Monday, 2020-05-04

Oksana_r0kk3rz: After installing Sailfish OS anew, I didn't run any zypper update. I didn't have to, OS automagically did it itself. Camera stopped working soon, then after reboot cellular network went down as well. Glad that it is still capable of booting, at least. But would like cellular network back! And camera, too.05:48
Oksana_Sooo, should I somehow try to recover Sailfish on Fxtec Pro1 to working state, or should I try to switch to Sailfish in hopes it will work better?05:49
r0kk3rzit wont automagically update things05:50
Oksana_zyp in --oldpackage doesn't work, Package not found for both nemo-gstreamer1.0-interfaces and the other thing.05:50
r0kk3rzyeah you cant really go backwards05:50
Oksana_Well, the GUI might have decided to automagically check for Sailfish OS updates. Who knows this Jolla thing...05:51
r0kk3rzno, the gui will do no such thing05:51
Oksana_Storeman doesn't run zypper update on non-OpenRepos packages, right?05:51
r0kk3rznot sure, i should hope not05:52
Oksana_Doesn't Sailfish OS during installation check for operating system updates?05:52
Oksana_What does Sailfish OS updates thingie in Settings do, and based on what triggers?05:52
r0kk3rzthe settings thing is for official devices05:53
r0kk3rzwhich does nothing on ported devices05:53
Oksana_Because I definitely didn't run zypper update. I didn't even run zypper refresh.05:53
Oksana_zypper --verify and version --dup, these two I have run. They wouldn't have run update, right?05:54
r0kk3rzdid you deliberately update a version?05:55
Oksana_These two have given errors - zypper has said something about 'product', version has said something else.05:55
Oksana_Sailfish is And I don't know what version --dup does, besides saying nothing provides needed by geoclue-provider-hybris-binder-0.2.28-1.4.5.jolla.armv7hl05:56
r0kk3rzso you flashed a 3.2.1 image?05:57
Oksana_zypper --verify gives product:# This file is copied as hw-release-0.0.1+master.20200502113015.be0be38.2.analogous to os-release is not installableSolution 1: deinstallation of product:# This file is copied as hw-release-0.0.1+master.20200502113015.be0be38.2.analogous to os-release05:57
Oksana_Yes, I flashed
r0kk3rzok good05:58
r0kk3rzwhy did you do version --dup then?05:58
Oksana_I don't remember the particular order. It is quite possible that I have only run it after problems started. But, difficult to be certain.06:01
r0kk3rzi suspect its the cause06:02
r0kk3rzclearly theres packages in the OBS repo, that if installed will break things06:02
r0kk3rzand you will need to wait until 3.3.0 is ready06:03
r0kk3rzversion --dup will probably install them06:03
Oksana_So, what does Sailfish OS updates thing in Settings do? Because I find it more believable that I clicked the GUI thing than that I started running random things in command line while everything was still working.06:04
r0kk3rzfor ported devices, nothing06:05
r0kk3rzand you should never run random things in command line :P06:06
* Oksana_ is still sure that every time I run version --dup manually in command line, it backed out after 9 retried due to the error and didn't actually do anything - but alright, cannot undo whatever was done, let's try updating to, since there is no way to downgrade...06:12
r0kk3rzprobably wait before doint that06:12
Oksana_Hmm, wait for what?06:12
Oksana_Ah, more polished 3.3?06:13
r0kk3rzwait for the porters to say its ready06:13
Oksana_Where are the updates? I have heard of Telegram, but, I don't have a Telegram account.06:15
Oksana_Somebody is already using on Fxtec Pro1, apparently, but, they do seem to have a "black camera" problem...06:18
r0kk3rzyeah, evidently there is something that needs fixing06:19
r0kk3rzupdated droidmedia maybe06:20
T42<birdzhang> use latest droidmedia and gst-droid can fix06:24
r0kk3rz@birdzhang did you fix your version package?06:31
Oksana_Are there any problems visible in the rpm -qa output attached to the forum post ? If there are old versions available somewhere to downgrade to, I could install them manually.06:31
