*** M4rtinK has quit IRC | 00:11 | |
*** arcean has quit IRC | 00:32 | |
*** gabriel9 has quit IRC | 00:49 | |
*** gabriel9 has joined #nemomobile | 00:50 | |
*** zalan_ has quit IRC | 01:05 | |
*** vakkov has joined #nemomobile | 01:25 | |
*** mjones has quit IRC | 02:16 | |
*** chriadam_ has joined #nemomobile | 02:16 | |
*** chriadam has quit IRC | 02:16 | |
*** mjones has joined #nemomobile | 02:17 | |
*** Morpog_PC has joined #nemomobile | 02:44 | |
*** sletta has joined #nemomobile | 02:45 | |
*** gabriel9 has quit IRC | 03:12 | |
*** gabriel9 has joined #nemomobile | 03:12 | |
*** hurrian has quit IRC | 03:14 | |
*** marquiz has quit IRC | 03:17 | |
*** marquiz has joined #nemomobile | 03:17 | |
*** mjones has quit IRC | 03:56 | |
*** martyone__ has joined #nemomobile | 04:09 | |
*** zalan has joined #nemomobile | 04:16 | |
*** VDVsx has quit IRC | 04:34 | |
*** zhxt_afk has quit IRC | 04:35 | |
*** Oksana has quit IRC | 04:41 | |
*** jonwil has quit IRC | 04:43 | |
*** Venemo_j has joined #nemomobile | 04:45 | |
*** VDVsx has joined #nemomobile | 04:45 | |
*** zhxt_afk has joined #nemomobile | 04:48 | |
*** Pat_o has joined #nemomobile | 04:48 | |
*** mjones has joined #nemomobile | 04:48 | |
*** bob has joined #nemomobile | 04:50 | |
*** spiiroin has quit IRC | 04:54 | |
*** Oksana has joined #nemomobile | 04:55 | |
*** Morpog_PC has quit IRC | 04:57 | |
*** WWDrakey has joined #nemomobile | 05:08 | |
*** Oksana has quit IRC | 05:14 | |
*** spiiroin has joined #nemomobile | 05:17 | |
locusf | morning | 05:18 |
---|---|---|
*** simbrown has joined #nemomobile | 05:18 | |
lpotter | hello | 05:22 |
*** spiiroin has quit IRC | 05:23 | |
*** filippz has joined #nemomobile | 05:24 | |
*** phaeron has joined #nemomobile | 05:28 | |
tbr | moaning | 05:33 |
*** spiiroin has joined #nemomobile | 05:34 | |
Stskeeps | morn tbr | 05:36 |
*** zhxt_afk is now known as zhxt | 05:45 | |
*** Venemo_j has quit IRC | 05:46 | |
*** mjones has quit IRC | 05:54 | |
*** mjones has joined #nemomobile | 05:55 | |
*** mjones has quit IRC | 05:59 | |
*** arturo182 has quit IRC | 06:05 | |
*** zhxt has quit IRC | 06:12 | |
*** mjones has joined #nemomobile | 06:15 | |
*** plfiorini has quit IRC | 06:17 | |
*** Sfiet_Konstantin has joined #nemomobile | 06:18 | |
*** gabriel9 has quit IRC | 06:26 | |
*** veskuh has joined #nemomobile | 06:29 | |
*** zhxt has joined #nemomobile | 06:30 | |
*** piggz has quit IRC | 06:37 | |
*** flash1 has joined #nemomobile | 06:42 | |
*** Sfiet_Konstantin has quit IRC | 06:46 | |
*** cxl000 has joined #nemomobile | 07:11 | |
*** jonwil has joined #nemomobile | 07:11 | |
*** sletta has quit IRC | 07:19 | |
*** sletta has joined #nemomobile | 07:25 | |
*** sletta has quit IRC | 07:31 | |
*** hedayat has joined #nemomobile | 07:32 | |
faenil | morning o/ | 07:35 |
hedayat | faenil: morning :) | 07:43 |
*** alien_ has joined #nemomobile | 07:48 | |
*** SK_work has joined #nemomobile | 07:51 | |
*** arcean has joined #nemomobile | 08:22 | |
*** r0kk3rz has joined #nemomobile | 08:22 | |
*** kjokinie has quit IRC | 08:23 | |
*** kjokinie has joined #nemomobile | 08:23 | |
*** olesalscheider has quit IRC | 08:27 | |
*** notmart has joined #nemomobile | 08:28 | |
*** olesalscheider has joined #nemomobile | 08:32 | |
*** jpetrell has joined #nemomobile | 08:35 | |
*** jpetrell_ has joined #nemomobile | 08:46 | |
*** jpetrell has quit IRC | 08:47 | |
*** jmlich has joined #nemomobile | 08:51 | |
*** ryukafalz has quit IRC | 09:12 | |
*** kjokinie has quit IRC | 09:15 | |
*** krnlyng has quit IRC | 09:24 | |
*** krnlyng has joined #nemomobile | 09:26 | |
*** ryukafalz has joined #nemomobile | 09:26 | |
*** jpetrell_ has quit IRC | 09:32 | |
*** kjokinie has joined #nemomobile | 09:35 | |
*** jpetrell has joined #nemomobile | 09:36 | |
*** niqt has joined #nemomobile | 09:37 | |
*** hurrian has joined #nemomobile | 09:42 | |
*** hurrian has quit IRC | 09:42 | |
*** hurrian has joined #nemomobile | 09:42 | |
locusf | hedayat: I'm very interested on how your solution works in practise :) | 09:50 |
filippz | hello o/ | 10:13 |
filippz | locusf, faenil: I've set up my N9 with the files from jolla-session-qt5 (including autologind binary) and I'm in lipstick for more than 20 minutes without problems | 10:13 |
locusf | filippz: nice | 10:13 |
filippz | I'm not sure how to proceed - autologind is not FOSS, and to be honest, I'm not sure what exactly it does. Maybe we could use some different mechanism for autologin to avoid this binary? | 10:13 |
* Stskeeps looks | 10:13 | |
Stskeeps | rpm -qf it? | 10:13 |
faenil | yes, Stskeeps should definitely give us a hand here :p | 10:14 |
* Stskeeps starts laughing | 10:14 | |
* faenil takes out the Mer whip | 10:14 | |
locusf | user session does survive for a long time on android frankensteins | 10:15 |
Stskeeps | filippz: it does while (true) sleep(long time) | 10:15 |
Stskeeps | i'll let you reverse engineer that | 10:15 |
Stskeeps | :P | 10:15 |
faenil | lol | 10:15 |
filippz | Stskeeps: that's it? I just love systemd | 10:16 |
Stskeeps | that's it. | 10:16 |
* faenil scratches his head | 10:16 | |
filippz | why not make it public/part of systemd-config-mer then? | 10:17 |
faenil | because they don't want to show that they have a crap solution :D | 10:17 |
Stskeeps | neglegt? | 10:17 |
Stskeeps | neglect | 10:17 |
Stskeeps | nono, it makes technical sense.. | 10:17 |
faenil | does it? it doesn't look like something you're supposed to be doing | 10:18 |
tbr | "valuable differentiator, proprietary IPR" | 10:18 |
*** sletta has joined #nemomobile | 10:19 | |
Stskeeps | as in | 10:19 |
Stskeeps | the solution makes sense | 10:19 |
faenil | really? | 10:20 |
filippz | faenil: suggestions? Fork systemd and put jolla stuff inside? Or be annoying and talk jolla guys into making this a part of systemd? | 10:21 |
Stskeeps | you're likely to have your own sessions packages, afaik | 10:22 |
faenil | sure we do | 10:22 |
locusf | https://github.com/nemomobile/nemo-mobile-session <- ? | 10:22 |
faenil | filippz: so, are we saying systemd doesn't have enough tools to bring up a user session? | 10:22 |
faenil | sorry guys I'm a bit missing the big picture here | 10:23 |
filippz | faenil: yes and no - we need some kind of autologin capabailities for systemd - ie. systemd would work with some login mechanism,but we want for user to be autologged in | 10:25 |
Stskeeps | you should take a look at the /lib/systemd/system/ stuff | 10:26 |
faenil | filippz: how was that different before | 10:26 |
filippz | newer systemd now runs systemd --user as root, or something to that affect | 10:27 |
faenil | while before? | 10:28 |
filippz | https://wiki.archlinux.org/index.php/Systemd/User says that changed with systemd 206 (note in Basic setup part) | 10:29 |
filippz | I've seen solutions to autologin using agetty - maybe we could try that route, but I think there's a reason jolla did with that autologind binary | 10:30 |
faenil | it doesn't say much | 10:31 |
phaeron | the only reason maybe that we started with an older systemd that didn't have user session capability | 10:31 |
phaeron | but I am not a qualified source for this info :) | 10:31 |
tbr | systemd proudly removing necessary embedded linux features | 10:32 |
faenil | ah, so older systemd didn't have user session at all? :O this sounds strange | 10:32 |
tbr | but at least they are getting better | 10:33 |
tbr | someone from pengutronix has been submitting embedded related patches and they got in | 10:33 |
jonwil | What I want to know is why so many people have jumped on the systemd bandwagon even when they admit its not so good for their use case or otherwise crap... | 10:33 |
jonwil | At least there are people willing to say "systemd sucks, lets find a better way" | 10:33 |
