Wednesday, 2018-10-03

*** frinring_ is now known as frinring03:03
dcalistechriadam: hello, I was not able to reproduce the lipstick-security-ui bug yesterday, but lipstick-security-ui was still in the boggus state when I begun my tests. I mean, the process was alive, but it didn't register to any new sailfish-secretsd and any request always ended with "No password agent registered".07:14
chriadamdcaliste: if you run sailfish-secretsd manually, then kill lipstick-security-ui (and it should get restarted) does it then reconnect to the sailfish-secretsd?07:15
chriadamI guess you're saying "no, it does not" ?07:15
dcalisteI tried to attach a GDB session to lipstick-security-ui, but GDB complained that the operation was not possible but the /proc/sys/kernel/yama/ptrace entry doesn't exist.07:15
chriadamI wonder if the issue is in the passwordagentplugin (i.e. it should remove the old/stale connection, and allow a new connection from the lipstick-security-ui)07:15
dcalisteThen, I killed lipstick-secrity-ui and then the new one registered properly to the secret daemon and during all day, the bug didn't reappeared…07:16
dcalisteYeh, annoying. I cannot exclude that I also restart the secret daemon before checking that it was working. So it's not sure it's an issue with passwordagent plugin.07:17
chriadamso somehow the connection is getting into a stuck state - the lipstick-security-ui exists, but the connection isn't working, and since lipstick-security-ui doesn't get restarted it doesn't recover07:18
dcalisteBut I cannot figure out what steps are required to make the bug appears. I wonder if just stop and restart the sailfish-daemon continuously may lead to it or not…07:21
dcalisteI'm going to add debug messages to the connection / disconnection in passwordagent plugin to try to follow what is happening and exclude (or not) a failure from plugin side.07:22
chriadamsounds like worth trying.  I wonder if device suspend might be related also07:23
chriadamsailfish-secrets PR#121 provides a bunch of manual test scripts which could be used to stress-test in case that might trigger07:25
chriadamjust run in a privileged terminal07:25
chriadamand PR#142 expands those more07:26
dcalisteI see, will try all these approaches indeed to gather more information since the reproduction is somehow difficult to achieve.07:27
chriadamthanks very much!07:27
chriadamI appreciate you doing the legwork to find a repro for this issue.  definitely bad if the device can get into a stuck state requiring user to reboot to perform secrets operations :-(07:28
dcalisteYep, same feeling ;)07:28
dcalisteBut now, I know that just lipstick-security-ui need to be killed which is nicer than restarting full lipstick (not a day to day solution though).07:29
dcalisteBesides, looking at passwordagent code, I understand what to put to get double check windows for new password. So it's working now!07:31
chriadammy gut feeling is that we should make lipstick-security-ui a dbus service which kills itself automatically after 10 seconds of inactivity, and which passwordagentplugin is a client of, rather than the other way around..07:31
chriadambut I need to discuss that with Andrew perhaps, he presumably had good reasons for doing it the current way.07:31
chriadamthat's good news :-)07:32
dcaliste(Just a bit annoying (API-wise) that we need to override default repeat message to get the repeat window, as far as I understood.)07:32
dcalisteAbout the way of client / server, I would agree with you also, but as you said I don't have the full picture.07:32
chriadamI wonder if the default repeat message is simply defined wrongly (and it "should" be the "correct" one), or if that was deliberate to prevent double-check-dialog by default?07:34
dcalisteWell, the dialog switch to CreatePassword state if promptText contains RepeatInstruction, see line 653. But IMHO, it's not possible to set RepeatInstruction with an empty string because of line 208 in interactionparameters.h07:37
dcalisteSo to be in the CreateWindow state, one needs to give a string to the repeat instruction, which disable the possibility to use the default one.07:38
chriadamI see.  we'd need some placeholder like "default" or something07:41
chriadamthen if prompt.value(RepeatInstruction) == "default" then use the actual default one (or set it to empty etc to cause it to be the default one)07:41
dcalisteYes, and it would be more intuitive from API to have some prompt constructor doing this, like PromptText::PasswordCreationPrompt(). I mean more intuitive than having to provide a specific text with setRepeatInstruction() to get the repeat capability.07:45
chriadamright, I agree07:45
chriadamor some flags to the constructor to specify the behaviour07:45
dcalisteWithout reading the code, I just put setNewInstruction("") to do what I wanted.07:45
dcalisteI mean, it's nice to be able to tune every message individually, but it lacks a default job-oriented setter (or constructor).07:46
chriadamyeah.  that sort of thing would definitely be nice.  I won't have time in the next two weeks to look into making that change, but perhaps after that I can do that.  can you create a github issue so I don't forget?07:47
