Friday, 2021-05-28

piggzlbt: great news :)15:32
poetasteroh. that is great news.15:34
poetasternow if someone would convince me I should use it :)15:36
Nico-old-defunctIf you have no need, don't use it. It will probably still go away at some point15:36
piggzpoetaster: you do ports, why wouldnt you??15:37
piggzit genuinly makes everything easier15:38
piggzand the more use it, the less it will go away15:38
piggzat last i can trigger a build of software for multiple architectures and versions15:38
poetasterI'm the type of person that looks at each port, tries it on multiple devices and archs.15:38
poetasterI mean, I don't trust a build until I've used it.15:38
lbtI can vouch for the " <piggz> and the more use it, the less it will go away"15:39
Nico-old-defunctGood point :315:39
piggzpoetaster: you have a port right?15:39
poetastererrr. no. you have ONE of my POrts.15:40
lbtpoetaster: we use OBS internally for guaranteed consistent/clean builds15:40
lbthappy to talk about the benefits of OBS if you're not familiar15:40
poetasterI understand that. I'm just a lowly developer. But I could probably build a better workflow.15:41
piggzpoetaster: why build manually like a noob, when you can do distributed builds in the cloud :D15:41
poetasterpiggz: because all the work I'm doing now has GUI components and I need to 'see' to believe.15:42
lbtpoetaster: I've not been around for quite some time but FWIW I've been working with jolla and before that Nokia/Mer on using the OBS to create sane workflows for over 10 years.15:42
piggzsure, do a test build locally ... but then trigger a build on obs, and have it create armv7/64/i486 versions for you15:43
lbtyou can even add the OBS build repos to your devices for direct install15:43
piggzi do ^^15:44
poetasterpiggz: ok, so 'just' test armv7 on the oldest device and trust it'll just work.15:44
poetasterpiggz: might work form most of my stuff.15:44
poetasterlbt: ah, so, eat my own lunch directly out of obs.15:45
piggzi mean, i cant test arm64 (yet) anyway15:45
lbtpoetaster: yes - and it's a proper clean build15:45
lbtsome developers won't have multiple devices so simply cannot test15:45
poetasterlbt: I'm offended. All my builds are clean. The code, on the other hand ....15:45
lbtbut they can use OBS repos to allow the rest of the community to download and test devel builds15:45
lbtpoetaster: not as clean as an OBS build ;)15:46
lbtit will update packages in the buildroot if you *or someone else* has changed it15:46
lbtagain - a lot of this is more about collaboration and there's not so much of that nowadays15:47
lbtin terms of multiple people working on a bigger group of packages which go together (like Nemo)15:48
lbtbut it's still great to share binaries with other users via a consistent community resource prior to releasing15:48
piggzlbt: so, jolla has ducked out of a difficult answer to my community meeting quiestion? :D15:57
lbtI missed that...16:03
lbtahhh  - actually I think that was mentioned internally :)16:05
lbtI think the question still stands - OBS needs to be part of the community so I think there is a responsibility on both sides16:07
poetasterpiggz: the volla claims to be  aarch6416:08
poetasterok. I'll get to obs as soon as I get my latest 3 releases out the door.16:09
poetasterlbt: the question if often how much community responsibility jolla would like and under what conitions.16:10
lbtwhat do you mean?16:10
poetasterlbt: 'be part of the community' in the sense of community is responsible for hosting?16:13
poetasterlbt: I may not haveunderstood you ;)16:14
lbtOh, I see16:14
malI think he means community needs to use it enough16:14
lbtyes, I did16:14
lbtit needs to be part of how the community works together16:15
lbtnot everyone of course - but it should be a meaningful service16:15
poetasterlbt: ok, got it. it seemed though that it's a resource drain? wouldn't that suggest outsourcing? to community?16:15
malin a way it's a shame most of the community ports are maintained by one person, some rare ones have more than 1 person16:15
lbtpoetaster: I ran it as a community service for many many years - all the Mer infra16:16
poetastermal: that is very true. I just jumped in to do a bunc of maintenance because of that very problem.16:16
lbtjolla contributed to costs in the early days and eventually took over all the costs16:16
poetasterlbt: ok, got it. manpower was donated, hosting covered?16:16
lbtyes, essentially16:17
poetasterlbt: mal has a very important point that not sharing the load, getting others involved can endager a project.16:17
poetastersorry, you all know that.16:17
lbtwe do16:18
lbtbut it's worth saying out loud16:18