faenil | please let's not switch to an ethical systemd discussion | 10:34 |
tbr | it's OK for a certain set of things, but I wouldn't have it on a very constrained embedded system (<64/128M RAM) | 10:34 |
faenil | we have to fix nemo issues here :D | 10:34 |
tbr | yeah, let's have an unethical discussion instead! | 10:34 |
faenil | ah damn you :P | 10:34 |
bencoh | (nemo as it is wouldnt run on "very constrained embedded" as it is anyway ;p) | 10:39 |
filippz | any conclusion - or will we sit on this for a while? | 10:40 |
faenil | filippz: so... | 10:40 |
faenil | the main issue is that systemd only starts the user session *after* a user has logged in, right? | 10:41 |
faenil | so we *just* need something to autologin user nemo, right? | 10:41 |
faenil | (and keep the session open) | 10:41 |
filippz | faenil: it seems so | 10:43 |
filippz | does anybody know why agetty or similar solution wasn't chosen? | 10:43 |
faenil | something like https://wiki.archlinux.org/index.php/Automatic_login_to_virtual_console#Virtual_console ? | 10:44 |
*** zhxt_ has joined #nemomobile | 10:45 | |
*** jmlich has quit IRC | 10:46 | |
filippz | afk for 30 minutes :( | 10:47 |
*** jmlich has joined #nemomobile | 10:51 | |
*** jmlich has quit IRC | 10:51 | |
*** zhxt_ has quit IRC | 10:52 | |
r0kk3rz | isnt that wha logind is for? | 10:52 |
*** niqt has quit IRC | 10:53 | |
faenil | Aren't linux distros already handling this? what is fedora doing? | 10:57 |
faenil | maybe not relevant because it's still x11 based? | 10:57 |
faenil | Stskeeps: isn't it enough to login the user? why the continuous sleep to keep the pam session oppen? | 10:58 |
Stskeeps | i think it's not, but i don't have a systemd expert at hand | 10:59 |
faenil | ok | 10:59 |
*** niqt has joined #nemomobile | 10:59 | |
*** niqt has quit IRC | 11:00 | |
faenil | let's look for experts in the appropriate channel | 11:00 |
*** spiiroin has quit IRC | 11:02 | |
r0kk3rz | just get lennart to explain it all to you | 11:04 |
*** tg has quit IRC | 11:07 | |
*** phaeron has quit IRC | 11:08 | |
faenil | I wish he would | 11:08 |
faenil | why is Pekka not on #nemomobile anymore :D :P | 11:09 |
*** Venemo has joined #nemomobile | 11:11 | |
*** jmlich has joined #nemomobile | 11:20 | |
*** jmlich has quit IRC | 11:20 | |
*** msava has quit IRC | 11:24 | |
*** vakkov has quit IRC | 11:26 | |
*** tg has joined #nemomobile | 11:27 | |
*** flash1 has joined #nemomobile | 11:27 | |
*** spiiroin has joined #nemomobile | 11:32 | |
*** flash1 has quit IRC | 11:34 | |
*** sletta has quit IRC | 11:35 | |
*** DrCode has quit IRC | 11:37 | |
*** mhall119|vacatio has quit IRC | 11:37 | |
*** mhall119 has joined #nemomobile | 11:37 | |
*** Merbot has joined #nemomobile | 11:58 | |
*** martyone__ has quit IRC | 12:03 | |
*** vakkov has joined #nemomobile | 12:12 | |
*** DrCode has joined #nemomobile | 12:12 | |
*** phaeron has joined #nemomobile | 12:26 | |
*** jonwil has quit IRC | 12:29 | |
*** hurrian has quit IRC | 12:33 | |
*** flash1 has joined #nemomobile | 12:35 | |
*** vakkov has quit IRC | 12:35 | |
*** Venemo has quit IRC | 12:36 | |
*** hurrian has joined #nemomobile | 12:39 | |
*** hurrian has quit IRC | 12:39 | |
*** hurrian has joined #nemomobile | 12:39 | |
*** msava has joined #nemomobile | 12:44 | |
faenil | a sum up of Sailfish login process files | 12:54 |
faenil | http://pastie.org/private/ahygzoj72gi6lq9fvb2cdq# | 12:54 |
faenil | wait-user-session basically just sleeps (or returns error) until $(systemctl is-active $USER_SERVICE) returns "active" | 12:55 |
faenil | with a timeout of 45secs | 12:55 |
faenil | if the timeout expires and user@.service is not active yet, it restarts the user session, a maximum of two times | 12:56 |
*** filippz has quit IRC | 13:07 | |
*** filippz has joined #nemomobile | 13:07 | |
*** filippz has quit IRC | 13:08 | |
faenil | mmm | 13:29 |
faenil | PAMName= Sets the PAM service name to set up a session as. If set, the executed process will be registered as a PAM session under the specified service name. This is only useful in conjunction with the User= setting. If not set, no PAM session will be opened for the executed processes. See pam(8) for details. | 13:29 |
faenil | [15:29] <faenil> so systemd can actually start pam sessions by itself? [15:30] == bagackiz [bagackiz@Why.Is.A.Raven.Like.A.Writing.Desk.PanicBNC.org] has joined #systemd [15:30] == bagackiz [bagackiz@Why.Is.A.Raven.Like.A.Writing.Desk.PanicBNC.org] has quit [Max SendQ exceeded] [15:30] <grawity> yes, though those are generally missing the information about X11 seat/display [15:30] <faenil> grawity: our OS is wayland based [15:30] <g | 13:31 |
faenil | ouch | 13:31 |
faenil | [15:29] <faenil> so systemd can actually start pam sessions by itself? | 13:32 |
faenil | [15:30] <grawity> yes, though those are generally missing the information about X11 seat/display | 13:32 |
faenil | [15:30] <faenil> grawity: our OS is wayland based | 13:32 |
faenil | [15:30] <grawity> so not particularly useful for things that rely on logind for permission checking | 13:32 |
Stskeeps | that's why we have the Environment= stuff.. | 13:32 |
faenil | [15:30] <grawity> same concepts | 13:32 |
faenil | [15:31] <faenil> mm, maybe then sailfish doesn't rely on logind for permission checking? :/ | 13:32 |
faenil | Stskeeps: right.. | 13:33 |
faenil | then I guess the picture is complete! | 13:35 |
faenil | Stskeeps: now I understand why the sleep...it's a hack, it really is...but, well, so it goes :p | 13:40 |
Stskeeps | welcome to modern mobile devices. | 13:41 |
faenil | is it too heavy to install a login manager? | 13:41 |
Stskeeps | i don't see why you'd need one | 13:42 |
faenil | to open a pam session and close it when neede instead of using tricks to keep a session open | 13:42 |
faenil | at this point I wouldn't be surprised if the system shutdown began with a "kill -9 autologind" :p | 13:45 |
*** zalan has quit IRC | 13:46 | |
*** zalan has joined #nemomobile | 13:49 | |
*** Behold has joined #nemomobile | 13:57 | |
*** WWDrakey has left #nemomobile | 13:59 | |
*** Venemo has joined #nemomobile | 14:03 | |
faenil | Stskeeps: next steps for Mer ? (I gave you monday for free :P ) | 14:10 |
*** VDVsx has quit IRC | 14:11 | |
Stskeeps | ar, don't ask me that 15 mins before i have to leave from work.. | 14:14 |
Stskeeps | ask in the mornings | 14:14 |
faenil | Stskeeps: gah :p | 14:14 |
*** veskuh has quit IRC | 14:22 | |
hedayat | locusf: Thanks! I'm leaving for now, but n9_easy_boot.tar.xz will be hopefully show up in https://hedayat.fedorapeople.org/n9/ in about 5 minutes when its upload is finished. I'll write the TMO announcement when I'm back. | 14:24 |
locusf | hedayat: ok see ya | 14:25 |
hedayat | I'll be happy if anybody tries it and see if it has any problems. | 14:25 |
hedayat | :) | 14:25 |
faenil | :) | 14:26 |
*** VDVsx has joined #nemomobile | 14:37 | |
*** r0kk3rz has joined #nemomobile | 14:44 | |
*** cxl000 has quit IRC | 14:57 | |
*** cxl000 has joined #nemomobile | 15:00 | |
*** alien_ has quit IRC | 15:06 | |
*** alien_ has joined #nemomobile | 15:19 | |
*** gabriel9 has joined #nemomobile | 15:38 | |
*** flash1 has quit IRC | 15:42 | |
*** Morpog_PC has joined #nemomobile | 15:50 | |
*** vakkov has joined #nemomobile | 15:51 | |
*** zalan has quit IRC | 15:58 | |
*** filippz has joined #nemomobile | 15:59 | |
*** Morpog_PC has quit IRC | 16:00 | |
*** Morpog_PC has joined #nemomobile | 16:01 | |
filippz | locusf, faenil: fixes used in ks file for SailfishOS also work well for nemo: https://www.dropbox.com/s/uh1dhnp59oqhed7/20141021%20006.jpg?dl=0 https://www.dropbox.com/s/qd2e8pfrh17f05c/20141021%20008.jpg?dl=0 | 16:02 |
Morpog_PC | lovely filippz! | 16:03 |
locusf | filippz: excellent! | 16:03 |
