| arno11 | inky: also got it with proot | 00:45 |
|---|---|---|
| arno11 | iirc the culprit is hildon-input-method | 00:45 |
| inky | ohhh... | 02:44 |
| inky | my friend wrote very good keyboard, and added rotation support. so it rotates together with the screen, adjusts its size. so in portrait mode it is very convenient to use onscreen keyboard now. but yes, whitespace rotates back to landscape. then another whitespace rotates to portrait. :/ | 02:50 |
| inky | maybe this is here... https://github.com/maemo-leste/hildon-desktop/blob/e79fbd450eae55e832feb2f1eb481a146ee5143e/src/launcher/hd-app-mgr.c#L1408 | 02:55 |
| sicelo | interesting that whitespace rotates orientation :-D | 09:15 |
| sicelo | inky: and no, i don't think it's that line. that seems to be checking for a space before some command, so it can remove the spaces before running the command. | 09:16 |
| sicelo | at least that's what i think is happening there, with my limited C-fu | 09:16 |
| * enyc meows sicelo | 10:24 | |
| enyc | Wizzup: AIUI in practice zram seems more helpful/used than zswap, but things like n900 might now be considered a niche case | 10:25 |
| enyc | POP!_OS includes some zram setup heuristic out the box that apparently works well | 10:26 |
| enyc | not sure Ubuntu/Mint/Debian/MX/Devuan do at all though; tried to suggest it ........ | 10:27 |
| uvos | inky: i dont think i quite grok what you are reporting | 11:20 |
| uvos | so in your cusom vkb pressing space causes the device to rotate? | 11:20 |
| uvos | i would suggest that maybe something is causing a different window to be (momentarly) focused on space | 11:21 |
| uvos | h-d would then check the protraitable flag of that window | 11:21 |
| uvos | find it missing and rotate to landscape | 11:21 |
| uvos | maybe try setting the "allways allow portrait" setting in transition.ini to check if its that | 11:21 |
| uvos | otherwise maybe describe whats happening in more detail | 11:22 |
| inky | uvos__: so diejuse also experiences it, and i even found it in the logs, h e reported it in this chat in february 18. though he uses onboard. | 12:28 |
| inky | ok, interesting, i tried with wired keyboard, and hardware wired keyboard does not rotate the screen. but onscreen keyboards do. will write to the author again. | 12:40 |
| arno11 | could someone explain me what is the role of hildon-desktop-wait in xsession.post plz ? | 13:21 |
| arno11 | (can't find relevant info in logs) | 13:21 |
| uvos | arno11: h-d sends a signal hat it is ready (im not sure what the criteria for ready is) this xsession script halts untill the signal is recived | 13:25 |
| uvos | arno11: this is so that stuff that requires h-d to be up can be started for sure after its up | 13:26 |
| uvos | arno11: this is kindof a dependancy management hack maemo dose because it lacks a session manager with real depedancy managment. | 13:26 |
| arno11 | ok ty | 13:28 |
| arno11 | i wonder if it could be the culprit for random boot issues: maybe sometimes hd is never 'ready' (?) | 13:32 |
| uvos | the only reason for that to be that i can think of is oom | 13:34 |
| uvos | without some kind of log output during a failed boot i dont think i makes sense to speculate more | 13:35 |
| Wizzup | inky: maemo detects attached keyboards | 13:35 |
| Wizzup | and then prevents rotation, I think | 13:35 |
| uvos | yes | 13:35 |
| uvos | but i think the real issue is | 13:35 |
| Wizzup | if you want osd to not rotate, or to rotate, you need to either set certain X11 atoms, or work with us to see if there's an issue in h-d where it special cases hildon keyboard | 13:35 |
| Wizzup | I think we wanted to look at the latter anyway for qtvirtualkeyboard | 13:36 |
| Wizzup | s/osd/osk/ or whatever :) | 13:36 |
| uvos | that when he clicks on the osd hildon sees a window raised that that dosent support portait | 13:36 |
| uvos | he should be able to check that by setting transiition.ini to ingore those atoms | 13:36 |
| dsc_ | hopefuly can commit Matrix stuff in the next few days | 13:46 |
| dsc_ | it was a bit complicated | 13:46 |
| dsc_ | :-P | 13:46 |
| arno11 | cool | 13:48 |
| dsc_ | I'm now an expert on Matrix and federation | 13:52 |