T42<birdzhang> yeah, i use @elros34 's function06:31
r0kk3rzhow do i do that?06:34
piggz@elros34 yeah, it like the other tag name06:44
piggzOksana_: camera stopped working becuase you are on devel: repo and some packages got updated in the common repos06:44
piggzif you update to 3.3 in the devel: repo you should be ok, but you should also switch to the testing: repo for users which has been announced a few time06:45
Oksana_So, common repos have newer packages than devel: repo, and the update breaks things because they are not from devel: repo anymore.06:45
piggzwell, the devel: common repo expects to always be on the latest version, and for the pr1 we hadnt yet updated the droidmedia and gst-droid packages to match06:47
piggzif you are on 3.2 still, you should switch your repos to testing:06:48
Oksana_piggz: Would you mind pointing me to how switch to testing: repo? I know that ssu re (is it the correct version?) can change version from, with version --dup to follow afterwards, but not much besides that...06:48
piggzare you in the sfos-users-telegram group?06:48
piggzsorry, sfos-pro1-users06:49
Oksana_No, I don't have a Telegram account06:49
piggzok, i think we also posted in the forums, but... basically06:49
r0kk3rzOksana_: you would ssu dr the devel repo, and ssu ar the testing repo06:49
piggzedit /usr/share/ssu/features.d/adaptation-community.ini06:50
piggzAdam Pigg, [01.05.20 22:13]06:50
piggzAt the moment, the file will likely be pointing to the devel: repository
piggzAdam Pigg, [01.05.20 22:15]06:50
piggzThis should be changed to:
piggzAdam Pigg, [01.05.20 22:16]06:50
piggzThen issue the command:06:50
piggzssu re && ssu ur && zypper ref && zypper dup06:50
piggzOksana_: and, id quite like to know if everything in the testing repo works ....06:53
piggzyou probably also need to edit the adaptation-community-common repo file too06:53
piggzto match06:53
piggzmal: you approved this in november not merged yet07:30
piggzmal: also, could go in common or common-native?07:31
r0kk3rzyou cant merge it?07:31
piggzi dunno, it =s mer-hybris, do i have permission there07:31
piggzno i cant07:32
piggzr0kk3rz: porters-ci question for ya .....07:38
piggzwe added the repo and package name for harbour-storeman to our pinephone package, along with webcat07:38
piggzwebcat gets installed, but there is no mention of storeman in the logs at all07:38
piggzit just seems to be logs show the repo being refreshed07:39
r0kk3rzyou added the things to the ks?07:39
r0kk3rzi would've thought it would abort if it couldnt find the package07:40
r0kk3rznot sure, i see no obvious error07:46
piggzr0kk3rz: i would have thought that too!08:14
malpiggz: do you want to make tiny change to both specs in broadcom-bluetooth, change license from Apache 2.0 to ASL 2.009:20
piggzmal: sure09:20
T42<elros34> @Oksana_: I stopped using storeman sometime ago because it upgrades other apps when I was installing something using it. Not sure if this is still a problem or it affects system packages.09:47
T42<edp_17> Hi guys. When I am building hybris-hal, I am getting this error:
T42<edp_17> This error doesn't make sense to me as I thought this step will create the kernel and boot.img files in out/target/product/$DEVICE/12:10
T42<edp_17> Or am I missing something?12:10
piggzmal: where does ssu gets its device vendor name from?12:13
atlochowski<elros34> yes, when you want to update just one app in storeman, it always updating all packages (apps).13:11
malpiggz: I think from file generated from this during build
malor something like that13:12
piggzmal: abranson: looks ok then ... file looks correct, and latte says (latte /latte ) in ssu status too13:16
piggzwe should request store access then maybe, but im not sure if we can read imei yet13:17
abransonmaybe that vendor should be something else?13:17
DylanVanAsscheWe can't read IMEI yet13:17
piggzabranson: the settings file is correct
piggzDylanVanAssche: whats your UID in ssu status ?13:19
