libera/#maemo-leste/ Tuesday, 2021-08-31

konaSweet, USB kb works on my pinephone00:12
konaFor some reason I thought that wasn't working yet for Maemo Leste00:13
Wizzupkona: :)01:05
Wizzupfreemangordon: shouldn't this be NULL? https://github.com/maemo-leste/dsme/commit/d866deb122101cffec00620947843fea1396568f#diff-4ef413819eba6cc91cb6f776025a84426e68ae2d4c965fa0a1a0f7275f4f37d8R36001:10
kona also, i may have ordered an n900.01:14
Wizzupkona: cool01:15
Wizzupit's not ideal as dev env, but it's a good (and supported) device01:15
konahopefully i can score a pinephone kb when they put them up for order/preorder too :)01:16
konaI have my vm for dev, or I saw that we have images for pi[123] and i have a couple of those in my parts crate01:17
enyckona: do you know about making sure the micro-usb is firmly soldered down and not to get ripped off board, such a common problem ;-(01:18
konacurrently trying to understand tny-vfs-stream.[ch] so I can port it from gnomevfs to gvfs.01:19
konaenyc, for the n900? i think i heard something about this.01:19
enyckona: yes01:19
enycwell known issue, most common ''failure'' of them imho01:20
konayeah, i think i (just barely) have the necessary "lack of skill" to reattach that.01:20
Wizzupkona: something you can also do is look for commits where we ported from gvfs to gio01:20
konaand solder wick to clean up my mess. :)01:20
konawould that probably be better? it seems like gio is an abstraction layer above gvfs, so maybe?01:21
Wizzupgio replaces gvfs afaik01:23
Wizzuphttps://github.com/maemo-leste/osso-pdf-viewer/commit/610b9012f95de2bcc90ebaa3176dfabf1ba300d801:23
Wizzuphttps://github.com/maemo-leste/libcomapp/commit/ada20f434909f8bf2cb28dc0f3234e136e73400601:23
Wizzupsome random one, not from us https://github.com/maemo-leste/evolution-data-server/commit/2622a74884ff4670d3ec6ba003293649efeaf63101:23
Wizzupnot sure if it's helpful01:23
Wizzuphttps://github.com/maemo-leste/hildon-application-manager/pull/101:24
konaWizzup, those links are helpful, thanks!03:21
freemangordonWizzup: tho others are inited with 0, see https://github.com/maemo-leste/dsme/commit/d866deb122101cffec00620947843fea1396568f#diff-4ef413819eba6cc91cb6f776025a84426e68ae2d4c965fa0a1a0f7275f4f37d8R35507:36
freemangordon*the others07:36
freemangordonI am just following the style07:36
freemangordonWizzup: anything else?07:59
freemangordon(I got a power outage here, might have missed something in the logs)07:59
freemangordonparazyd: is this https://github.com/maemo-leste/dsme/commit/ae0a385af685249cc878e7666450790169d270b2 released?09:24
parazydIt should be09:32
freemangordonthis is a mess :(09:33
freemangordonno changelog, eh?09:33
freemangordonalso, seems somehow stop() is missing from the init script, I still didn;t find why09:34
freemangordonwhy is https://github.com/maemo-leste/dsme/commit/9b49e8047a73a38d40258b5692b11843accdb915#diff-d89e0e8d06d3ced1f6f4f4c0710a4d71f2d2b123da2c54a3e952a31358399b2cL36 removed?09:35
freemangordonany idea?09:35
freemangordonparazyd: ^^^09:35
parazydOpenRC has a default stop() function09:38
parazydSo that's redundant09:38
freemangordonok09:38
parazydWhat's a mess?09:40
freemangordonno changelog for the release09:40
freemangordonso you can;t really tell if a commit is included or not09:41
freemangordonalso, tag is on previous commit09:41
freemangordonsee https://github.com/maemo-leste/dsme/releases/tag/0.61.709:41
parazydYeah we don't usually make tags when stuff in debian/ is changed.09:41
freemangordonhmm, I was under the impression we are tagging debian/changelog change commit09:42
freemangordonbecause basically this is what makes a relese, no?09:43
freemangordonhmm, wait09:43
freemangordonthe tag is on debian/changelog commit, which is right09:43
parazydWell the way I see it, the HEAD of beowulf-devel branch should be released in devel09:44
freemangordonand then there is one more commit09:44
parazydThe HEAD of beowulf branch should be released in stable09:44
parazydAnd development should happen on master09:44
freemangordonok, but IIRC you release the latest tag, no?09:44
freemangordonor I forgot how that all was supposed to work09:44
parazydYes, but tags serve to tag the source code itself, unrelated to the debian directory.09:45
freemangordonhmm09:45
freemangordonkinda makes sense, though sounds weird to me09:45
freemangordonbut ok, I'll keep it in mind09:46
parazydhttps://leste.maemo.org/Development/Building_Packages#The_git_tag09:46
parazydSee N.B here09:46
freemangordonany objections if I tag debian/changelog commit?09:46