| dsc_ | rip | 13:52 |
| inky | dsc_: i use xmpp. i even join this irc via xmpp gateway. xmpp is much easier to host. and it is a democratically governed protocol. yes, i wish maemo had some representation on fediverse. | 13:54 |
| Wizzup | we're also making progress on omemo | 13:55 |
| Wizzup | for xmpp | 13:55 |
| inky | in transision.ini i don't see any "allways allow portrait" like setting. i see forcerotation = 0 (also i tried it with 1, but didn't help). | 13:57 |
| inky | i also added those apps to whitelist, but still pressing whitespace with x11 keyboard, not our maemo keyboard, rotates. | 13:57 |
| inky | hm, i see osso-xtrem is in blacklist, but it rotates happily. | 13:58 |
| dsc_ | inky: I also prefer xmpp, at least from an protocol/implementation standpoint | 13:59 |
| dsc_ | Matrix clients are hard to code, and you see that in the existing clients | 14:00 |
| Wizzup | inky: can you make a bug report with this rotation issue so that we can discuss / keep track there? | 14:04 |
| inky | so okay, i am using coolkbd (diejuse reported the same with onboard). | 14:04 |
| inky | that's an onscreen keyboard developed by my friend. he even added maemo docked mode. | 14:04 |
| inky | his keyboard has 3 modes. it can cover the screen from up, from down, and it is able to resize active window and fit under it (i think he described it as maemo toolbar mode). | 14:04 |
| inky | i connected that keyboard to the power button with xbindkeys, so when i press the power button on pinephone, the keyboard appears, then it disappears (gets killed). | 14:04 |
| inky | the biggest problem is that in landspace mode, if the onscreen keyboard is used/sized comfortably, there is not enough space for the app in the upper part of the screen. | 14:04 |
| inky | so portrait mode is indeed more convenient for using onscreen keyboard. right now our keyboard takes 1/3 of the screen in portrait mode and is very convenient. | 14:04 |
| inky | i rotate dino, or other apps, and i use them in portrait mode. the only problem is that pressing whitespace triggers screen rotation. | 14:04 |
| Wizzup | pls also link the keyboard in question and the exact problem | 14:04 |
| inky | Wizzup: i'll open a ticket right now. i was searching for diejuse's issue to continue there but didn't find. still he has the same issue with onboard keyboard. | 14:04 |
| sicelo | dsc_: yes, matrix protocol is huge | 14:18 |
| inky | https://github.com/maemo-leste/bugtracker/issues/742 | 14:25 |
| dsc_ | sicelo: also, on the server side of things, I heard it has performance issues with the reference server daemon written in Python (?) | 14:25 |
| dsc_ | anyway mostly growing pains | 14:26 |
| dsc_ | this 'libquotient' library I have been using is high quality | 14:26 |
| inky | yes, correct. synapse is written in python and is very heavy. also matrix (according to protocol) syncs all the images, all the data from the known other nodes of the network. | 14:27 |
| inky | so my friend who runs it, just disabled federation. | 14:27 |
| inky | because his server space gets exhausted in days otherwise. | 14:28 |
| inky | i think he mentioned about 50g per day back then. | 14:28 |
| dsc_ | \o/ | 14:28 |
| dsc_ | oh god, its written in Twisted | 14:29 |
| dsc_ | the predecessor to asyncio | 14:29 |
| Wizzup | dsc_: don't demotivate yourself xP | 14:30 |
| inky | Wizzup: did you see my github issue link? | 14:30 |
| dsc_ | Wizzup: ever done stuff in Twisted? | 14:30 |
| dsc_ | :P | 14:31 |
| inky | dsc_: i demotivated myself long ago. i am unable to use matrix. but i am always on xmpp. i don't have anything else. | 14:31 |
| dsc_ | im not demotivated I'm just motivated to fork Matrix, remove 70% of the features and write a daemon in c++ | 14:31 |
| dsc_ | but maybe in another life :P | 14:31 |
| inky | my biggest concern is that the protocol is governed by a private corp. not by community, like xmpp. | 14:32 |
| dsc_ | yep | 14:32 |
| dsc_ | well I dont really care about that tbh | 14:33 |
| uvos__ | matrix syncs all the content incl videos/images for channels a user of that matrix server is registered on | 14:34 |
| dsc_ | but I can see how some people care | 14:34 |
