| inky | apply for the repo (: | 00:32 |
|---|---|---|
| tmlind | freemangordon: yup making them serdev_device_drivers would clarify things quite a bit | 04:47 |
| freemangordon | tmlind: ok, I think I am sorting this out. What I don;t understand is why tty/serdev (no ngsm, ordinary stuff) are allocating serdev controller per port. we shall have only one, right, not one per serdev device? | 08:10 |
| dsc__ | hmm yes, I can see how this vkb Qt stuff was a bit complicated xD | 11:57 |
| dsc__ | there are 2 issues I have right now with the vkb Qt module | 16:18 |
| dsc__ | him vkb is misbehaving by stealing focus, its not how virtual keyboards should work, am pretty sure | 16:18 |
| dsc__ | the other is interpreting incoming xcb events | 16:18 |
| Wizzup | how does it 'steal' focus, by bringing up the keyboard? | 16:28 |
| dsc__ | yeah, I'm not sure what mechanism takes the focus | 16:30 |
| dsc__ | https://github.com/community-ssu/qt-x11-maemo/blob/b3282763fc0a95b7f6eddc7b8caf7406003975ac/src/gui/inputmethod/qhildoninputcontext_x11.cpp#L492-L495 | 16:31 |
| dsc__ | this poor Nokia engineer had the same issue | 16:31 |
| dsc__ | dont think I can replicate his hack due to changes in Qt | 16:31 |
| dsc__ | but I can certainly write a different hack ^^ | 16:32 |
| dsc__ | the correct way: https://github.com/valve-project/steam-qt-keyboard-plugin/blob/master/src/steaminputcontext.cpp#L48-L66 | 16:32 |
| dsc__ | cant use hideInputPanel() because it triggers as soon as focus is stolen | 16:33 |
| freemangordon | dsc__: I am not sure this is really a bug, as you can have 2 types of VKB | 16:45 |
| freemangordon | like, the one in android (for example) is not fullscreen, so it shoudl not steal the focus | 16:46 |
| freemangordon | however, HIM VKB *is* fullscreen, with its own edit field | 16:46 |
| freemangordon | so, it makes all the sense for it to steal the focus | 16:47 |
| freemangordon | also, VKB already hides itself when clicked outside, there is no need for you to hide it once again | 16:48 |
| freemangordon | I can bet hideInputPanel() is for cases when VKB behaves like android one | 16:48 |
| dsc__ | oki makes sense | 16:49 |
| freemangordon | maybe, just maybe VKB shall take care to return the same focus as before stealing | 16:50 |
| dsc__ | this is quite awkward though, as vkb sends Qt events, and nothing is in focus, so first need to ensure the widget is in focus again | 16:50 |
| dsc__ | at which point a focus event triggers, but vkb is already opened, etc | 16:50 |
| dsc__ | as vkb sends events to Qt* | 16:50 |
| freemangordon | wait, why would vkb send qt events before it is closed? | 16:51 |
| dsc__ | this was my understanding, but maybe I am misinterpreting how it works | 16:51 |
| freemangordon | the whole point (iiuc) is to do whatever has to be done in vkb edit field and return the end result to qt | 16:51 |
| dsc__ | makes sense | 16:51 |
| freemangordon | you should not send any keys or whatever to qt before vkb is done, unless I am missing somethoing | 16:52 |
| freemangordon | Wizzup: ^^^ | 16:52 |
| dsc__ | well the reason why I think that | 16:52 |
| dsc__ | sec | 16:52 |
| * freemangordon has to go to do some shopping, bbl in 15 minutes | 16:53 | |
| dsc__ | user presses 'space' on the keyboard: https://github.com/community-ssu/hildon-input-method-plugins/blob/b83654e5cfb1dbccd90926295c282f0465c35a16/hildon-im-fkb/hildon-im-western-fkb.c#L2809 | 16:54 |
| dsc__ | it goes to here: | 16:54 |
| dsc__ | https://github.com/maemo-leste/hildon-input-method/blob/master/src/hildon-im-ui.c#L2112 | 16:54 |
| dsc__ | which does hildon_im_ui_send_event() | 16:54 |
| dsc__ | this event goes through the underlying application, so it seems it does send events while the vkb is active | 17:04 |
| dsc__ | goes to the* | 17:04 |
| dsc__ | cant type anymore :D | 17:04 |
| dsc__ | and also, it would make sense for vkb to constantly emit these events as its not just keys, it also needs to communicate key modifiers and such - it would be strange to send those in bulk at the end of the vkb operation... maybe | 17:12 |
| dsc__ | bbl | 17:13 |
| freemangordon | dsc__: seems to depend on commit mode | 17:20 |
| freemangordon | also, this sends event to him framework | 17:22 |
| freemangordon | I guess this is done to enable plugin to filter some keys or dunno | 17:24 |
| freemangordon | or, could be related to word prediction | 17:26 |
| freemangordon | tmlind: oh, there is a comment in the code "/* Only a single slave device is currently supported. */". that explains it why we have a controller per port | 17:38 |
