libera/#maemo-leste/ Sunday, 2025-02-02

freemangordonWizzup: update-initramfs: Generating /boot/initrd.img-6.1.0-30-rt-armmp07:10
freemangordonwhat the?07:10
freemangordonsicelo: wait, the driver reports capacity 0 if not calibrated?07:39
siceloyes, because that's what the hardware reports07:41
freemangordonthis is on d407:41
sicelothe registers are reset -> 007:41
freemangordonhmm, lemme check the driver code, that does not make sense07:41
sicelowhy do you say that?07:42
freemangordonbecause the driver should now it is not calibrated and report ENODATA or something07:43
freemangordonnot 007:43
sicelothen what if the capacity is really 0% :-)07:44
freemangordonwhat is capacity 0%07:44
freemangordoncapacity is absolute, no?07:44
freemangordoncharge_level might be 007:44
freemangordonbut then, you have battery low irq07:45
sicelowhat does "absolute" mean?07:45
freemangordonmAh07:45
freemangordonlemme check what driver does07:45
siceloin any case, even with ENODATA, upower will still be unhappy iirc. N900 driver used to report the ENODATA07:46
freemangordonok, but ENODATA is something we can use to argue with upower maintainers07:47
freemangordonit is not 007:47
freemangordonthe ^^^ statement (garbage in, garbage out" is correct, if we disregard the tone07:47
sicelook, we can try that approach. it's easy to implement In the kernel. if capacity = 0 && voltage > VOLTAGE_MIN_DESIGN, return -ENODATA :-)07:48
siceloactually that's why i recently implemented VOLTAGE_MIN_DESIGN for the N900's driver ... i think D4 already reports it07:49
freemangordonsicelo: BTW, which exactly property we talk about?07:49
freemangordonPOWER_SUPPLY_CAPACITY?07:49
siceloyes07:50
freemangordonhmm https://github.com/maemo-leste/droid4-linux/blob/maemo-6.6.y/drivers/power/supply/cpcap-battery.c#L75807:51
freemangordonit seems it already return ENODATA under some conditions07:51
sicelopart of the upower problem is that they made the assumption that capacity_level is only valid for peripherals not supplies, e.g. bluetooth devices.07:55
sicelofreemangordon: an idea I had recently is ... we leave kernels as they are, and upower capacity/percent as is, but just teach it to export the capacity_level property straight from kernel08:11
siceloso no assumptions get made by either kernel or upower08:11
siceloa upower consumer (like us) can then make its own decision to react to capacity_level if it wishes08:12
freemangordonbut the real issue now is that upower does a shutdown, no?08:13
sicelowe can suppress that shutdown in latest versions of upower, so it's no longer a problem08:14
freemangordonthen, what is the problem?08:14
sicelothe shutdown is there in the version found in daedalus ... but we can build newest, i guess08:15
siceloright now there's no major problem really :-)08:15
siceloexcept ... yes we don't shutdown, but, what do we report in the battery applet & mce08:16
freemangordonif the property is not reported, we can just say "battery not calibrated" in the applet and show half-empty icon08:17
freemangordonmce does not care about charge anyways, it cares about 'battery low' signal only08:18
freemangordonand voltage level08:18
freemangordondsc_: please, fix conversations to not connect accounts if availability is offline08:19
siceloi completely agree with that approach ... and that's why i was thinking about exposing capacity_level in upower, so at least we don't always show the 50%, but can show 75%, 25%, etc. solely from capacity_level08:21
freemangordonahaa, now I see08:21
freemangordonyeah, makes sense08:21
freemangordon1maemo-summoner: died loading invoked application: '/usr/bin/osso-abook-home-applet.launch: cannot dynamically load position-independent executable'09:21
freemangordon1will have to fix that09:21
dsc_freemangordon1: alright09:45
freemangordon1thanks!09:46
Wizzupsicelo: wait, upower does a shutdown through elogind? surely we can disable that10:58
Wizzupdo you have some links10:58
Wizzupfreemangordon: are you sure this is conversations doing this and not status applet? cc dsc10:58
Wizzupfreemangordon: I see status applet blinking green upon reboot trying to connect when there is no ntework10:58