parazydIt's fine if you do09:46
freemangordonok09:46
freemangordonparazyd: wait, this is a mess :D09:47
freemangordonyou say that changelog version is used to checkout the tag to build, right?09:47
parazydYes, but the debian directory is always taken from the branch's HEAD09:48
freemangordonbut in dsme tag 0.61.7 is *before* commit ae0a385af685249cc878e7666450790169d270b209:48
freemangordonso, how is this commit included?09:48
freemangordonae0a385af685249cc878e7666450790169d270b2 I mean09:48
parazydBecause that's inside debian/09:48
freemangordonoh, sorry09:48
parazydSo as I wrote already, it gets picked up09:48
freemangordonyeah, init script is in debian09:49
parazydmhm09:49
freemangordonhmm, doesn;t sound clean to me09:49
parazydIt is also possible to maintain them outside09:49
parazyduvos does this in one or two packages09:49
freemangordoninit script is not part of debian packaging09:49
freemangordonbut ok, now it is clear what happened09:49
freemangordonthanks!09:50
parazydYou're welcome09:50
lelfreemangordon opened a pull request: https://github.com/maemo-leste/dsme/pull/5 (Graceful exit)09:50
lelIMbackK synchronize a pull request: https://github.com/maemo-leste/hildon-desktop/pull/14 (add support for rotating xinput touchscreens when entering or leaving portrait mode)11:16
lelfreemangordon closed a pull request: https://github.com/maemo-leste/hildon-desktop/pull/14 (add support for rotating xinput touchscreens when entering or leaving portrait mode)11:34
Wizzupfreemangordon: we do not have stop in dsme because we don't want to ever restart it at runtime afaik11:35
Wizzupfreemangordon: so stop should be a stub11:35
freemangordonok11:36
freemangordonWizzup: please review dsme pr when you have some spare time11:36
Wizzupok11:36
freemangordonBTW, are those PRs still valid https://github.com/maemo-leste/hildon-desktop/pulls ?11:38
Wizzupparazyd: freemangordon: I think we don't want openrc to be able to stop dsme11:46
parazydI agree11:47
parazydStopping dsme, even on shutdown, could lead to weird behaviour11:47
Wizzuphm, wait, I think it does maybe need to do that on shutdown, that's a good point..11:47
parazydYou need to make sure it's the last in line somehow11:47
Wizzupwell, maybe we don't worry about that now11:48
bencohdon't we have dependencies for that?11:48
Wizzupit seems to 'work'11:48
Wizzuplet's just fix the srv restart problem first11:48
parazydbencoh: not really split for start/stop11:48
bencoh?11:48
parazydYou have dependencies in general, but I'm not sure you can set dependencies for the stop() function11:49
bencohah11:49
parazydSo from the top of my head, I can't see how one would make dsme be the last initscript to stop11:50
parazydWell, not _last_ last, but after everything Maemo11:50
bencohRequired-Stop sounds like what we would want though (dunno how it translates to with openrc)11:50
bencoh(dunno if it works either to be honest)11:51
Wizzupin any case this is not actually a practical/current problem11:53
Wizzupdsme does watchdog handoff now I think, so we don't have the resets anymore11:53
Wizzup(when it exits)11:53
freemangordonWizzup: this is already fixed11:59
freemangordonhttps://github.com/maemo-leste/dsme/commit/2d4db8f5e14bc0f131ee5402915188b01968f8e012:00
freemangordonthat's why we should have stop() that does nothing12:00
freemangordonIIUC12:00
Wizzupah, ok, yeah12:05
WizzupI thought the commit you linked showed it as removed12:05
Wizzuphttps://github.com/maemo-leste/dsme/blob/master/debian/dsme-thermal.init I don't see a stop() stub here12:05
Wizzupoh that's thermal12:06
Wizzupfreemangordon: whatever https://github.com/maemo-leste/dsme/commit/2d4db8f5e14bc0f131ee5402915188b01968f8e0 did before, it's gone now in master12:06
Wizzupoh, ne12:06
Wizzupok12:06
WizzupI just need to wake up12:07
Wizzupthe stop() stub is gone though12:07
freemangordonyes, and I asked why it is gone on which parazyd replied that openrc has default stop() so this is redundant. I only hope the default one does what removed stop() does12:15
Wizzupwhat is removed is not what the default stop does afaik, but I don't know for sure12:16
freemangordonif default does something else, then we shall restore the removed stop(), I did it like that on purpose back then and I see no reasoning in the commit that removed it.12:18
freemangordonparazyd: ^^^ any comment?12:19
parazydAll good12:19
freemangordonyou're fine with restoring the removed stop()?12:19
freemangordonok, I'll update the PR12:20
parazydYeah I'd say it's fine12:21