filippz | I'm thinking of leaving this as is - without fiddling with the settings in kernel | 16:03 |
faenil | filippz: great, which fixes? | 16:04 |
filippz | fbset-N9.service and 60-n9-n950-ui.conf are patched from ks file | 16:04 |
faenil | filippz: ah ok it's only for the rotation stuff | 16:05 |
faenil | I didn't follow the issue so I barely know the subject :) | 16:05 |
filippz | faenil: yes | 16:05 |
faenil | great anyway :) | 16:05 |
faenil | filippz: did you see my notes about systemd login stuff? | 16:05 |
faenil | I think we've got all pieces now, re sailfish | 16:05 |
faenil | and what systemd expects | 16:06 |
*** flash1 has joined #nemomobile | 16:06 | |
filippz | faenil: basically Qt Quick control used for glacier and jolla homescren don't work OK for landscape displays | 16:07 |
*** flash1 has quit IRC | 16:07 | |
filippz | as for systemd I'll get back to the subject this weekend - I've lost too much time on it already | 16:08 |
faenil | I did the orientation implementation for qqc-nemo (not sure if this is what you're referring to, or it's general qqc issue) | 16:08 |
faenil | it was assuming that lipstick worked as it did before | 16:08 |
faenil | don't know if something changed | 16:08 |
locusf | faenil: its the same age old issue with using Page as root component for Glacier homescreen | 16:09 |
locusf | instead of ApplicationWindow | 16:09 |
filippz | colorfull works OK even now - It's orientated properly and responds to rotation | 16:09 |
locusf | yeah our implentation for rotation is only supported for ApplicationWindow | 16:10 |
filippz | Isn't that jolla uses it's own ApllicationWindow from silica components? | 16:10 |
*** alien_ has quit IRC | 16:10 | |
faenil | locusf: ah ok it's not qqc, it's lipstick | 16:10 |
faenil | (afair) | 16:11 |
locusf | filippz: yes and our implementations differ | 16:11 |
locusf | faenil: yeah lipstick does handle rotation but Page component doesn't respond to it like ApplicationWindow does | 16:12 |
filippz | faenil: wouldn't colorfull also be affected if the lipstick itself is picky about orientation? | 16:12 |
locusf | filippz: nope, colorful home works because it uses PageStack from the old components which has rotation handling | 16:12 |
faenil | there ^ | 16:13 |
filippz | well It's working with those fixes, so we can move past that | 16:14 |
faenil | locusf: what was the issue with using applicationwindow for glacier homescreen? I forgot | 16:14 |
faenil | filippz: it has to be seen if the fixes have any sideeffect on the real glacier orientation implementation | 16:14 |
faenil | for the rest of the cases :D | 16:14 |
locusf | faenil: using applicationwindow created a black screen for the compositor | 16:15 |
filippz | It seems my N9 needs it's battery recharged :( | 16:15 |
locusf | faenil: I propose we clone the rotation handling from applicationwindow | 16:15 |
filippz | locusf: I vote yes | 16:16 |
faenil | locusf: what? it makes no sense :I | 16:17 |
locusf | hold your horses, its apparently not that simple | 16:17 |
faenil | do you really want the orientation code to be in each of your application pages? | 16:17 |
faenil | I do not | 16:17 |
locusf | indeed | 16:17 |
faenil | (beside, I don't remember if there were issues with that, it's been too long) | 16:17 |
locusf | yeah its been waaayyy too long since anyone touched qqcn | 16:19 |
faenil | but anyway, let's fix booting issues first, we'll get to that later, we have a workaround for now | 16:19 |
locusf | yup | 16:19 |
locusf | filippz: so the files you used on sailfish works on nemo as well? | 16:20 |
locusf | filippz: for rotation? | 16:20 |
filippz | locusf: yes | 16:20 |
faenil | if I only think about how long it took to get the drawer orientation change right....god :D | 16:21 |
filippz | I'll transfer them to nemo ks for time being. We can decide wether to put those fixes into fbset-N9.service or try to set them from kernel and drop that service | 16:22 |
*** alien_ has joined #nemomobile | 16:22 | |
*** cloanta_ has joined #nemomobile | 16:22 | |
filippz | QT_QPA_EVDEV_TOUCHSCREEN_PARAMETERS="rotate=90" in 60-n9-n950-ui.conf can't be avoided | 16:23 |
locusf | we should upstream that | 16:24 |
*** tanghus has quit IRC | 16:24 | |
faenil | and the dbus symlink | 16:25 |
filippz | locusf: yes | 16:26 |
filippz | faenil: dbus symlink is a part of systemd mess that needs maybe up to 10 files changed | 16:27 |
*** tanghus has joined #nemomobile | 16:27 | |
faenil | filippz: no problem :p | 16:27 |
faenil | the sooner we get it back up and running, the sooner we can have other people join with easier qml tasks :) | 16:28 |
*** zalan has joined #nemomobile | 16:29 | |
*** mart has joined #nemomobile | 16:29 | |
filippz | In my current setup I can't even kill my N9 - it restarts by itself when shutdown from nemo (even when using 8 sec press) - probably that ACTDEAD thing, or maybe wrong runlevel target for systemd | 16:29 |
*** notmart has quit IRC | 16:30 | |
filippz | Did I mention that I have developed strong affection to systemd in the last few days? | 16:30 |
faenil | haha | 16:31 |
Stskeeps | systemd is great. compared to the other crap that could exist. | 16:31 |
Stskeeps | :P | 16:31 |
faenil | yeah systemd doesn't have issues on its own imho...the problem is the rest of the world wanting to depend on it | 16:32 |
faenil | *that's* the real issue | 16:32 |
filippz | Stkeeps: maybe so - but it doesn't exactly keeps things simple/stupid (on paper maybe - but in reality....) | 16:33 |
faenil | filippz: do you plan to try with agetty in the weekend? | 16:33 |
faenil | filippz: http://forkfedora.org/ | 16:34 |
filippz | faenil: I'm still under imrepssion that there should be some easy fix for our situation - I'll try to find it (that agetty thing looks promising) | 16:34 |
faenil | filippz: yeah well we just need something that opens a pam session | 16:35 |
faenil | agetty should do it | 16:35 |
faenil | and then hope that login binary keeps the pam session open, as I'd expect a login service to do | 16:36 |
faenil | (or not?) | 16:36 |
filippz | faenil: still jolla made that autologind thing - and they are not stupid - there is probably a reason for that approach | 16:37 |
faenil | filippz: the autologind just keeps the autologin service open...because once that service dies (i.e. if the binary would exit) the pam session goes with | 16:38 |
*** m4g0g has joined #nemomobile | 16:38 | |
faenil | and systemd brings everything back into the grave when pam session disappears | 16:38 |
faenil | goes with it* | 16:39 |
faenil | it's a hack to make it so that the pam session is kept open | 16:39 |
filippz | I will try agetty and see wether that is a viable solution - we have it in /sbin already - why not use it | 16:39 |
faenil | otherwise systemd would kill everything of the user-session | 16:39 |
faenil | filippz: sure...I'm just not sure if login binary keeps the pam session open, as I know nothing pam | 16:40 |
faenil | if it does, we have a winner, otherwise, we need the same hack(or an alternative) | 16:40 |
faenil | I wouldn't be surprised if jolla went for that solution because of lack of time | 16:40 |
filippz | faenil: until a couple of days I just knew that systemd is the reason we have to move to kernel 3.0+ - now I know so much more about it | 16:41 |
faenil | know nothing ABOUT* pam...why do I keep missing words? | 16:41 |
faenil | filippz: :D | 16:42 |
filippz | If pam is to be looked at then we'll look at it and be smarter by the day :) | 16:42 |
faenil | I learned a bit about the services when because I had to, when I helped transition nemo from x11 to wayland last summer | 16:42 |
faenil | :) | 16:42 |
faenil | nobody in #systemd could help a lot | 16:42 |
faenil | they just said "use gdm or something like that and their autologin features" | 16:43 |