DylanVanAssche02ba177f6d77 piggz (IRC) yours?13:19
DylanVanAsscheSo it's unique? That's not good?13:20
piggzno, it is good13:20
DylanVanAsscheBut it needs to be stable across reboots?13:21
piggzDylanVanAssche: mine is stable across boots13:23
DylanVanAsschepiggz (IRC): Mine is stable too 🎉13:28
malabranson: it's normal to have latte/latte13:29
malpiggz: DylanVanAssche is that related to wlan mac in any way?13:30
DylanVanAsschemal (IRC): ssu device id? Without IMEI it is based on MAC addresses if I remember correctly13:32
malyeah, it should be13:33
piggzmal: abranson: this seems naughty14:05
piggz[nemo@PinePhone ~]$ rpm -qa --qf '[%{obsoletes}\n]' store-client14:05
piggzDylanVanAssche: ^^14:05
abransonpiggz: Obsoletes: harbour-storeman <= 0.0.1514:06
abransonbecause of that old packagekit problem14:07
piggzthought i had it then14:07
abransonthere are a couple of those around to remove packages that break upgrades14:10
piggzmaybe ill try and cheat, and install the package manually at the end of the kickstart14:11
DylanVanAsscheThat's an option 😁 Cheating all the way!14:14
piggzDylanVanAssche: building now, if its works, ill MR :D14:16
DylanVanAsschepiggz (IRC): Nice 🙂 Should we add Jolla Store access first or just announce the 3.3.0 release when storeman is in there?14:17
piggzsee what pketo says ?14:17
piggzthe holder of store access :D14:17
T42<Nebrassy> so bluetooth works for me if I restart bluebinder after boot, but I can't get it to work normally15:34
T42<Nebrassy> is it possible to delay its start ?15:45
T42<elros34> check sources there is already delay, maybe it doesn't work for you15:53
T42<Nebrassy> I tried adding sleep 20 in various places in and but didn't help15:56
T42<Nebrassy> Also I'd like to build for multiple devices on the same machine, is it simple to do so?15:59
T42<elros34> have you checked in journal that it really delay start, not just timeout or smth16:01
T42<Nebrassy> With sleep 10 in , when I restart bluebinder it takes 10 seconds and works so it is delaying16:02
T42<Nebrassy> In it times out16:03
DylanVanAsschespiiroin (IRC): We talked a while ago about polling in mce for battery states. However, we haven't defined how we actually can implement it. For the PinePhone, cable states and battery full state are reported and the gauge is updated. Between those events, no updates are passed to the mce's battery state. We would only need to poll a sysfs interface in a certain interval, but I have no idea atm how to properly16:05
DylanVanAsscheimplement that in mce.16:05
T42<elros34> I mean on system boot16:05
spiiroinDylanVanAssche: I've planned something like that in my head. But that was mostly geared towards not missing unnotified battery percent changes during discharging = pace does not matter that much, heartbeat could be used.16:37
spiiroinhaving also major/rapid changes go without notifications are a bit more problematic16:38
spiiroinIIRC pine phone was silent about some things on cable disconnect -> getting some things immediately (disconnect) and some up to 12 seconds later (battery state) might cause hiccups16:38
spiiroinso there might be need for both time and event based polling - which easily turns ugly and unreliably (basically back to statefs problems)16:40
spiiroinI might have a change to get my hands on / remote connection to a pine phone - which could make things a bit easier to gauge16:41
DylanVanAsschespiiroin (IRC): That would be nice 🙂 I was thinking about a hacky hardcoded implementation (just to test 😁 ): mce_wakelocked_timeout_add -> function to read a file (the battery capacity) and send the value over the mce datapipes16:51
piggzspiiroin: still around?17:21
T42<Nebrassy> log for camera, after taking a video, switching back to photo crashes cam, also happens on jolla camera17:42
T42<Nebrassy> and needs restart to work again17:42
DylanVanAsschespiiroin (IRC): I quickly hacked together my idea here:
DylanVanAsscheWorks on the PinePhone, hardcoded like hell, drains the battery like hell but the battery gauge gets updated 🎉18:44