| tmlind | freemangordon: oh ok | 17:50 |
| freemangordon | I am about to test something, just wonder how would DTS for child nodes should look like | 17:52 |
| tmlind | reg = <1> for dlci1 | 17:53 |
| tmlind | channel@1 { reg = <1>; ); | 17:53 |
| freemangordon | who is going to read that? current cares about compatible only | 17:54 |
| freemangordon | but anyway, lets get to that | 17:54 |
| freemangordon | tmlind: getting there :D https://pastebin.com/XmqpSYHL | 17:58 |
| tmlind | ok great | 18:13 |
| freemangordon | not really, is I don;t see why it would error, but meh | 18:13 |
| freemangordon | tmlind: ok, of_serdev_register_devices() is doing for_each_available_child_of_node() | 18:36 |
| freemangordon | tmlind: getting there :) https://pastebin.com/T66DdMMB | 20:15 |
| freemangordon | tmlind: https://pastebin.com/zi0jUFxs YAY! | 21:15 |
| freemangordon | using serdev | 21:15 |
| inky | also him rotates the screen when whitespace is pressed | 21:53 |
| inky | i have to disable him to prevent that. | 21:53 |
| inky | it happens with all onscreen keyboards. | 21:54 |
| Wizzup | freemangordon: sweet :) | 21:55 |
| sicelo | inky: just tried on my droid 4, from both landscape and portrait orientations. no such problem. or what are the specific steps to reproduce? | 21:56 |
| inky | install any onscreen x11 keyboard | 21:56 |
| inky | remember diejuse? | 21:57 |
| sicelo | you said him | 21:57 |
| inky | he had exactly the same issue with bboard or something | 21:57 |
| inky | yes when you use any x11 onscreen keyboard and press whitespace | 21:58 |
| inky | him rotates the screen | 21:58 |
| inky | the only way to prevent it and use x11 onscreen keyboard in disabling him from startup | 21:58 |
| sicelo | you mean *both* him *and* the additional keyboard become active at the same time? | 21:58 |
| inky | yes | 21:59 |
| inky | well no | 21:59 |
| inky | vkb is not visible | 21:59 |
| inky | just your arbitrary onscreen keyboard | 21:59 |
| inky | you press whitespace and screen rotates | 21:59 |
| sicelo | so where is him when the additional vkb is there? how do you know him rotates the screen if it's not even there? | 21:59 |
| inky | i disable him, reboot, and no probrem. without him service running everythiwg works good. | 22:00 |
| inky | > so where is him when the additional vkb is there? how do you know him rotates the screen if it's not even there? | 22:00 |
| inky | because if i remove him then screen does not rotate | 22:00 |
| sicelo | anyway, i think leste is not planning to change keyboards anytime soon | 22:01 |
| inky | well i am not using vkb or him because of that. | 22:02 |
| inky | every time i install leste on a device first hhing to bring it to usable state is to disable him and use other onscreen keyboard. | 22:02 |
| sicelo | what the problem with him? | 22:03 |
| inky | i dont know where to start | 22:04 |
| sicelo | nearly no one has reported problems :-) | 22:04 |
| mkf | inky how many devices do you have running leste | 22:05 |
| sicelo | and thousands have used it on Fremantle without problems either. i think everyone agrees that it has its flaws, but that device is unusable... | 22:06 |
| inky | i complained here and everywhere so many times. thet i helped uvos to create a header file that partially solved some problems | 22:06 |
| inky | ok let me come to the computer and explain | 22:06 |
| inky | sicelo, you don't notice the problems because you: | 22:17 |
| inky | * use hardware keyboard | 22:18 |
| inky | * don't use non latin characters. | 22:18 |
| inky | now imagine you use pinephone. | 22:18 |
| inky | or anything without hardware keyboard. | 22:18 |
| inky | https://github.com/maemo-leste/hildon-input-method/commit/5dcba543c36c12ac70288a7b652310e275eebee7 | 22:18 |
| inky | this commit mentions me. | 22:18 |
| inky | since that it is possible to type in lets say regular xterm (not osso xterm) | 22:19 |
| inky | but | 22:19 |
| inky | * only if you can trigger the keyboard and | 22:19 |
| inky | * only if you can switch the layout which requires typing setxkbmap by hand in other terminal. | 22:19 |
| inky | on pinephone you cannot trigger the keyboard by pressing a hardware button like "search" button on d4. | 22:20 |
| inky | so no way to type almost everywhere. | 22:20 |
| inky | i can use leafpad, type there, but then no way to select and copy paste to other place. | 22:20 |
| inky | you have no hwkbd to do ctrl+c | 22:20 |
| inky | you just can type in leafpad | 22:20 |
| inky | in any other regular linux app you need to be able to trigger keyboard, it needs to appear so that you would typee | 22:21 |