WizzupI don't think it is conversations that onlines accounts10:59
Wizzupfreemangordon: hmm I am getting an error with X and permissions11:36
Wizzupalso our old upower remains because we epoch'd it :)11:42
Wizzuptry dpkg -l | grep '+m7'11:42
Wizzupsicelo: https://gitlab.xfce.org/xfce/xfce4-power-manager/-/issues/146#note_9519511:45
Wizzupso we'll need *newer* upower not to have it be dumb :D11:46
Wizzupfreemangordon: so yeah I am seeing issues with seatd not running11:48
Wizzuplooks like my X wants to use libseat/seatd11:48
freemangordonthat's not an issue11:50
freemangordonBTW, I successfully dist-upgraded my d411:51
freemangordonso not sure what issues you have on yours11:51
Wizzupfreemangordon: well, this is mz617, but it is the same12:01
freemangordonif you dd an image, does it work?12:02
Wizzupnot possible12:03
Wizzupno sd card12:03
WizzupI'll check user groups12:03
Wizzupseems upower will be a real nuisance12:16
WizzupI think it's powering off my tablet immediately due to this 0 thing12:16
freemangordonyeah12:16
freemangordonyou should boot on charger connected12:16
Wizzupcan't12:16
Wizzupneed serial/keynoard12:16
WizzupI chmod -x'd upower12:17
Wizzupnow X starts12:17
freemangordonok, maybe put some value in /var/lib/$wherever_the_value_is12:17
freemangordonthe calibration value12:17
WizzupI don't know what file you're referring to, but that doesn't work on mz617 I think12:18
Wizzupwill check12:18
freemangordonsec12:18
freemangordon/var/lib/droid4-battery-calibration/charge_full12:18
Wizzupno idea what to write tbh12:22
Wizzupin any case, with upower disabled it boots12:22
freemangordonyeah, we have to fix upower, somehow12:23
freemangordonbut let me see if I can properly fix gconf first12:23
Wizzupagreed12:23
Wizzuplet me charge tablet and  then consider what to write there12:24
freemangordonit shoud not matter12:25
freemangordonput 1500000 there12:25
WizzupI think gtk chimaera is 2:2.24.33-2+m7.8 and daedalus is 2:2.24.33-712:26
Wizzupso we are not yet using daedalus gtk12:26
Wizzupnot sure why/how12:26
Wizzupmaybe pkg names just changed12:26
freemangordonBTW, scrooling an addressbook on daedalus seems way faster12:27
freemangordoncould be placebo, but...12:27
Wizzupdon't know, but I think driver seems a bit better12:33
freemangordonmhm12:34
freemangordonwe shall enable fullscreen in the browsers12:34
Wizzupseems libecal also isn't on newer version12:35
Wizzupii  libecal-2.0-1:armhf                                 3.38.3-2+m7                            armhf        Client library for evolution calendars12:35
Wizzupii  libecal-2.0-2:armhf                                 3.46.5-1+4m7.1                         armhf        Client library for evolution calendars12:35
Wizzupolder one is installed12:35
Wizzupdoesn't make sense to me12:35
freemangordonapt upgrade?12:35
Wizzupclutter too12:36
Wizzupdist-upgrade12:36
freemangordonnow, what happans if you do 'apt update/upgrade'?12:37
Wizzupnothing, as dist-upgrade is a superset of it12:37
freemangordonhmm, those packages have epoch or?12:38
Wizzupno12:38
WizzupI removed them with dpkg -r and it seems like nothing depends on it, strangely enough12:38
freemangordonis calendar installed?12:38
freemangordonalso, which clutter?12:39
freemangordon1.0?12:39
Wizzupii  libclutter-0.8-0:armhf                              0.8.2.79+m7.1                          armhf        Open GL based interactive canvas library12:39
freemangordonnothing depends on that?12:39
freemangordonthat doesn't make sense12:39
WizzupI was talking about libecal etc12:40
freemangordonah12:40
freemangordonhmm, see apt-cache policy libclutter-0.8-012:41
WizzupI wonder if one or two pkgs in daedalus don't have the prefix12:41
freemangordoncould be12:42
Wizzuplet's rebuild clutter once the gconf jenkins fix is in12:42
freemangordonbut better rebuild tham12:42
Wizzuphttps://maedevu.maemo.org/pkgweb/daedalus/main/binary-amd64/libclutter-0.8-0.html12:42
Wizzupyeah this one needs a rebuild12:42
freemangordonI think gconf is better now12:43
freemangordonbut, I am afraid I cannot fix it for dist-upgrade12:43