faenil | (though they were nice and tried to explain what systemd requires) | 16:43 |
filippz | the world still hasn't moved away from X11 - that's why we have so much things to do IMHO | 16:44 |
*** phaeron has quit IRC | 16:44 | |
filippz | well my N9 is on charger, so there's not much I can do. See you all tomorrow o/ | 16:48 |
*** filippz has quit IRC | 16:48 | |
Aard | what getty issues do you have? | 16:49 |
faenil | Aard: we're trying to move to systemd206, the user session stuff changed | 16:50 |
faenil | and it seems we need a pam session to be opened to trigger systemd user session starting mechanism | 16:50 |
Aard | 206/ don't you mean 208? | 16:50 |
faenil | jolla does it with a hack, it starts an exe with PAMName and that exe just keeps running forever so that the PAM session stays on | 16:50 |
faenil | we'd like to look for an alternative | 16:51 |
faenil | yeah 208 | 16:51 |
faenil | so the alternative would be to use agetty to autologin the user, so that systemd starts the user session | 16:51 |
Aard | remind me later or tomorrow, I can see if I find something useful about that in git history | 16:51 |
faenil | but we're not sure if it will work (i.e if login binary will keep the pam session open) | 16:52 |
faenil | Aard: ok, thanks | 16:52 |
faenil | ! | 16:52 |
*** flash1 has joined #nemomobile | 16:54 | |
*** flash1 has quit IRC | 16:55 | |
hedayat | I'd like to ask two (unrelated!) questions | 17:07 |
hedayat | (If you have time to explain) 1. What exactly do you use "systemd --user" for? And currently, how 'nemo' user is 'logged in'? | 17:09 |
hedayat | and 2. AFAIK, Nemo is a layer on top of Mer. Is it possible to run Nemo on top of another core with reasonable effort? | 17:10 |
faenil | 1) systemd --user starts the user session, i.e. it tries to achieve /usr/lib/systemd/user/default.target if I understood correctly. About Nemo login, it doesn't auto login at the moment, so the user session isn't started, that's our issue | 17:18 |
hedayat | I just wonder if the only use of 'systemd --user' in Nemo is to start dbus-session.service, why don't you run dbus-session without using systemd? Like the way you start other things like the glacier/lipstick UI? | 17:18 |
faenil | 2) No idea about the effort, but I think it's considerable because all the middleware is done to work on Mer...I don't know | 17:18 |
faenil | lipstick UI should be started by that systemd --user, if there are no errors | 17:19 |
hedayat | faenil: about 2, do you have lots of patches to upstream packages in Mer? | 17:19 |
faenil | because of the chain rections when systemd tries to achieve default.target | 17:19 |
faenil | hedayat: you mean packages in Mer vs upstream of those packages? | 17:20 |
hedayat | faenil: How do you actually login the nemo user now? | 17:20 |
hedayat | Yes | 17:20 |
faenil | we don't...afaik | 17:20 |
*** mart has quit IRC | 17:21 | |
faenil | I know Qt has got quite a few patches, plus a lot of other stuff, you can check /github.com/mer-packages to see the packages which had to be patched | 17:21 |
hedayat | OK, if there are patches, then it'd be hard. | 17:21 |
faenil | also, why port it to another Core? we're merging Nemo middleware with Mer Core... | 17:22 |
faenil | we can't afford testing a whole new core, jolla is doing the hard job for us | 17:22 |
hedayat | Building a "Fedora Nemo Product" just crossed my mind :D :P. | 17:22 |
hedayat | No no, I was not thinking the you move to another core | 17:22 |
Stskeeps | also, not to mention that there's been substantial effort to optimize towards mobile use/60fps/etc.. | 17:23 |
hedayat | Was just think if OTHERS can bring Nemo to other cores. | 17:23 |
Stskeeps | it's a PITA to do that | 17:23 |
Stskeeps | :P | 17:23 |
faenil | hedayat: I think it would be quite easier to do it from scratch | 17:23 |
faenil | than to port it to another core xD | 17:23 |
*** M4rtinK has joined #nemomobile | 17:24 | |
faenil | hedayat: don't believe Stskeeps' 60fps promises :P | 17:24 |
hedayat | faenil: :)) OK OK. Just was curious about it; as Fedora has started to have a number of "Products"; and as I'm a Fedora fan! :P | 17:25 |
hedayat | faenil: :)) | 17:25 |
faenil | :) | 17:25 |
hedayat | faenil: well, how do you currently actually 'login' the nemo user? | 17:25 |
faenil | currently, nobody does | 17:26 |
* Stskeeps slaps faenil | 17:26 | |
faenil | afaik | 17:26 |
hedayat | ?! so everything is running as root? | 17:27 |
faenil | I don't know, I'm confused about thow things were handled pre-new systemd | 17:27 |
faenil | I never saw any service logging in | 17:28 |
faenil | (or I forgot about it in the meanwhile) | 17:28 |
Stskeeps | no, systemd starts a user session, running as nemo | 17:28 |
Stskeeps | through a bit of user@ magic | 17:28 |
faenil | Stskeeps: that's the new setup | 17:28 |
Stskeeps | yes | 17:28 |
faenil | what about the old one | 17:28 |
faenil | which I guess is what he was asking | 17:29 |
Stskeeps | there was some changes needed for the new setup indeed, to get systemd going with it | 17:29 |
hedayat | Stskeeps: wow, I felt that 'systemd --user' is run AFTER the user is logged in ... somehow | 17:29 |
faenil | hedayat: is IS | 17:29 |
faenil | it* | 17:29 |
Stskeeps | systemd --user is probably getting fork-setuid-exec'ed | 17:29 |
faenil | you login, pam session is opened, systemd detects that and starts user-session | 17:29 |
faenil | with all its services | 17:29 |
faenil | when you close the pam session, systemd detects it, and closes all user-sessio services | 17:29 |
hedayat | and who does that?! apparently that is the process which should be running while user is logged in. | 17:30 |
faenil | who does what? the login? | 17:30 |
hedayat | yes | 17:30 |
hedayat | ! | 17:30 |
faenil | in Nemo, NOTHING (we're deciding how to fix that!) , in Jolla, autologin@.service | 17:31 |
hedayat | So, in Nemo, how systemd --user is being started in the first place?! :P (sorry if I'm missing something) | 17:31 |
faenil | our issues is exactly that...new systemd requires a user to be logged in, so it can start user-session and services | 17:31 |
faenil | hedayat: it is NOT! XD | 17:31 |
faenil | that's our issue! XD | 17:31 |
Stskeeps | start-user-session.. are you sure | 17:32 |
Stskeeps | check user@ and /lib/systemd/system | 17:32 |
faenil | user@ is started after you login | 17:32 |
hedayat | faenil: but I heard that it runs, then terminates, and this is your problem! | 17:32 |
faenil | hedayat: well actually, sometimes it runs, sometimes it doesn't, because guys had a mixed old-new setup | 17:32 |
Stskeeps | faenil: not exactly. | 17:32 |
Stskeeps | :P | 17:32 |
faenil | they still had user-sessio@.service which was the *OLD* setup | 17:33 |
faenil | Stskeeps: really? that's how I got it so far | 17:33 |
hedayat | hmmm OK, let me see what happens on my Fedora. I see 'systemd --user's running, but with no childs. Probably nothing is managed by it yet. | 17:34 |
hedayat | I'm going to kill them to see what happens! :D | 17:34 |
faenil | # graphical.target.wants -> start-user-session.service -> # autologin@.service -> user@.service -> systemd --user | 17:34 |
faenil | Stskeeps: ^ | 17:34 |
faenil | no login, no user@.service ;) | 17:34 |
hedayat | (I might disappear because of it!) | 17:34 |
faenil | lol | 17:35 |
hedayat | In the meantime, has anybody looked into my n9_easy_boot? | 17:35 |
hedayat | :P | 17:35 |
faenil | not me | 17:35 |
locusf | I didn't look at tmo today, sorry :/ | 17:35 |
hedayat | killed one of them, nothing happened. going for the second one... | 17:35 |
faenil | Stskeeps: check start-user-session ;) it only runs autologin bin | 17:35 |
faenil | (close your ears guys, Stskeeps is spreading FUD to make our life impossible! :P ) | 17:36 |