lelfreemangordon synchronize a pull request: https://github.com/maemo-leste/dsme/pull/5 (Graceful exit)12:27
uvosfreemangordon: https://github.com/maemo-leste/hildon-desktop/pull/13 is obsolte per say, but please do something along the same lines by rerunging some tool on the code base and then closing the pr. (since h-d is sutch a mess code formating wise)14:59
uvosfreemangordon: https://github.com/maemo-leste/hildon-desktop/pull/10 works fine for me and i find it usefull, if its something you want is up to you15:00
uvosfreemangordon: https://github.com/maemo-leste/hildon-desktop/pull/7: mce currently provides this via the old interface, but its bad imo. mce has to implement an interface that just mirrors the iio-sensor-proxy one under a different dbus name to provide a sensor interface just for this (and mce provides no other external interface to other sensors anywhere makeing this module a really odd duckling). The way this causes the rotation15:05
uvospolicy to be split currently is bad too. i consider the interface mce provides to be deprecated and would request slowly moving stuff to use iio-sensor-proxy directly, which provides equivalent dbus interfaces. mce currently warns when the if is used. However this pr is just a proof of concept and would require some more error checking and sutch to be really in a mergable state.15:05
uvossidenote lifeguard should check for a gid not a uid to assign permissions15:21
uvosreally a group name but a gid would be less offensive allready15:22
freemangordonuvos: why gid and not uid? you have one and only one interactive user under maemo16:03
freemangordonlets not discuss if it is good or bad16:04
freemangordon(only one user)16:04
freemangordonuvos: are yuo going to review https://github.com/maemo-leste/dsme/pull/5 ?16:19
freemangordon*you16:19
inky_so, people16:25
inky_i have compiled megapixels (camera app for pinephone) under maemo16:25
inky_and it sort of works.16:25
inky_but you know it doesn't work well under manjaro as well.16:25
inky_so i guess that's the best we can have now.16:26
Wizzupis it pinephone specific?16:26
inky_may be with devel kernel it'll be better.16:26
inky_yes can you imagine that?16:26
inky_the only camera for pinephone today is pinephone specific.16:26
inky_because underlying layers don't support pinephone hardware well.16:26
inky_i got already photos made with it. very close to what you get under manjaro. some low saturation shots.16:27
inky_it actually creates several dng files with each shot.16:27
inky_and then converts to tiff with dcraw.16:27
inky_the redraw is very slow. but i remember it was also slow under manjaro16:27
inky_may be not as slow16:28
inky_and pinephone under maemo has this problem - sometimes the part with controls of the camera app disappears and shows the controls of terminal app, if it is running.16:28
inky_then you press it it comes back.16:28
inky_such redraw issues are common on pinephone + maemo, but absent on pinephone + something else.16:29
inky_it is not really really useful now, but it is not really really really useful on other distros as well.16:29
inky_so i don't know if i should try to package it.16:29
inky_may be it just bit worse on maemo.16:30
inky_also when the screen blanks, i cannot bring it back. i can only kill the app via ssh.16:30
Wizzupmight still make sense to package it somewhere if you made any changes16:30
Wizzupyes, 3d problems on the pinephone are frequent with the lima driver16:31
inky_do we use other driver than others?16:32
Wizzupwe use X11+glamor16:32
inky_they use wayland, and it has a better driver?16:32
Wizzupothers use wayland16:32
Wizzupyes, basically16:32
inky_is the work being done to improve x11 somewhere? or we are the only ones interested?16:33
inky_is it in mesa?16:33
inky_i have some problems on pinebook pro and they say it is in mesa, not the kernel driver.16:33
inky_i don't really understand what is glamor. i understand we have userspace drivers in xorg that talk to kernel drivers.16:34
inky_let's say nouveau in xorg talks to nouveau in kernel.16:34
inky_and mesa has the opengl implementation for each hardware.16:34
WizzupI think people work on it, but they don't work on the specific problems we have. It's either in mesa or glamor, which uses mesa16:35
Wizzupcould be partially kernel driver too16:35
inky_i think i'll package it. yes, in order to build it, i had to change its ninja.build a bit. and at runtime it demands ~/Pictures, may be I should add its creation to the source code? don't know.16:35