DylanVanAsscheIf this idea is okay, the code can be rewritten, improved, ... to get something usable. Or do you have a better idea?18:44
piggzmal: what should the permissions be in /etc/dconf/db ?19:15
piggzmy pro1 has r+x on the directories, but pinephone doesnt19:16
piggzr+x seems more normal19:16
piggzsure this must be related
malpiggz: folders 755 and files 644 root:root19:21
malpiggz: why would only your device have issues19:25
malpiggz: why is there a exit 0 here
piggzmal: iirc, as we were experimenting with how to add users/groups (ended up in kickstart) when i commented out the pre section, it required an exit 019:28
piggzthe whol section can be removed now19:29
piggz----and it has been!19:32
malpiggz: are all folders and files in /etc/dconf with wrong permissions?19:33
malpiggz: is the issue only with 3.3.0?19:33
piggzmal: the perms are wrong becuase of the fix we have in the kickstart ...19:33
piggzthe fix "fixes" it for us but isnt quite right19:34
piggzbefore the perms are totally off19:34
piggzmal: see
piggzmaybe ishoudadd an ls -lhR to the start to print the perms?19:36
piggzmal: and yes, only in 3.319:36
piggzand we do it again here
malhmm, that is very strange19:39
piggzmal: im running a pipeline now that wil print the permissions before edititing .... and i learned some chmod magic to set the correct permissions in a single command19:45
malpiggz: I'm making a test build for fp2 to see if the image looks good19:53
malpiggz: can you point me to some broken build in ci19:57
piggzmal: if you watch this log you will see the perms shortly
piggzmal: this is probably the first broken one
piggzmal: you can see here that after package installation, the oneshats fail
piggzand here, missing read perms for non root
piggzhmmmm, at the start the perms look reasonable
piggzthe oneshot for dconf update fails
piggzi bet its because we havnt created users by that point20:06
piggzactually, dconf oneshot is supposed to fail
T42<elros34> just an idea, did you regenerate ks file for 3.3.0? IIRC "user --name=nemo --groups" is no longer there20:11
piggz@elros34 we didnt ... but we have to do a lot of user/group manlging on pine becuase we dont have DHD to do it for us20:20
piggzbut a diff would be nice to see changes20:20
T42<elros34> basically just the user part:
piggz@elros43 thats interesting in itself ... i wonder where that happens20:32
piggzmust be sailfish-setup?20:32
T42<elros34> I guess, I didn't investigate it. I created 3.3.0 image for mainline device some time ago but didn't test it much20:34
piggzthe oenshots that are triggering the permission errors are all enable-*-hints20:47
piggzit could be that su -c ......20:49
piggzlets see20:49
piggzmal: that was it!!!!!21:05
piggz@elros34 thx!21:05
malpiggz: what?21:05
piggzsu -c "/usr/bin/oneshot --mic"21:06
malah, so old .ks21:06
piggz /usr/bin/oneshot --mic21:06
malgood to update my .ks files then21:06
malI always use the same files21:06
piggzas do we!21:06
piggzthe diff is pretty small21:06
piggzsee [21:28] <T42> <elros34> basically just the user part:
malpiggz: I think I had a lot older .ks because my diff was much bigger21:07
malcould be even years old .ks21:08
piggzours is about to get smaller with all those hacks taken  out21:08
malPSA: when you build images for recreate your .ks file, do not use old .ks files21:14
malpiggz: people using ci need to remember that .ks file issue21:14
malprobably easy to use old file there21:15
piggzi need to look into sailfish-setup to see what that does for us ... maybe we could drop some uer stuff if its handled there21:16
malpiggz: hmm, I recreated the .ks but it still has the wrong oneshot command21:17
piggzmal: how is su -c "command" different to command ? ... isnt mic running as root anyway?21:17
piggzmal: asl @elros34 where he made the diff from i guess :)21:17
malah, forgot to update target21:18
malpiggz: yep, now my build log looks clean21:33

Generated by 2.17.1 by Marius Gedminas - find it at!