lbtI think it would be a good time to try and rekindle some collaborative work16:18
poetaster:lbt phwew. I've been a bit 'too present' on the forums just trying to get up to speed. and now I'm maintaining 'too' much.16:18
poetasterlbt: hear, hear.16:19
lbtyou should find OBS will help with maintenance if you learn to use it correctly16:19
poetasterlbt: some of the work I took on had immediate involvement from others just by ending up on githb.16:19
poetasterlbt: reading the obs-user-quide.16:20
Nico-old-defunctI was so inactive, I lost my regular status on the forum >.>16:20
lbtthe user guide is interesting but we tend to use the OBS as a builder triggered by git tags/webhooks16:21
lbtosc and suchlike are mainly used for advanced build debugging nowadays16:21
poetasterlbt: makes sense. I'm trying to do proper releases with binaries on github though I'm sure I'm not doing it correctly.16:21
lbthehe - well, I'm currently re-writing years worth of "release" cruft for SFOS - if you could see what I see you'd know that "properly" isn't a high bar :D16:23
poetasterlbt: ah, I can imagine, in too many languages ...16:23
lbtyah... bash/python2/3/ruby .. too many servers/services ... eek!16:24
poetasterlbt: nodejs, go ... so I have a tag/release based hook/action workflow in future?16:24
lbtyes, I hope so16:25
lbtI need to get webhooks on community working again though16:25
poetasterlbt: ok, so point me in the general direction....16:25
Nico-old-defunctWhen will I be able to deploy from OBS directly to harbour? :316:25
poetasterNod: now that is a good question :)16:26
lbtNico-old-defunct: that was on the cards way back but it won't happen now16:26
poetasterthe QA in harbour is good, though, so I'm ok with the 'labour'16:27
Nico-old-defunctIt would be nice to have some API for harbour at least16:27
Nico-old-defunctpoetaster: You don't run sfdk check in CI? ;p16:27
poetasterNod: be evil, use curl.16:27
lbtI pushed to have the maemo-builder-like devel/release happen on OBS but we (the community) didn't go that direction16:27
Nico-old-defunctWell, it would have been awesome!16:28
lbtYou can/should submit OBS built binaries to harbour though16:28
poetasterNod: no, I'm prefer to do vcs stuff by hand. probably wrong though16:28
Nico-old-defunctWell, I build them in gitlab-ci and don't actually use OBS atm and my apps aren't even harbour compatible... but maybe some day! :D16:29
lbtoops - stray +
poetasterNod: One of my goals is to always have a harbour version. 3 of the apps I adopted from planet_os were not harbour compatible but easily made so.16:30
lbtI'm too used to matrix nowadays - could edit a previous comment16:30
poetasterNod: easily, with HELP from the community!16:30
poetasterlbt: I think we should go there so long as an irc bridge is in place16:30
poetasterlbt: thanks!16:31
lbttbh I'm not sure if wh is still available... it's been too long (6 yrs)16:31
lbtnp - give things a bit for OBS to upgrade and webhooks to work. I think it may be dead atm16:31
lbtI'm swamped internally for a few weeks but I'll probably end up doing OBS upgrade over some weekends soon16:32
poetasterlbt: I can get up to speed in the meantime.16:33
lbtfeel free to ping if you have questions on it - I'm keen to help people who want to establish some collaboration16:33
lbtI'd better get back to work o/16:33
poetasterlbt: ditto ...16:33
poetasterthe dumbest part of my job, doing what no one wants to do on a friday evening.16:34
lbthehe - what do you do?16:34
poetasterlbt: system admin/ programmer activist newsite, netzpolitik.org16:35
Nico-old-defunctlbt: Thank you for maintaining OBS and putting so much effort into it and also agreed on the Matrix stuff :316:36
lbtsounds familiar16:36
poetasterlbt: all the editor are afraid of the newsletter tool and gone for the weekend. so I do that, 'too'16:36
poetasterNod: DITTO!16:36
lbtNico-old-defunct: my pleasure16:36
lbtNico-old-defunct: now I have to persuade you to use it ;)16:36
Nico-old-defunctFor OBS? Yes :D16:37
Nico-old-defunctBut I just have one tiny app that builds perfectly fine in gitlab ci .-.16:38
poetasterlbt: I'm sold if I can use webhooks/tags with github.16:38
lbtNico-old-defunct: they all count!16:38
lbtMaybe it would be woth using Chum again...16:39
poetasterNod: I have 9 apps which build perfectly fine, by hand in the sdk :) I'm soooo slow.16:39
lbtit was a way to do "early-access" to users for testing16:39
Nico-old-defunctpoetaster: That sounds like you should use OBS16:39
poetasteryeah. and there is a lot of refactoring to do. sigh. well, it's worth it in a least 3 or 4 instances.16:40