hedayat | locusf: I've not posted it on TMO yet! Thought that you might try the file I linked, and have some feedback. I'm going to post it after killing my systemds :D | 17:36 |
hedayat | locusf: by the way, the package has a README with it which should be enough | 17:37 |
locusf | hedayat: oh okay | 17:37 |
hedayat | OK, nothing happened as expected. So, at least on Fedora 20, systemd --user manages nothing for the desktop. | 17:38 |
hedayat | It's just there to manage user provided services, if any. | 17:38 |
hedayat | gnome-session-manager is still used to manage gnome session. | 17:38 |
faenil | ok, interesting | 17:39 |
hedayat | but Fedora runs systemd-logind, which is probably used for session management | 17:39 |
faenil | yes, *after* the user logs in :) | 17:39 |
hedayat | (actually, systemd --user is not provided for session management, but for 'user services management' as I get it) | 17:40 |
hedayat | faenil: what? logind is started after login? | 17:40 |
faenil | I don't know when it is started, but it handles the login session *after* you login | 17:41 |
faenil | it's not the tool you use *to* login | 17:41 |
faenil | (I used to believe that as well, until today) | 17:41 |
faenil | but it's my first day inside login stuff, so I'm all confused, still trying to put pieces together | 17:42 |
faenil | atm I just want Stskeeps to confess he was wrong, haha | 17:42 |
Stskeeps | neverrr | 17:42 |
Stskeeps | :P | 17:42 |
faenil | :D | 17:42 |
hedayat | :)) | 17:42 |
hedayat | hmmm, you are right: "User sessions are registered in logind via the pam_systemd(8) PAM module." | 17:43 |
hedayat | let me see what pam_systemd is | 17:43 |
faenil | \o/ | 17:43 |
faenil | it should be the module which connects pam to systemd (captain obvious, I know) | 17:43 |
hedayat | :D | 17:44 |
locusf | damn nemo mobile forum is quiet | 17:53 |
locusf | fortunately this channel isn't as quiet | 17:53 |
faenil | which forum | 17:56 |
hedayat | faenil: "https://github.com/joukewitteveen/xlogin/blob/master/xlogin@.service" | 18:03 |
hedayat | faenil: I think you'll need something like this | 18:03 |
hedayat | faenil: and you can just run your main GUI element with it | 18:04 |
hedayat | IIRC, it can be set up in a way that even if it crashed, it is started automatically again by systemd | 18:04 |
hedayat | I'm not familiar with Wayland world, but you'll have a base graphical process in that setup too, right? (something act like X). It is called "compositor" or something I guess | 18:05 |
locusf | faenil: tmo | 18:06 |
locusf | now I've got here maps for android and now downloading belgium maps for FOSDEM preparation | 18:08 |
faenil | hedayat: yes well the compositor is lipstick binary | 18:08 |
faenil | and that should restart already if it crashes | 18:08 |
hedayat | Using something like xlogin, setting up a PAM session is actually done by systemd, and the long running process is that compositor. There should be no need for something like autologind | 18:09 |
hedayat | PAMName: http://www.freedesktop.org/software/systemd/man/systemd.exec.html | 18:09 |
faenil | hedayat: have you read the log from this afternoon? :p | 18:10 |
hedayat | faenil: Yes I did. | 18:10 |
hedayat | :) | 18:10 |
faenil | hedayat: ok, I pasted PAMName definition and the fact that syustemd starts the pam session :D | 18:11 |
hedayat | :) | 18:11 |
faenil | but lipstick is one of the services that are part of the user session startu | 18:12 |
hedayat | Since you only have a single user, nemo, you'll need a service without @, and "User=" entry should be User=nemo | 18:12 |
faenil | I don't know if it can be started as early as at the beginning of the user session | 18:12 |
hedayat | faenil: That's OK, this service can run an script, just like as what the xlogin service does | 18:13 |
faenil | https://github.com/nemomobile-ux/glacier-home/blob/master/rpm/lipstick.service | 18:13 |
hedayat | faenil: All it needs is that the script never finishes | 18:13 |
faenil | no I meant, maybe it has to follow systemd ordering | 18:13 |
faenil | but it seems it's not needed as it doesn't have big requirements | 18:13 |
hedayat | faenil: hmmmm.... well, maybe we can handle it somehow. | 18:14 |
hedayat | faenil: I guess none of its requirements are "user" srevices, am I right? | 18:16 |
faenil | hedayat: mmm so adding PAMName and User to lipstick.service could be enough | 18:16 |
faenil | maybe | 18:16 |
faenil | hedayat: yes that's what I meant with the previous msg :) | 18:16 |
hedayat | faenil: apparently it doesn't depend on anything in user session service, and I feel (hope!) that pre-user-session.target is actually a "PRE" session target | 18:16 |
hedayat | faenil: Yes, I think so! | 18:17 |
faenil | I think so :P | 18:17 |
faenil | mmm | 18:18 |
faenil | I'm not sure what would be of the bins here in that case | 18:19 |
faenil | https://github.com/nemomobile/nemo-mobile-session/tree/master/bin | 18:19 |
faenil | (killx is not needed anymore anyway) | 18:19 |
faenil | maybe that script is needed to support actdead | 18:20 |
faenil | (most likely) | 18:21 |
hedayat | sorry, I'll be back in about 10 minutes. BTW, PAMName= might depend on setting up pam_systemd which was mentioned earlier | 18:21 |
faenil | not sure it actually needs setting u | 18:21 |
faenil | p | 18:21 |
kimmoli | lbt: hmm... where i can get the wh script? | 18:23 |
faenil | this is strange though | 18:26 |
faenil | why another PAM session? | 18:26 |
faenil | /lib/systemd/system/user@.service:PAMName=systemd-user | 18:26 |
faenil | hedayat: btw, what happens if we call agetty in the login.service instead of lipstick directly? (so we don't have to changed the ordering) | 18:29 |
faenil | does agetty exit after logging in? (and what about the login process that agetty uses to login?) | 18:30 |
kimmoli | is it this? https://github.com/mer-tools/sdk-setup/blob/master/src/wh | 18:30 |
faenil | kimmoli: looks like it ? :) | 18:31 |
*** gabriel9 has quit IRC | 18:31 | |
*** gabriel9 has joined #nemomobile | 18:31 | |
*** phdeswer has quit IRC | 18:36 | |
hedayat | faenil: I'm back. sorry. | 18:36 |
kimmoli | faenil: yep.. | 18:36 |
*** phaeron has joined #nemomobile | 18:37 | |
faenil | hedayat: np ;) | 18:37 |
*** plfiorini has joined #nemomobile | 18:38 | |
hedayat | well, I didn't expect system-user to setup a session by itself. There is a problem: "systemd --user: This process will survive as long as there is some session for that user" | 18:39 |
hedayat | if it creates a session itself, so there is a session for the user while it is running! | 18:39 |
faenil | dinner, bbl | 18:40 |
kimmoli | pöh: the apiurl 'https://api.opensuse.org' does not exist in the config file. | 18:46 |
kimmoli | ..and it works \o/ | 18:49 |
kimmoli | .oscrc had apiurl to that opensuse... | 18:50 |
*** phdeswer has joined #nemomobile | 18:51 | |
*** gabriel9 has quit IRC | 19:07 | |
ZogG | o/ | 19:11 |
kimmoli | ...or not working. :( | 19:12 |
faenil | back | 19:14 |
*** LjL has quit IRC | 19:17 | |
*** cloanta_ has quit IRC | 19:17 | |
*** cloanta_ has joined #nemomobile | 19:17 | |
hedayat | faenil: I feel that user@.service is already doing what you expect: keeping 'systemd --user' running. because it opens a session for the user itself. | 19:19 |
hedayat | faenil: I decided to run this service manually, and see if it terminates. It's still alive | 19:20 |
faenil | hedayat: the problem is that that session isn't opened if there is no other session that triggers the start of user@.service | 19:20 |
hedayat | faenil: but you said that graphical target wants it? | 19:20 |
*** LjL has joined #nemomobile | 19:21 | |
hedayat | faenil: so... if you want to reach this target you should start user@.service, not? | 19:21 |
faenil | checking | 19:21 |
Stskeeps | user@UID | 19:21 |
