |
bluesabre | I'm assuming this also handles the positioning, and is probably tested? :) | 23:25 |
bluesabre | and do you suspect this is related? | 23:26 |
bluesabre | https://bugs.launchpad.net/lightdm-gtk-greeter/+bug/1300153 | 23:26 |
ubottu | Launchpad bug 1300153 in lightdm-gtk-greeter (Ubuntu) "login Screen is not showing correctly" [Undecided,New] | 23:26 |
ochosi | bluesabre: i haven't gotten around to testing it tbh | 23:26 |
ali1234 | that bug doesn't look related | 23:26 |
ali1234 | or rather, fixing the wallpaper won't fix it | 23:27 |
ali1234 | basically gtk-greeter is not multimonitor aware... at all | 23:27 |
bluesabre | that bug is single-display | 23:27 |
ali1234 | ok then in that case it is definitely not related | 23:27 |
bluesabre | but I hope to fix the multimonitor stuff with the next release | 23:27 |
bluesabre | :) | 23:28 |
ali1234 | brainwash' patch just makes it draw the wallpaper spanning monitors rather than once per monitor (even if they overlap) | 23:28 |
ali1234 | as i understand it, anyway | 23:28 |
sidi-valencia | ali1234, sudo make uninstall works for that guy over ~ askubuntu | 23:28 |
ali1234 | i have not tested it either | 23:28 |
sidi-valencia | ali1234, libindicator? that stuff from 2010 from Ubuntu? | 23:28 |
bluesabre | ah | 23:28 |
ali1234 | sidi-valencia: oh really? okay then | 23:28 |
bluesabre | he had mentioned improper zooming and positioning at one point | 23:29 |
bluesabre | I also have not tested it | 23:29 |
ali1234 | yes, of the wallpaper | 23:29 |
sidi-valencia | it'll erase the files regardless of whether they were created by his install though. it just removes files where it would have put them | 23:29 |
ali1234 | yeah, not all software has uninstall though | 23:29 |
sidi-valencia | with autotools i think it should be. | 23:34 |
bluesabre | anybody with an encrypted home and multiple languages want to verify if this is fixed in trusty? | 23:59 |
bluesabre | https://bugs.launchpad.net/lightdm-gtk-greeter/+bug/1002706 | 23:59 |
ubottu | Launchpad bug 1002706 in lightdm-gtk-greeter (Ubuntu) "Language setting ignored" [Undecided,New] | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!