inky_and report issues to the author.16:35
Wizzupdoes it really require that and not some XDG var?16:35
Wizzupshould be XDG_PICTURES_DIR16:36
inky_it creates /tmp/megapixels.randomstring and stores dng files there.16:36
inky_then it tries to copy them, and in console i see the debug output of command 'cp' that tries to copy photos to ~/Pictures and failes.16:36
inky_once I created the directory, it works.16:36
tmlindsicelo, uvos: fyi pushed out wlroots changes so far to https://github.com/tmlind/wlroots/commits/pvr-gles2, at least weston-simple-egl issue remains caused by commit 4e07d4cbf9c116:36
inky_let me try with exporting XDG_PICTURES_DIR16:36
inky_i confirm, it doesn't care about XDG_PICTURES_DIR16:43
inky_i run it with this environment, and it tries to save to ~/Pictures. If Pictures doesn't exist, it fails to cp to it.16:43
uvosfreemangordon: "uvos: are yuo going to review https://github.com/maemo-leste/dsme/pull/5 ?me but a gid would be less offensive allready" if you like17:23
freemangordonuvos: sorry, I don;t understand how gid/uid is related to this PR17:24
freemangordoncould you please elaborate?17:25
uvosfreemangordon: "uvos: why gid and not uid? you have one and only one interactive user under maemo" no i cant, and its a perfectly valid reason, we need to get away from this "only one user" thing with hardcoded user name and id if we ever intend to integrate with other distros (the main impediment being that this makes it impossible for hildon to be a regular session) or even sutch niceties as having your own user name on deivce for17:26
uvosssh etc.17:26
uvosthere is no technical reason to replace the hardcoded uid's and user names with groups as we encounter them.17:26
uvos*there is no technical reason to NOT replace the hardcoded uid's and user names with groups as we encounter them.17:27
freemangordonuvos: but, is this related to the PR in question?17:27
uvosno its a side note: this needs improvement17:27
freemangordonah, ok17:27
freemangordonbut, if we strictly discuss the PR as such, do you have comments/notes on it? Shall I improve/fix anything?17:28
uvosi only skimmed it  so far, i can read it in ditail later if you like17:28
uvosit looks fine at first glance at least17:29
WizzupI think it's pretty fine17:29
freemangordonyes, I would prefer more eyes to look through code changes done in such a critical component17:29
freemangordonWizzup: I think it is pretty fine too, but lets have uvos look at it as well17:30
freemangordon:)17:30
Wizzupsure17:30
konaweechat is workable on pinephone19:42
konauvos: i tried trojita with x11 vkbd, got it configured but then i never see the password prompt.  i'll check the issue list and log a new issue if necessary?19:48
uvostrojita only promts for a pw when 1. you dident specify the pw in settings 2. the conection to the server was sucessfull 3. an operation that requires the pw is initiated19:53
uvoskona: are you sure that all of these are true? if the server is missconifgured you wont ever see the promt for instance19:54
uvoskona: maybe try trojita on a linux desktop if you can and see if it works there19:54
konaes, was trying to not save password :)19:54
uvoskona: if it dosent on desktop, reporting upstream makes the most sense i think, as my changes to it are very minimal and only relate to the ui.19:55
konaserver logs seem to say yes but auth hangs.19:55
uvoskona: deleating the pw on my device makes it prompt fine here19:56
konacool, maybe this is more pinephone oddity19:56
uvosvery unlikely19:56
uvosit might not render the prompt due to mesa issues but hildon should at least show you that it got a new window19:57
uvosofc i assume you are starting trojita via its .dekstop file19:58
uvosnot via xterm via different env19:58
konavia .desktop, yes.19:59
uvostry menu->view->debugging->showprotocol log20:00
uvosand see if it even gets to the pw prompt with the server20:00
uvos(restart after activating that)20:00
konalast message: connCheckingcertificates (SSL)20:09
konaso i will sniff the packets, sorry for buggin :)20:10
uvosif it stays like that then thats wierd and a bug it should timeout and tell you your sever is missconfigured20:12
uvosbut i would direct the bug upstream20:12
konak20:16
lelfreemangordon closed a pull request: https://github.com/maemo-leste/dsme/pull/5 (Graceful exit)22:36
sicelotmlind: thanks @wlroots! haven't had time to play with this stuff lately, but intend to pick it up again, soon23:42

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