| uvos__ | so it only grows huge if your users are in channels that send alot of media | 14:34 |
| uvos__ | btw | 14:34 |
| uvos__ | so if you run a server with a freind you can avoid it mostly | 14:34 |
| Wizzup | inky: I did see it, yes, I can't look into detail now | 14:37 |
| inky | yay, good. | 14:38 |
| uvos__ | inky: if you have forcerotation = 1 and it forces to landscape on space thats pretty wierd behavior | 14:39 |
| inky | yes, it also forces with forcerotation 1 or 0, doesn't matter. | 14:40 |
| inky | and with orientation lock applet enabled. | 14:40 |
| uvos__ | dose space do anything else? | 14:40 |
| inky | and i checked that: | 14:40 |
| inky | $ gconftool-2 --get /apps/osso/hildon-desktop/ui_can_rotate false $ gconftool-2 --get /apps/osso/hildon-desktop/orientation_lock true | 14:40 |
| uvos__ | like a popup like android dose? | 14:40 |
| uvos__ | inky: as for the window being to small in landscape thats a hard to solve problem | 14:40 |
| inky | i noticed that when i press the left upper button, and see all the minimized apps, and then press space, they all disappear and i see the wallpaper screen, like if i pressed the left upper button once again. | 14:41 |
| uvos__ | android solves this by not allways makeing the window smaller but instead by paning the window to the input field | 14:41 |
| uvos__ | but that only works beacuse andoird allways knows what input field is selected | 14:41 |
| uvos__ | we dont beacuse it depends on the toolking | 14:41 |
| uvos__ | (altho with at-spi you could alllways know too) | 14:41 |
| uvos__ | inky: ok wierd | 14:41 |
| inky | this keyboard is very indeed useful for me. i even want to package it. | 14:42 |
| uvos__ | could you run xev | 14:42 |
| uvos__ | and record what events it gets when you type something (incl space) | 14:42 |
| inky | i can use it in portrait, and it types in everything. it also works on regular linux/x11. | 14:42 |
| inky | minute, i am running xev | 14:42 |
| uvos__ | sure i looks mutch better than him in general | 14:42 |
| uvos__ | him is kinda, not so great | 14:42 |
| inky | xev when opened, forces rotation to landscape. then when i press whitespace: https://bpa.st/ZJOQ | 14:44 |
| uvos__ | if you have forcerotation = 1 | 14:44 |
| uvos__ | you should be able to start with xev in portrait | 14:44 |
| inky | minute | 14:44 |
| inky | yes it started in portrait. | 14:45 |
| inky | should i type in xev window or in other window? | 14:45 |
| uvos__ | into xev | 14:45 |
| uvos__ | im looking to see if it sees any focus out events | 14:45 |
| uvos__ | during text input | 14:45 |
| inky | this is whitespace now: https://bpa.st/4MUA | 14:45 |
| inky | yeah, no focus out. | 14:46 |
| uvos__ | hmm | 14:46 |
| inky | let me try other button | 14:46 |
| uvos__ | but no keypress either | 14:46 |
| uvos__ | it sends a change to the xcb map (strange why) but no keypress | 14:46 |
| uvos__ | that dosent make any sense | 14:47 |
| inky | this is pressing button a https://bpa.st/RNPQ | 14:47 |
| inky | i think it tries to change the map to add the button to the current layout because it also has cyrillic and armenian layouts. | 14:47 |
| uvos__ | the whitespace trace makes no sense | 14:47 |
| uvos__ | could you do that again but | 14:47 |
| uvos__ | show the fous in event | 14:48 |
| uvos__ | when you switch to xev | 14:48 |
| uvos__ | then type some text | 14:48 |
| uvos__ | then space | 14:48 |
| uvos__ | then type some text again? | 14:48 |
| inky | i got keypress and keyrelease now with whitespace | 14:48 |
| inky | https://bpa.st/7SJA | 14:48 |
| inky | now i'll try as you say | 14:48 |
| uvos__ | i think there should at least alos be a window resize event | 14:49 |
| uvos__ | when rotation happens | 14:49 |
| uvos__ | (ofc start in portrait) | 14:49 |
| inky | https://bpa.st/ZUTA | 14:49 |
| inky | btw it doesn't rotate now for some reason. in xev. | 14:49 |
| inky | but if i press on something else it'll rotate. | 14:50 |
| inky | i hope xev will register the event, if i press on something else. | 14:50 |
| uvos__ | dose it rotate in xterm (not osso-xterm, regular xterm)? | 14:50 |
| inky | let me try | 14:50 |