Wizzupbtw, looks like the ebook stuff wasn't necessary because there was also a newer version installed12:43
freemangordonlemme try jenkins12:44
Wizzupit's not just dist-upgrade, I think also when debian say upgrades sgml12:45
freemangordonyeah12:45
freemangordonsgml does something12:45
freemangordonWizzup: jenkins is fixed, but I have no idea how to fix dist-upgrade12:47
freemangordonmaybe sgml-base shall pre-depend on gconf12:48
Wizzupfreemangordon: maybe we can fix some of this in our meta pkgs12:50
freemangordonyes, but ni idea how12:51
freemangordon*no12:51
freemangordontrying a fix now though12:51
Wizzupmaybe we can have gcon2 not depend on gconf-common, hope it gets pull in some other way and fix it that way12:52
freemangordonhmm...12:52
freemangordonmaybe12:52
Wizzupupon dist-upgrade now, indeed the gconf bug is back, btw12:55
freemangordonI know12:55
Wizzupyeah12:56
freemangordonone more try, if that does not work, I'll give-up13:07
Wizzuplet's hope :)13:11
freemangordonI don;t really believe this will work, but...13:12
freemangordonWizzup: hmm... *maye* it is really fixed13:32
freemangordongoing to dist-upgrade my VM13:32
Wizzupgreat13:34
freemangordonwell, lest confirm13:35
freemangordon*lets13:35
freemangordonyou can build clutter etc in thye meanwhile, jenkins is fixed13:37
freemangordonWizzup: ^^^13:37
freemangordonWizzup: dist-upgrade is not ok, I am out of ideas13:48
Wizzupwhat about some hacks with hildon-meta13:51
Wizzuphave gconf pkgs not depend on each other13:52
Wizzuphm I guess that is tricky for ci13:52
freemangordonmaybe that will work, dunno13:52
freemangordonyeah13:52
Wizzupor maybe merge gconf2 and gconf-common and add some provides/replaces13:55
Wizzupthen there might also not be a cycle13:55
freemangordonwould you try? as I am sick of gconf now :)13:56
freemangordonhmm, wait13:56
freemangordonmaybe we shall move triggers to another package13:57
freemangordonhave to run, bbl13:58
Wizzupthat works too @ triggers14:09
Wizzupfreemangordon: can gconf pre-depend on sgml-base?14:26
freemangordonsgml-base is not the issue14:34
freemangordonit is gconf that is the issue14:34
WizzupI think it's both, no?14:34
freemangordonnot sure14:34
freemangordoncould be14:34
Wizzupok, well, I liked the idea of moving the triggers to a separate pkg'14:35
freemangordonthat package shall pre-depend on gconf or something14:36
freemangordonI am not sure how dependency chain should look like14:36
Wizzupgconf can depends on gconf-triggers and gconf-triggers can pre-depend on gconf, no?14:37
freemangordonsounds weird but we can try14:38
Wizzupthis way gconf can configure without running triggers and without issue14:38
freemangordontry it14:40
Wizzupok, will today14:45
arno11Wizzup: btw for n900, did you move all stuff from devel to daedalus ?15:20
arno11i ask because the last chimaera img is finally unusable apart for basic stuff, and not upgradeable to devel because of random network/modem crash and h-d freeze :(15:26
WizzupI haven't observed these crashes or freezes15:28
arno11i mean from a fresh install15:28
Wizzupbut we can look at n900 specific stuff when the daedalus install works on some devices15:28
arno11i mean it is not possible to use it from a fresh install with chimaera15:29
arno11so ofc from our old install, daedalus works. different story for new user15:30
arno11when you say you didn t observe these crashes or freezes, you mean from a fresh install ?15:32
Wizzupno, I haven't dd an image in a bit15:40
arno11Wizzup: ok. even dts fixes are missing with 125MHz freq still there btw15:55
WizzupI am not super inclined to work backwards on chimaera and rather just make a daedalus image that works well15:55
Wizzupbut if you tell me exactly what to build I can do it15:55
arno11oh, ofc no problem for daedalus as a priority. i just mean we absolutely need all devel stuff for stable img15:56
arno11for chimaera or daedalus15:57
Wizzupdaedalus built from devel stuff for the most part15:57
arno11ok that's nice15:57
freemangordonWizzup: https://bugs.launchpad.net/ubuntu/+source/gconf/+bug/155162316:35