| inky | but you have no hardwarke key to trigger it. and if you tap it won't come, because it only comes if you tap on gtk2-3, not everywhere. | 22:21 |
| sicelo | i guess this - 23:18 < inky> * don't use non latin characters. ... | 22:22 |
| inky | but in gtk2, lets say in leafpad you cannot then select the text and then copy paste it to other place, no ctrl+c | 22:22 |
| inky | other problem is that in order to type non latin, you have to write by hand setxkbmap in osso-xterm | 22:22 |
| sicelo | hardware keyboard aint it, because we do use it with keyboard closed, and Librem5 does not have hardware keyboard. i didn't have a problem on it either | 22:22 |
| inky | then only in other place non latin will work. if you able to trigger keyboard which you cannot do because you have no hardware key mapped. | 22:23 |
| inky | how do you trigger keyboard on librem5? | 22:23 |
| inky | is there a hardware button for it? | 22:23 |
| sicelo | it just pops up when input field is focused | 22:23 |
| inky | in which browser? | 22:23 |
| sicelo | i didn't test browsers | 22:24 |
| inky | you cannot write anything in any browser | 22:24 |
| inky | and you cannot copy paste to the browser input field. | 22:24 |
| inky | so blogging is out of question. using webmail is out of question. web anything. | 22:24 |
| * sicelo doesn't use Qt if he can help it ... and yes, does recall that the leste browser is Qt | 22:25 | |
| sicelo | but the Qt problem is already being worked on ... so that's soon not going to be a problem either, i believe | 22:25 |
| inky | it's not a qt problem. if you want to tap in firefox, it's not qt | 22:26 |
| inky | tap in the address bar, it doesnt bring you the keyboard. | 22:26 |
| freemangordon | inky: browsers do not bring vkb up because they don;t follow the standards | 22:26 |
| inky | well i use other keyboard, i solved my problems. | 22:27 |
| freemangordon | it is really browser issue, not leste or vkb | 22:27 |
| freemangordon | sure | 22:27 |
| inky | but other users cannot solve them. | 22:27 |
| freemangordon | still, it makes sense someone to raise bugs against browsers | 22:27 |
| inky | i think the best thing i can do for others is to package the keyboard i use and explain how to use it in the wiki | 22:27 |
| inky | and also then my package i guess should disable him. and uninstalling it should enable him back. | 22:28 |
| inky | it's not only browsers. xterm also doesnt feel the tap. | 22:28 |
| freemangordon | no way | 22:28 |
| inky | not osso xterm. regular one. | 22:28 |
| freemangordon | ah, xterm | 22:28 |
| inky | yes. | 22:28 |
| freemangordon | anyway, I am not going to argue that HIM is the best option around | 22:29 |
| inky | freemangordon: i am extremely thankful to you for all your support and all what you did with him. i can use him with dino or some other programs. | 22:29 |
| inky | i just wanted to say to sicelo that it has problems, and on pinephone if you use non latin keyboard you just cannot survive because as i explained no way to trigger it, and necessity to write setxkbmap in osso xterm which not many would be ready to do. | 22:31 |
| freemangordon | yeah, we have to write that gesture handler that brings vkb up someday | 22:32 |
| inky | oh that would be amazing! | 22:32 |
| inky | just for you to know, i mentioned it, but for you to know, there is also a bug which, when you write in vkb, sometimes non latin characters are not copied to the input field of underlying program. | 22:33 |
| inky | let's say you type in vkbd and what you typed has to appear in leafpad or our email program or anything supported | 22:33 |
| inky | but then only "," or "." symbols are there | 22:34 |
| inky | instead of other non latin symbols you get whitespaces. | 22:34 |
| inky | it happens rarely, but it happens. | 22:34 |
| inky | i have no idea how can you repeat it. it's not often. | 22:34 |
| inky | freemangordon: i am also very thankful that you enabled unicode in xmpp accounts. | 22:47 |
| inky | my account is unicode so thanks no you i can use conversations. | 22:47 |
| dsc__ | firefox is able to raise a vkb, but only on wayland it seems | 23:55 |
| dsc__ | going to guess its because wayland standardized the way virtual keyboards are suppose to work | 23:55 |
| dsc__ | people have been asking for firefox vkb support since 2010 on their tracker, i doubt x11 will get it | 23:56 |
| dsc__ | s/tracker/bugtracker/g | 23:56 |
| dsc__ | just a guess @ wayland, I dunno much about wayland internals :p | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!