hedayat | user@nemo works too | 19:21 |
faenil | it's not in graphical target wants | 19:22 |
faenil | as far as I can see | 19:22 |
faenil | or is it | 19:22 |
* faenil doublechecks | 19:22 | |
faenil | no | 19:22 |
hedayat | faenil: :P | 19:23 |
faenil | there's start-user-session, which triggers autologin, which then triggers user@service | 19:23 |
faenil | so, NO, user@service is not triggered by graphical target | 19:23 |
faenil | (not without the autologin procedure, I mean) | 19:24 |
hedayat | faenil: What if autologin is eliminated from the chain? | 19:24 |
faenil | then user@.service should never be started | 19:24 |
faenil | (if I understood correctly how the whole thing works) | 19:24 |
faenil | would* | 19:24 |
hedayat | No! I mean what if start-user-session triggers user@nemo.service directly | 19:24 |
lbt | kimmoli: you got it | 19:25 |
kimmoli | lbt: but not working :( | 19:25 |
faenil | don't know...I guess there's a reason why it has to be triggered via pam session :/ not sure though | 19:25 |
lbt | kimmoli: does 'osc ls' work | 19:25 |
lbt | the wh script should check to ensure osc is properly configured | 19:26 |
kimmoli | it created the webhook ok | 19:26 |
hedayat | faenil: but it opens a pam session by itself | 19:26 |
hedayat | ! | 19:26 |
faenil | hedayat: ah right, I forgot about that already :p | 19:26 |
faenil | then who knows :) | 19:26 |
kimmoli | lbt: but pushing to devel branch doesnt msg on mer-boss | 19:27 |
faenil | let's see if they know something in #systemd | 19:27 |
lbt | kimmoli: which pkg ? | 19:27 |
kimmoli | home:kimmoli:devel > toholed-daemon | 19:27 |
kimmoli | i did already copy-paste _service to have place to manually paste sha | 19:28 |
lbt | right - but there's no wehook service in there | 19:28 |
kimmoli | ? | 19:29 |
lbt | https://wiki.merproject.org/wiki/Packaging/webhooks | 19:29 |
lbt | specifically https://wiki.merproject.org/wiki/Packaging/webhooks#Web_Interface | 19:29 |
lbt | edit this: https://build.merproject.org/package/view_file/home:kimmoli:tohs/toholed-daemon/_service?expand=1 | 19:29 |
kimmoli | yes. it was like that after wh -script, but didn't work | 19:29 |
lbt | looks OK to me: https://webhook.merproject.org/webhook/api/webhookmappings/mobs/home:kimmoli:devel/toholed-daemon/find | 19:31 |
kimmoli | i edited _service, now "broken" | 19:32 |
lbt | mmm | 19:34 |
kimmoli | now i pushed commit to bb | 19:34 |
lbt | I'm looking at OBS and last change to _service was 2 days ago | 19:35 |
lbt | and it looks the same | 19:35 |
lbt | also i it says "broken" then it's not likely to work :) | 19:35 |
lbt | if | 19:35 |
kimmoli | home:kimmoli:devel/toholed-daemon | 19:35 |
lbt | oh, my bad - I must have had :tohs in my browser history | 19:36 |
kimmoli | home:kimmoli:tohs/toholed-daemon is 2 days ago https://build.merproject.org/package/show/home:kimmoli:tohs/toholed-daemon | 19:36 |
kimmoli | the _service is now according to wiki | 19:37 |
*** piggz has joined #nemomobile | 19:37 | |
lbt | let me go check boss is all good | 19:37 |
*** DrCode has quit IRC | 19:41 | |
faenil | hedayat: nothing, nobody has replied in #systemd so far | 19:42 |
*** gogeta has joined #nemomobile | 19:42 | |
hedayat | faenil: probably testing is the easiest solution! | 19:44 |
faenil | hedayat: not sure we know enough to even say if it's working or not...it could seem to be working, at first glance... | 19:45 |
hedayat | faenil: BTW, at least it works in Fedora. I just start user@username.service, and it runs | 19:45 |
faenil | sure | 19:45 |
hedayat | and it doesn't terminate | 19:45 |
faenil | I don't see why the service wouldn't run...I'm just not sure of the sideeffects for the rest of the user session | 19:45 |
hedayat | I'm still waiting to see if it will terminate after some time | 19:45 |
hedayat | hmmm... unless autologind actually does other things too (rather than just sleeping?!) | 19:47 |
lbt | kimmoli: push again - I wonder if some commit events were missed | 19:47 |
kimmoli | ok | 19:47 |
faenil | hedayat: we were told it just sleeps, we have no reason to believe otherwise ;) | 19:47 |
kimmoli | pushed | 19:48 |
kimmoli | dodi | 19:48 |
hedayat | faenil: :) | 19:48 |
*** DrCode has joined #nemomobile | 19:49 | |
hedayat | faenil: I fail to see why autologind is needed. It might be needed for the days of user-session@.service! but with user@.service, it really seems to be useless. | 19:49 |
hedayat | IF I've understood things correctly! | 19:49 |
faenil | hedayat: I agree, IF we understood correctly ^^ | 19:50 |
*** gabriel9 has joined #nemomobile | 19:54 | |
faenil | hedayat: I feel much more positive now :) | 19:55 |
hedayat | faenil: :) | 19:56 |
*** Guhl has joined #nemomobile | 20:00 | |
*** juergbi has quit IRC | 20:04 | |
faenil | hedayat: feel like testing it ? :) | 20:06 |
hedayat | Yes, but I've not setup Mer SDK yet. | 20:06 |
*** juergbi has joined #nemomobile | 20:07 | |
faenil | ouch | 20:07 |
hedayat | :P | 20:08 |
faenil | I think locusf uploaded an image already? | 20:08 |
hedayat | If there's an image, I'd gladly download and test it. And would be able to test such things. | 20:09 |
*** gabriel9 has quit IRC | 20:09 | |
hedayat | otherwise, I'll setup Mer SDK when I found some free time again. | 20:09 |
*** gabriel9 has joined #nemomobile | 20:10 | |
faenil | hedayat: http://images.devaamo.fi/sfa/n9/ | 20:10 |
faenil | it's from 4 days ago, it should have new packages already | 20:10 |
faenil | don't worry if graphics is broken (in case you get it to boot ;) ) | 20:10 |
hedayat | faenil: great, thanks. I'll download it tonight. | 20:13 |
hedayat | :) | 20:13 |
faenil | I think the needed changes are: align user@.service to the one from SailfishOS, create a symlink to dbus.service /usr/lib/systemd/user/pre-user-session.target.wants/dbus.service | 20:13 |
faenil | delete user-session@service | 20:14 |
hedayat | OK | 20:14 |
faenil | add something in graphical-target-wants that runs user@.service with the right id | 20:14 |
*** gogeta has quit IRC | 20:14 | |
faenil | not sure if there's any other change which is needed that the other guys know | 20:15 |
*** jonwil has joined #nemomobile | 20:16 | |
hedayat | user@.service can be just user-nemo.service, and nemo being hardcoded as the user | 20:16 |
locusf | the image should get you running lipstick | 20:16 |
faenil | get you running lipstick via user-session@service, I hope you mean | 20:17 |
faenil | not via user@service | 20:17 |
locusf | yes | 20:17 |
faenil | so it's important to delete user-session :D | 20:17 |
hedayat | locusf: so it doesn't have the new stuff? I already have nemo-n950-rnd-20140224.tar.bz2. Is it much different? | 20:17 |
faenil | yes it's A LOT different | 20:18 |
locusf | hedayat: a lot | 20:18 |
hedayat | :D OK! | 20:18 |
faenil | 8 months worth of upgrades different, precisely :P | 20:18 |
faenil | and now I wonder... | 20:19 |
faenil | if we want to start the .service manually | 20:19 |
hedayat | :P I thought that maybe the updates were pushed into the repositories recently. | 20:19 |
locusf | I should be sleeping :p | 20:19 |
faenil | then how is that different from keeping user-session@service? | 20:19 |
faenil | hedayat: not sure if you can update to latest core from your image, could be, don't know | 20:20 |
faenil | but now I'm puzzled :/ | 20:21 |
faenil | why are we transitioning from user-session if we're creating something which is very similar | 20:21 |
locusf | not lilely to produce anything viable via zypper dup with newmer core | 20:21 |
hedayat | faenil: good point. I wonder why exactly start-user-session script is needed. BTW, what is ACT_DEAD mode?! :P :D | 20:22 |
faenil | hedayat: I think it's needed because of actdead... | 20:22 |
hedayat | faenil: maybe to be compatible with newer systemd anyway?! | 20:22 |