dcalisteAt once ;)07:47
chriadamI have to head home now.  Good luck with finding the repro for the stuck lipstick-security-ui, let me know what you discover there, and I'll definitely prioritise trying to fix that one as it sounds very serious.07:49
chriadamhave a great night :-)07:49
dcalisteThank you have a safe trip back home. See you.07:50
*** Mirv_ is now known as Mirv11:33
*** flx_ is now known as flux12:31
lolekmonich: hi, any news regarding hd voice support?13:52
monichno :/13:52
lolekmonich: not good. Do you think that in the nearest (say a year) future, it will be possible to get it working?13:57
monichlolek: within a year, I would estimate the probability at roughly 30%14:00
monich(don't take this number too seriously)14:00
r0kk3rzgenerally unless a customer pays for it, it'll get pushed down the priority list14:09
lolekmonich: that's very very baad :(14:15
lolekr0kk3rz how much? Maybe community could make some donation for a feature?14:15
r0kk3rzno idea, it would be up to jolla to quote for such work14:16
r0kk3rzwhich i doubt they will do14:16
DennisRoczekr0kk3rz: yeah, i do want hdvoice.14:49
* Tekk_ didn't even know hd voice was a thing. He just uses voip for most things...14:59
DennisRoczekbecause the quality of the simple normal voise calls are much much more better15:00
tbrif that in any way relies on an OS side SIP stack, then: LOLOLOLOLO15:03
r0kk3rz[m]it doesnt15:03
*** ced117_ is now known as ced11716:38
tortoisedocSfietKonstantinW : how did you get the obs to use rust122 ?18:00
tortoisedoctbr : do you know about obs?18:01
tbrtortoisedoc: rumor has it that I indeed do18:01
tortoisedoctbr : im a total obs n00b (huzzah!) and im struggling with getting obs to be able to install built packages; do they need to be separately be deployed after having been published, or does publishing make them available to any build task?18:02
tortoisedocas in, is publishing the packages enough to use them as dependencies in builds of other rpm packages, or are additional steps required?18:05
tbras long as they are in the same project, yes18:05
tbryou don't even need to publish them for building against them (only relevant for closed sauce)18:06
tbrif they are not in the same project you need to tell your project that it can pull from the other project18:07
tortoisedoctbr : then there's something else wrong I guess
tbrdoesn't rust need a rust build to boostrap itself?18:08
tortoisedocwell apparently not the rustc compiler (cargo & co do)18:09
tortoisedocrust 1.23 builds out of 1.22, but I cant build 1.2418:09
tortoisedocout of 1.23 due to "unavailable"18:09
tbrI only see one rust package18:10
tortoisedochmm okay18:10
tortoisedocI have built only rpm's, made no packages18:11
tbras in you can't build against yourself18:11
tortoisedocah! okay, that explains now18:11
tortoisedocso I need a new package for each version18:11
tbror maybe you can, but it's non-trivial (haven't done such stuff)18:12
* tortoisedoc goes googling about obs and packages18:12
tbrso the result of that thing seems to be packages of 1.2318:13
tbrif you don't want to have a second rust thing, then you'd need to do shenanigans with a bootstrap project18:14
tbrbasically a project that provides a rust package until you've built your own rust package18:14
tbrI don't expect that to be very well documented18:14
tortoisedocits pretty much non-documented yes18:15
tortoisedocthis is all a bit poc18:15
tortoisedocor actually just playing :)18:15
tortoisedoci was surprised to find such configure scripts18:15
tortoisedocin the first place18:16
tortoisedocand yes I'd definitely avoid a bootstrapper for now at least18:16
tbrtortoisedoc: scroll to the very bottom of the "META" tab
tbrthat's how a bootstrap looks like <path project="mer-core:aarch64:bootstrap" repository="Core_aarch64"/>18:18
tortoisedocok thanks18:19
tbrjust guessing, but I suspect you could just do the same but with "home:sfietkonstantin:sailfish:rust" and have your local package build 1.24 on top of that. Both packages could retain the name rust and would just differ in versioning18:20
tortoisedocrust is a special beast18:21
tortoisedoc1.22.0 wont build 1.24.0, heck I know how that's possible18:21
tortoisedocits like a rolling turd18:21
tbryes, but isn't that package building 1.23? then your package can build on top of that18:22
tortoisedocyes, that's what I was doing wrong; I was rebuilding 1.24 in the same package18:25
tortoisedocthat built 1.2318:25
tortoisedocfunny fact, adding a new package (rust-1.24) nuked my meta info of the rust package18:27
tortoisedochere be dragons :)18:27
tbrI guess you possibly could engineer it to be one long build process that chains all this and results in packages18:30
tbrit's somewhat inelegant though to every time build 1.23 from 1.22 to then build 1.24 from 1.2318:31
tortoisedocguess its due to the "infant" nature of rust18:35
tortoisedocwith the complice of "lean" sw development methodologies18:35
tortoisedoceveryone gets left behind, unless they move forward.18:36
tortoisedocwhat a beatiful attitude, in 202018:37
tbrchooo chooo18:44
tortoisedockimmoli : Click dat19:26

Generated by 2.17.1 by Marius Gedminas - find it at!