lbtI need to hand over my openGl game to someone who could take it forward16:40
Nico-old-defunctI need to finish my Sailfish matrix client ;p16:40
lbtooh yes!16:40
lbtanyhow... I'm going!16:40
Nico-old-defunctSee ya <316:41
poetasterNod: yes!16:41
poetasterNod: which client was that, nheko?16:41
Nico-old-defunctWell, and konheko, the smaller version for Sailfish16:44
Nico-old-defunctneko (cat) -> koneko (kitten)16:45
poetasterNod: sorry, that's what I meant.16:46
Nico-old-defunctI gave so much thought to the name!16:47
poetasterNod: I've been in one of the marathon group testing phases, e2e crypto was so focal. sigh.16:47
poetasterNod: my last app, just for practice, is named after the service it uses. ZERO imagination.16:48
Nico-old-defunctI mean, I did call my app Matrix at first and then spoon, but...16:49
poetasterspoon is good but probably the wrong conotations ;)16:57
Nico-old-defunctAlso I since pretty much took over nheko and someone floated the konheko idea, which I simply loved :317:09
poetasterNod: ! ok, danger, danger! I've been compaing nheko and mirage on desktop.17:17
poetasterNod: I'd prefer less fancy, aka nheko, but the crypt opts and retention stuff....17:18
poetastergit push17:18
Nico-old-defunctcrypt opts?17:26
piggzlbt: so, will you align with internal obs to make maintenance easier?17:58
T42<adampigg> @b100dian
piggzdid you resolve that?18:12
T42<b100dian> I did a partial revert of some hybris patches and got it resolved, but am now running without them (so I still have that output) and it seems benign18:27
piggzok ... i have lipstcik and hwcomposer crashing atm18:28
piggzthat was the only log, but if its unrelated i'll leave it18:28
T42<b100dian> it's system overlay, piggz18:28
T42<b100dian> remove it18:28
T42<b100dian> Sorry I thought you said camera and hwcomposer18:29
T42<b100dian> Whats the crash?18:29
piggzyes i have a system overlay18:32
T42<b100dian> I had a stack corruption on camera and hwcomposer services with the system overlay18:34
piggzill remove it and bind mount the individual files18:36
T42<b100dian> first check if the crashes are gone by disabling it18:40
T42<b100dian> maybe you dont have my error18:40
piggz@b100dian well, test_hecomposer now works!!18:44
T42<b100dian> good:) One step closer to amazfish on 4.118:54
piggzone step close to actually having something to demo on sunday, when im supposed to be showing this off!!!!!!!!18:55
T42<b100dian> i know. You should at least have lipstick, right?19:00
T42<b100dian> Whats the next error19:00
piggzyeah, i just need lipstick19:00
piggzi mean, which idiot agrees to demo sailfish at the volla developer event .. then wipes his phone that week and starts a fresh port!19:01
T42<b100dian> Is dbus running?19:02
T42<b100dian> 'Always have lipstick on stage'?19:02
piggzyeah, dbus is running...19:06
piggzdoes anyone know how to make a file appear in /system without an overlay?19:07
piggzi need to make the icu files appear, but as they dont exist, a bind mount doesnt work19:07
T42<b100dian> what's the error? can you also share journal?19:08
piggzsure, just rebooting and getting a clean journal19:09
T42<b100dian> maybe overaly works with files if you'd like to try
piggzmal: ^^19:11
malpiggz: this looks perhaps odd "dbus-daemon[913]: SELinux: Class dbus not defined in policy."19:14
T42<b100dian> Is this needed? > Failed to start Droid bootctl.19:24
malsome device needs it, some don't19:24
piggzthats a/b devices19:24
malyou will see if the device stops booting after 5-7 reboots19:24
T42<b100dian> yes, but the question is, does it block getting to graphical target?19:25
T42<b100dian> mal: oh, interesting. And then what do you do, flash lineage?19:26
piggzno, reset you slots19:26
piggzboot loader will switch to slot b after too many fails on slot a19:26
malif that happens then fix that service (you can make device boot again by flashing kernel again for example or resetting things with fastboot)19:26
mal@b100dian you can try to disable the service for now19:27
mal@b100dian systemctl mask droid-bootctl.service19:27
T42<b100dian> Thanks for the explanation piggz and mal. I have it disabled with a /dev/null link in sparse, I don't think I've seen the device not booting till now..19:28
T42<b100dian> piggz "jolla-startupwizard-pre-user-session.service: Failed with result 'core-dump'.19:28
malthat is unrelated19:29
piggzb100dian: on pro1, bootctl needed a lot of hacking to get the correct parittion names, as they were hard coded by qcom19:29
T42<b100dian> I'll look into that19:31
mal@b100dian I'm a bit confused with your problem19:34
T42<elros34> @adampigg sparse-10/usr/bin/droid/ looks like a sane approach for me instead creating bunch of systemd services19:37