faenil | actdead is the mode which is used when you use charger and the phone is off | 20:22 |
faenil | (iirc) | 20:22 |
*** r0kk3rz has quit IRC | 20:23 | |
faenil | hedayat: to be compatible with the new systemd you have to start a pam session, not trick the system like we're trying to do :D | 20:23 |
*** r0kk3rz has joined #nemomobile | 20:23 | |
faenil | we're basically tricking it by going back to the obsolete mode which we're getting rid of :/ | 20:24 |
hedayat | faenil: Probably yes and no. PAM session should be created by the login manager. And if you are using user@.service as the login manager, that's fine | 20:25 |
faenil | not sure...that PAMName is in the upstream systemd user@service | 20:25 |
hedayat | IMHO, user@.service is actually a session manager since it opens a new session for the user. However, 'systemd --user' is a user service manager, not dependent on any given session. | 20:26 |
hedayat | faenil: at least, it exists in Fedora version too, while Fedora doesn't use it. | 20:26 |
hedayat | (so it is probably upstream) | 20:28 |
*** mhall119 has quit IRC | 20:35 | |
*** mhall119 has joined #nemomobile | 20:35 | |
hedayat | Well, apparently user@.service is usually started by pam_systemd if not already started. Which is a little confusing :P | 20:40 |
hedayat | http://www.freedesktop.org/software/systemd/man/pam_systemd.html, part 3 | 20:41 |
faenil | hedayat: alright, so just start user@id yourself :) | 20:43 |
faenil | instead of hardcoding the id | 20:43 |
*** jonwil has quit IRC | 20:44 | |
*** jonwil has joined #nemomobile | 20:45 | |
faenil | see what start-user-session does, to get the id from login defs ;) it's a simple cmd | 20:45 |
faenil | so that we also keep the same behaviours | 20:46 |
hedayat | yes, I know. But I wonder what are the benefits of a generic code in a system which always have a single user | 20:46 |
hedayat | or not?! | 20:46 |
hedayat | Does Sailfish/Nemo have other users? | 20:46 |
faenil | hedayat: don't think so...not sure if other users are used for some purposes though | 20:47 |
faenil | but, you know, better keep it generic if it is cheap | 20:47 |
*** zalan has quit IRC | 20:47 | |
*** dr_jolla has joined #nemomobile | 20:47 | |
*** zalan has joined #nemomobile | 20:48 | |
hedayat | faenil: hmm OK. But then, I'd probably have a config file somewhere rather than reading the minumum UID! anyway, not important | 20:49 |
hedayat | faenil: Ah! I feel that I understand it correctly now! :D | 20:49 |
faenil | :) | 20:49 |
hedayat | faenil: user@.service opens a new session, so its 'systemd --user' never terminates | 20:49 |
hedayat | pam_systemd starts user@.service when a user is logging in, if it is not running. | 20:50 |
faenil | yes, ok ;) | 20:50 |
hedayat | systemd --user is expected to be running while there are some open user sessions, so it will run forever while user@.service is running, since it has a dedicated session | 20:50 |
*** Oksana has joined #nemomobile | 20:51 | |
hedayat | And pam_systemd terminates user@.service when the last user session (except the one created by user@.service) is terminated | 20:51 |
faenil | yeah, no news there :) | 20:51 |
hedayat | So, this is why user@.service open a separate session when a new session is already created | 20:52 |
hedayat | :D | 20:52 |
faenil | wait, except the one created by user service? then how do you stop it? | 20:52 |
faenil | so we do need another session? | 20:52 |
hedayat | Well, actually I was confused that why user@.service should create a new session, if it is started by pam_session, which is called when a new session is already created! | 20:52 |
hedayat | pam_session does | 20:53 |
hedayat | 'session' PAM services are called both when a new user session is created, and when a session is terminated | 20:54 |
*** krnlyng has quit IRC | 20:54 | |
faenil | yes but if there's no other session except the one created by user@service.. | 20:54 |
hedayat | So, pam_session can actually keep track of real user sessions | 20:54 |
dr_jolla | good evening | 20:54 |
faenil | dr_jolla: o/ | 20:54 |
*** LjL has quit IRC | 20:54 | |
hedayat | No, pam_session is called on each session creation/termination, not just when the last session terminates | 20:55 |
*** lbt is now known as lbt_censored | 20:55 | |
faenil | yes, but... | 20:55 |
*** lbt_censored is now known as lbt | 20:55 | |
dr_jolla | i saw many hysteria against Saìlors , something happened? | 20:55 |
faenil | you want to start user@service without opening a new session *before* right? | 20:55 |
hedayat | when the last user session (except the one created by user@.service) is closed, pam_session stops user@.service | 20:55 |
faenil | dr_jolla: what happened? | 20:56 |
faenil | hedayat: but there is no session beside the one created by user@service the way you wanted to do it | 20:56 |
dr_jolla | i got the twitter timeline full of hate | 20:56 |
dr_jolla | btw i see you are mangling with systemd user session | 20:56 |
faenil | yea | 20:57 |
hedayat | faenil: I just described what happens in regular systems, which can affirm that our findings about user@.service is correct | 20:57 |
dr_jolla | what are you doing? | 20:57 |
faenil | ah, ok | 20:57 |
faenil | then, my question remains, how do we stop it if we have no other session beside the one created by user@? we stop it manually? | 20:58 |
hedayat | So, I'm more confident that starting user@.service from start-user-session should be enough | 20:58 |
hedayat | faenil: Yes, you can stop it when you want to shut down the device | 20:58 |
*** LjL has joined #nemomobile | 20:58 | |
hedayat | because there is no 'logout' in a phone | 20:59 |
*** zalan_ has joined #nemomobile | 20:59 | |
faenil | hedayat: ok so stop it manually | 20:59 |
dr_jolla | faenil: question is , will nemo become multiuser ala android? | 21:00 |
faenil | who knows | 21:00 |
faenil | we'll think about it when the time comes? | 21:00 |
dr_jolla | not now | 21:00 |
hedayat | shutting down is a manual process anyway! Even on the desktop, user logs out manually. | 21:00 |
faenil | I don't think it will become multiuser without jolla doing the real job, so... | 21:00 |
faenil | hedayat: sure | 21:01 |
dr_jolla | faenil: problem are many framewőrk use uid 1000 by default | 21:01 |
dr_jolla | saw on sfa phone | 21:01 |
dr_jolla | but for now i'm involved into a brewing system project | 21:02 |
faenil | dr_jolla: nemo is 10000 | 21:02 |
dr_jolla | lost 0 | 21:02 |
*** zalan has quit IRC | 21:02 | |
hedayat | ok, good night. Hopefully, I'll come with a new Nemo image tomorrow! | 21:02 |
dr_jolla | any news here? | 21:03 |
dr_jolla | i was totally out | 21:03 |
faenil | hedayat: wait! you're leaving without telling me if it works? :D | 21:03 |
faenil | dr_jolla: not much, still working on having a stable system boot | 21:03 |
dr_jolla | ahaha | 21:03 |
* ZogG tickles faenil | 21:03 | |
faenil | gaaaaaaaaaaaaah | 21:04 |
faenil | he left? come oooon :D | 21:04 |
hedayat | faenil: :D I'll tell you tomorrow. It takes some time for me to download the image | 21:04 |
faenil | hedayat: ooooh, I see...okay ;) | 21:04 |
faenil | if you get it working don't forget to send PullRequest to https://github.com/nemomobile/nemo-mobile-session | 21:04 |
faenil | so that we can update the package and go on with the fixes ;) | 21:05 |
hedayat | :) sure | 21:05 |
ZogG | faenil, rpi images? | 21:05 |
faenil | ZogG: ? | 21:05 |
faenil | don't know, I've never worked on an rpi | 21:06 |
faenil | but vgrade used to boot nemo on it iirc | 21:06 |
ZogG | :\ | 21:07 |
ZogG | faenil, the image itself is optimized for specific HW right? | 21:07 |
*** hedayat has left #nemomobile | 21:07 | |
dr_jolla | armv6hf i think | 21:07 |
faenil | armv7hl usually | 21:07 |