| inky | interesting, xterm is not forced to portrait when i even added it to transisions.ini whitelist. | 14:53 |
| inky | but i can try dino or leafpad | 14:53 |
| uvos__ | forcrotation = 1 should force it to portrait | 14:53 |
| inky | it is set... | 14:53 |
| uvos__ | something is wrong then | 14:53 |
| uvos__ | xterm can be protrait on my device with that set | 14:53 |
| uvos__ | are you restarting hildon-desktop? | 14:53 |
| inky | oooooh | 14:53 |
| inky | so i opened leafpad. | 14:53 |
| inky | typed a | 14:54 |
| inky | i didn't get any event in xev output | 14:54 |
| inky | then i pressed whitespace in leafpad and got looots of events | 14:54 |
| inky | let me give you | 14:54 |
| uvos__ | thats notmal xev only shows events sent to its window | 14:54 |
| uvos__ | oh thats not normal | 14:54 |
| uvos__ | unless its the xcb map event again | 14:54 |
| uvos__ | that is sent to every client | 14:54 |
| inky | https://bpa.st/BWGA | 14:56 |
| uvos__ | that looks like a normal rotation sequence (no input) | 14:57 |
| inky | interesting. | 14:58 |
| inky | i am trying onboard, diejuse said onboard behaves the same. | 14:58 |
| inky | https://bpa.st/WTEQ | 15:04 |
| uvos__ | inky: ok these traces are unfortionatly not of any use, it would seam its not what i was suspecting it might be | 15:28 |
| uvos__ | well i gues that dose make them of some use, they disproved that theory | 15:28 |
| inky | what was the theory? | 15:57 |
| inky | house m. d. was saying 'nothing means something'. | 15:57 |
| inky | sorry... 'nothing tells us something'. | 15:57 |
| uvos__ | inky: dose "xdotool key a space b" | 16:11 |
| uvos__ | trigger portait rotation? | 16:11 |
| uvos__ | when done via ssh wih export DSIPLAY=:0.1 | 16:11 |
| uvos__ | *when done via ssh wih export DSIPLAY=:0.0 | 16:11 |
| uvos__ | inky: i was expecting a focus out event that suggests some other window is takeing over for some time | 16:12 |
| uvos__ | maybe even with a force landscape atom or no atoms etc | 16:12 |
| uvos__ | since that would cause h-d to reevalutate portrait mode | 16:12 |
| inky | let me try | 16:18 |
| inky | no, xdo tool types that 'a b' in the dino input field but does not rotate. | 16:18 |
| inky | it's hard to check with onboard, i only have to believe to diejuse and arno11, who said same is in their android proot. | 16:20 |
| inky | onboard is too big and goes to background for me. | 16:20 |
| inky | i will work with coolkbd author as well, to try to understand what is happening. | 16:20 |
| arno11 | inky: out of curiosity, is the issue still there if you remove (temporary) hildon-input-method from xsession ? | 17:14 |
| arno11 | Wizzup: btw i have cmtspeech ready for 8000Hz (but need new shm value as well) | 18:00 |
| Wizzup | right | 18:07 |
| Wizzup | I think we can push a leste-config with client and daemon changes, agreed? | 18:07 |
| arno11 | yep | 18:08 |
| arno11 | we already have daemon stuff btw | 18:08 |
| arno11 | *custom daemon | 18:09 |
| arno11 | i'll make a PR for cmtspeech this evening/night | 18:11 |
| inky | arno11: let me try. | 18:17 |
| arno11 | ok | 18:17 |
| inky | root@devuan-pinephone:/etc/X11# find . | grep -i hildon-input | 18:18 |
| inky | ./Xsession.post/25hildon-input-method-configurator | 18:18 |
| inky | ./replace/Xsession.post/hildon-input-method-configurator | 18:18 |
| inky | i guess the file that starts with 25? temporary move it somewhere? | 18:18 |
| arno11 | yep | 18:21 |
| arno11 | and reboot | 18:22 |
| arno11 | (on proot, it avoids weird behaviour with other keyboards) | 18:25 |
| inky | arno11: thank you! when i remove hildon-input-method from xsession, my keyboard works!!! | 19:36 |
| inky | THANK YOU. | 19:37 |
| inky | arno11, i don't remember what you also suggested for so that my status won't be... broken on boot? | 19:38 |
| inky | i need to document those changes. i don't know what did i do and why already. | 19:38 |
| arno11 | inky: no probs. the other stuff was hildon-status-menu on boot: the tweak is to start it later to avoid it to crash. but that's not a proper fix since nobody knows what's going on with it actually | 21:52 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!