piggz@elros34: that only works if the target file exists19:38
piggzalso, i just tried an overlay, with just the symlinks needed, but that brought the crashes back!19:38
T42<elros34> ah. Somebody mentioned about typo in this patch (PATH3 <-> PATH4), maybe is also worth to check it:
piggz@elros34: i need mini* services to load (and related) .. but theyre in /apex/.... and i need them to appear in /system19:51
mal@elros34 I think that patch might not be used really on 17.1, at least I see no files in hw/ folders in libexec19:51
piggzback in the old days, was the way :D19:51
mal@elros34 that patch was only really needed on some older devices which needed patches hwcomposer etc19:51
T42<elros34> ok19:52
T42<elros34> I think It was even useful for me long time ago fix some null pointer access in hwcomposer19:53
malI think I made the original patch when I was porting Xperia Pro19:55
malquite many years ago19:55
T42<b100dian> mal: re confusing, sorry I haven't shared a problem yet, I was mostly looking into piggz's logs20:11
T42<b100dian> piggz: re crashes, that's interesting, which file did you tried?20:12
piggzsod it, for now, im remounting rw!20:18
piggzoh, apparently i cant!20:18
malpiggz: why not?20:21
malyou might need to remount both system and system_root in correct order20:21
piggz/ # mount -o remount,rw /system_root20:22
piggzmount: /system_root: cannot remount /dev/mmcblk0p31 read-write, is write-protected.20:22
piggz/ # whoami20:22
piggz/ # mount -o remount,rw /system20:22
piggzmount: /system: cannot remount /dev/mmcblk0p31 read-write, is write-protected.20:22
malhmm, strange20:22
piggzmal: maybe i also had the overlay20:24
piggznope, same20:25
piggzso, any other options like making the system look in other folders for libs?20:30
malpiggz: what lib do you want it to look for20:31
piggzmal: the icu ones20:31
malpiggz: that's what my planned libhybris patch would do :)20:31
piggzmal: well, pass my test patches!20:32
malpiggz: there is no patch yet20:32
piggzmal: small hack made20:33
piggzon this system, /odm isnt used, so, i can create it!20:34
malnever thought about that20:35
malpiggz: not sure if you could hack ld.config.29.txt also20:35
malthe problem is that some processes don't load that20:36
piggzok, so back to dbus/lipstick20:51
T42<b100dian> Is /sys/selinux/enforcing readable?20:54
piggz@b100dian doesnt exist20:59
piggzno /sys/selinux20:59
malpiggz: probably /sys/fs/selinux/21:01
piggz/ # cat /sys/fs/selinux/enforce21:01
T42<b100dian> 0 is good, but is it world readable? ls -l /sys/fs/selinux/enforce21:09
piggz@b100dian it is21:16
T42<b100dian> ok, I'm out of ideas. maybe dbus is just fine. is it expected to be able to launch lipsitck as root? maybe try defaultuser21:20
T42<b100dian> but are you manually starting it because it crashes or because it doesnt reach that stage?21:21
T42<b100dian> Do you have repeated 'Waited one second for ... waiting another' in logcat for a particular service?21:21
piggzi did for service manager21:22
piggzbut ive got rid of that by disabling the gps service21:22
T42<adampigg> (Document)
piggzmal: ^^ couldnt find a paste service that would take such a large file, so i will abuse Thaodan's telegram bridge21:35
T42<elros34> that is why minimer is better idea unless you already tried it and works good but only lipstick crash22:06
T42<b100dian> How to debug camera not working? I see dump-camera-parameters tries to open /dev/v4l2 but I have /dev/v4l-subdev0..4022:21
T42<b100dian> actually it tries /dev/video0..n first22:23
T42<b100dian> are hadk instructions are up to date for camera debugging?22:25
T42<ruspartisan> I managed to run pinephone image in QEMU, but resolution is 640*480, can I change it to something better?22:44
T42<ruspartisan> (Document)
T42<b100dian> Looks good:) no cmdline param in qemu for resolution I guess?23:23
T42<b100dian> My gst-inspect doest show droidcamsrc.. That should come with gst-droid, shouldn't it?23:24
T42<ruspartisan> @b100dian [Looks good:) no cmdline param in qemu for reso …], Qemu usually relies on guest to set resolution. Maybe there's some magic settings file in sailfish or linux kernel param23:34
mal@b100dian is minimediaservice running?23:38
T42<b100dian> it shows up in ps aux, yes23:40
T42<b100dian> the only output from it in logcat is `/usr/libexec/droid-hybris/system/bin/minimediaservice: open() failed for /dev/cgroup_info/cgroup.rc:` though23:42
T42<b100dian> (which is every process liinking to libprocessgroup)23:52
malyou could try removing the ~/.cache/gstreamer-1.0/* file and try camera again23:58

Generated by 2.17.1 by Marius Gedminas - find it at!