freemangordonWizzup: going to test the fix16:37
arno11Wizzup: last d4 build failed with same submodule error (as n900)17:26
Wizzupfreemangordon: do we really want to fork dpkg too? I'd prefer to try my fix17:32
Wizzuparno11: ok, ty, I'll try to see what's up17:33
freemangordonWizzup: no dpkg fork17:33
freemangordonbut change gconf trigger types17:33
Wizzupok17:33
freemangordonWizzup: I force-pushed, so keep you local changes for a while17:34
Wizzupyeah, I did, I also have the local changes you had17:34
Wizzupjust in case17:34
freemangordonso far it seems this change helps, however, we may have to upgrade gconf in chimaera too17:34
Wizzupsounds like a good idea (let's do -devEL first)17:35
Wizzup-devel17:35
freemangordonI'll do locally forst17:35
freemangordon*first17:35
freemangordonand then -devel17:35
freemangordonin either case, there will be a need for dpkg-reconfigure a, most-probably17:36
Wizzupregardless of a fix? strange17:36
freemangordonyes, because we already have gconf with broken triggers in chimaera17:36
Wizzupif we replace gconf it doesn't seem like it would be a problem17:37
freemangordonwill have to try17:38
freemangordonlemme first check if it will allow dist-upgrade to not break17:38
freemangordonstarting dist-upgarde with new gconf installed, lets see17:42
freemangordonWizzup: seems it will work18:02
freemangordonI think we shall issue a chimaera HAM metapackage that depends on that gconf18:02
freemangordonto  give users chance to upgrade18:03
freemangordondist-upgrade still goes on though, will report when its done18:04
Wizzupok18:21
freemangordonWizzup: done, no errors18:22
freemangordonso, the plan is:18:22
freemangordon1. put fixed gconf from daedalus to chimaera-devel and check if it installs (with apt upgrade) with no issue, then test dist-upgarde. I'll do that18:23
freemangordon2. move gconf to chimaera18:23
freemangordon3. check if image builds still work (both chimaera and daedalus)18:24
freemangordon4. add new gconf as a dependency to chimaera hildon-meta and mark hildon-meta as HAM system package18:25
freemangordonwhat do you think?18:25
Wizzupsounds good18:28
Wizzuphildon-meta is already a system package in ham for experimental btw18:28
Wizzupre: 5, we might also want to make some hildon-meta-pin pkg that gets update monthly with specific versions18:30
freemangordonthat was my idea from yesterday, to auto-generate such package, somehow18:45
freemangordonright before I asked if we can hire repo maintainer :D18:46
freemangordonWizzup: hmm, maybe I shall put it in experimental, if metapackage is there18:47
freemangordonhm?18:47
Wizzupeither way18:50
freemangordonok, doing for experimental first18:51
freemangordonWizzup: hmm, HAM lists "core meta package" too19:06
Wizzupyeah I did that19:09
Wizzupwe can remove it though19:09
freemangordonok19:11
freemangordonwe shall, it does not have a nice icon :)19:11
freemangordonI will do it19:11
freemangordonhmm, conflicts with libgconf2-dev (3.2.6-7)19:15
Wizzupstrange, why19:16
freemangordonno idea19:16
Wizzupmaybe you apt-get instlal'd something that depends on this specific version19:17
Wizzupwould be strange19:17
Wizzupor ham is crazy :)19:17
freemangordonHAM19:17
freemangordonwill try one more thing19:17
freemangordonhmm, why is core package "not provided by maemo leste"19:18
freemangordonoh19:18
freemangordonI guess I have to install hildon-application-manager-settings-standard first, right?19:19
freemangordonWizzup: ^^^?19:19
freemangordonWizzup: did you mark our packages as system?19:21
freemangordonWizzup: somehow the upgrade is not trusted by HAM, tha's why it refuses to install it19:24
freemangordonWizzup: what is the key of experimental?19:27
freemangordonWizzup: oh, HAM uses its own gpg store19:36
freemangordonsomeone shall read the docs :)19:40
Wizzupfreemangordon: ah yes, I need to do that still20:27
Wizzupfreemangordon: I probably didn't have that because I had it in red pill mode20:27
Wizzupbut yes there's the domains thing20:27
freemangordonwe have to fix domains/keys in order to use HAM20:42
WizzupI know21:10

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!