*** vakkov has quit IRC | 21:07 | |
dr_jolla | rpi is arm6hf sadly | 21:08 |
dr_jolla | ZogG: another crappy platform | 21:08 |
faenil | if Nemo used to run on rpi I don't see why it can't be done again :) | 21:08 |
dr_jolla | which qt are you using? | 21:09 |
dr_jolla | which display server | 21:09 |
dr_jolla | ? | 21:09 |
faenil | who | 21:09 |
dr_jolla | wayland works better then X | 21:09 |
faenil | are you talking about Nemo? | 21:10 |
faenil | it moved to Wayland more than a year ago | 21:10 |
faenil | and the current image is using qt5.2, like the next Jolla update (the one we're waiting) | 21:10 |
dr_jolla | i knoe | 21:10 |
* dr_jolla want rpi image asap | 21:11 | |
ZogG | faenil, most of my experience with linux except just installing from bootable media is setting gentoo from stage3. so while there i know how i set everything and as it's sourcebased i just compile things with flags i need for platform i need automagically | 21:11 |
*** LjL has quit IRC | 21:11 | |
ZogG | \with adoptating nemo to other platfomr i have no idea how it's done | 21:11 |
faenil | ZogG: you just wait for OBS to build the packages for you arch :P | 21:12 |
*** gabriel9 has quit IRC | 21:12 | |
faenil | and fix a package if you see OBS fails to build it | 21:12 |
ZogG | dr_jolla, if you get one, ping me :P | 21:12 |
dr_jolla | ok | 21:12 |
ZogG | if i'm not answering, pass it to faenil, he is like my secretary :P | 21:12 |
faenil | lol | 21:12 |
faenil | https://build.merproject.org/project/monitor/mer-core:armv6l:devel?blocked=0&building=0&dispatching=0&finished=0&scheduled=0&signing=0&succeeded=0 | 21:13 |
ZogG | faenil, pakage is not the problem, but how do i boot to system to the point i build packages :P | 21:13 |
faenil | that's the current situation with armv6 | 21:13 |
ZogG | faenil, with gentoo i just chroot | 21:13 |
faenil | ZogG: you build the image *after* you know OBS has built all the needed packages | 21:13 |
dr_jolla | with or without hardfloat? | 21:13 |
faenil | so first you get OBS to build the packages you need, *then* you build the image and you flash it :) | 21:13 |
faenil | ZogG: you probably have more experience than me, I only used chroot for mer sdk :P | 21:14 |
dr_jolla | is completely broken | 21:14 |
dr_jolla | good luck | 21:14 |
faenil | what's broken? | 21:15 |
dr_jolla | armv6 | 21:15 |
faenil | there's only 5-6 packages which fail to build, the rest is dependencies missing because of those 5-6 packages | 21:15 |
*** LjL has joined #nemomobile | 21:15 | |
dr_jolla | cross-arm* | 21:15 |
dr_jolla | chain reaction | 21:15 |
dr_jolla | whole image is broken | 21:16 |
*** olesalscheider has quit IRC | 21:16 | |
faenil | I don't know how to fix those few packages | 21:17 |
faenil | but I'm sure somebody in this channel knows | 21:17 |
faenil | but first we have to get nemo to work on armv7 at least | 21:17 |
faenil | and finish Nemo-Mer merge | 21:17 |
faenil | after that's all done maybe someone can think about fixing those failures | 21:17 |
faenil | someone like you, guys :p | 21:18 |
ZogG | i may try to fix | 21:21 |
*** olesalscheider has joined #nemomobile | 21:21 | |
faenil | ZogG: great :) | 21:23 |
Oksana | About armv7: does anybody know why xulrunner-qt5 still fails to build? https://build.merproject.org/package/live_build_log/mer-core:armv7hl:devel/xulrunner-qt5/Core_armv7hl/armv8el | 21:29 |
faenil | yes, I wrote the reason a week ago or two | 21:29 |
faenil | basically, the fixes I sent make it build on embedlite33 branch | 21:30 |
faenil | but nemo will be stuck on embedlite31 until the next LTS is out, it seems | 21:30 |
faenil | so, either you cherrypick all the rest of the fixes that make it compile on gcc4.8 and that mozilla committed between version 31 and 33 | 21:31 |
faenil | or, you move to version 33, and the fixes I sent will be enough to compile successfully | 21:31 |
Oksana | Why not move to version 33? | 21:32 |
dr_jolla | is last stable | 21:32 |
faenil | I don't know the specific reasons, but the embedlite maintainer for nemo said so | 21:32 |
faenil | that we'll only move to the next LTS version | 21:33 |
Oksana | LTS=ESR ? | 21:33 |
ZogG | faenil, but i'll need your help as always to understand how do i fix it in obs? i mean i fork repo and do my own build or how i push changes? | 21:33 |
faenil | Oksana: long term support | 21:34 |
faenil | ZogG: you look at the _service file, there is a git repo in there | 21:34 |
*** plfiorini has quit IRC | 21:34 | |
Oksana | Yes, yes, LTR=ESR, it is easier to renew xulrunner on Nemo every 7 versions, instead of jumping at every new version. | 21:34 |
Oksana | LTS=ESR | 21:34 |
faenil | you fork the repo, clone it on your pc, try to build it in your mer sdk using "mb2 -t <target_name> build" | 21:35 |
faenil | and when you manage to build it, you send a pull request to the official repo, with the fixes | 21:35 |
*** m4g0g has quit IRC | 21:36 | |
Oksana | Is it possible to fork it in github, make some changes, 'show' the fork to OBS, and if OBS compiles successfully, push the fork back to main repository? | 21:36 |
faenil | ZogG: you need a little bit special mer sdk and target, due to the Nemo-Mer merge being a work in progress | 21:37 |
faenil | the instructions to set it up are here http://www.merproject.org/logs/%23nemomobile/%23nemomobile.2014-10-01.log.html#t2014-10-01T15:06:34 | 21:37 |
faenil | and also the morning after | 21:37 |
faenil | ZogG: btw you can already see the build log on OBS, if you didn't notice | 21:39 |
faenil | so that you can have an idea if you know how to fix it or not | 21:39 |
faenil | but you have to first go to the package page | 21:39 |
faenil | and then click "failed" on the right | 21:39 |
dr_jolla | guys good night | 21:40 |
faenil | if you click "failed" from the packages overview which I linked, it won't work | 21:40 |
faenil | (bug) | 21:40 |
dr_jolla | see you soon | 21:40 |
faenil | dr_jolla: o/ | 21:40 |
Oksana | Why do different architectures have different errors? https://build.merproject.org/package/live_build_log/mer-core:i486:devel/xulrunner-qt5/Core_i486/i586 | 21:41 |
faenil | Oksana: the reason is in your msg :D | 21:42 |
*** dr_jolla has quit IRC | 21:42 | |
faenil | I'm going to bed btw, so tired... | 21:42 |
faenil | cya o/ | 21:42 |
Oksana | i486 as a whole has too many failed. Why? | 21:43 |
Oksana | https://build.merproject.org/project/monitor/mer-core:i486:devel?arch_i586=1&defaults=0&failed=1&repo_Core_i486=1 | 21:43 |
Oksana | Good night! | 21:43 |
faenil | Oksana: I think something bad happened | 21:44 |
faenil | all installation errors | 21:44 |
Morpog_PC | ph dlna packages | 21:44 |
Morpog_PC | oh | 21:44 |
faenil | Oksana: there should be only xulrunner not building on i486 | 21:44 |
faenil | Oksana: thanks for noticing, I'll try to ping someone in the morning | 21:45 |
Oksana | Thank you! Good night. | 21:45 |
Oksana | Something in MAKEDEV ? While armv7hl ( :) ) has sb2-init before initializing rpm db..., i486 ( :( ) does not do anything between init_buildsystem and initializing rpm db... | 21:51 |
*** Morpog_PC has quit IRC | 21:52 | |
*** gabriel9 has joined #nemomobile | 22:01 | |
*** Pat_o has quit IRC | 22:04 | |
*** Guhl has quit IRC | 22:11 | |
ZogG | faenil, i read the above thn you wrotw i'll analyze it at weekend and ask you more | 22:27 |
*** gabriel9 has quit IRC | 22:28 | |
*** cxl000 has quit IRC | 22:29 | |
*** gabriel9 has joined #nemomobile | 22:33 | |
*** faenil has quit IRC | 22:39 | |
*** vakkov has joined #nemomobile | 22:39 | |
*** gabriel9 has quit IRC | 22:40 | |
*** r0kk3rz has quit IRC | 22:42 | |
*** M4rtinK has quit IRC | 22:47 | |
*** krnlyng has joined #nemomobile | 23:02 | |
*** Venemo has quit IRC | 23:04 | |
*** vakkov has quit IRC | 23:59 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!