libera/#maemo-leste/ Thursday, 2022-12-29

moparisthebestis there an XMPP client that's recommended for running on Maemo ?07:10
Wizzupmoparisthebest: so we're working on finishing parts of conversations that allow for group chats and the like (probably not yet media), so that'd probably be recommended11:25
Wizzupmoparisthebest: what device do you have?11:48
norayrmoparisthebest: i use pidgin and dino.12:46
norayri like using pidgin because i like seeing different windows, and updates in those (in phosh that doesn't work, btw), and it is possible to write armenian in gtk2 apps with maemo onscreen keyboard.12:47
norayralso pidgin takes very little battery, and doesn't do heavy i/o.12:47
norayrdino draws battery and does heavy i/o, it is gtk3, so no maemo keyboard easy way, but it has http_upload, it has omemo encryption.12:48
norayri wish it would be possible to build dino from handy branch, it has audio/video calls. and it can redraw itself for small screens, by hiding one of the panes, left or right.12:50
norayrit is good to have conversations, of course.12:50
norayri was very much dependent on this way of xmpp chats, both in maemo and in sailfish.12:51
Wizzupnorayr: as fmg said, we might have hildon input method soon for gtk312:54
norayrgood. didn't notice this. usually i read everyhing, but not these couple of days.12:56
norayrwe have a big problem with kbd, maybe i need to fork it and work on it myself.12:56
Wizzupwhat is this 'big problem' ?12:57
norayrWizzup: let me start from the beginning. maemo kbd gives us utf-8 codes, not xkeycodes. before my contribution of the header file, we were not able to write in other languages in any place. today it is possible to write in any place in armenian, but i need to do 'setxkbmap am' before that. by using new uvos' backend, of course, not old.13:14
norayrwhich is also, btw, hard, because if i use a comma inside armenian text, it won't draw, i need to switch by doing setxkbmap us, then write a comma.13:15
norayrmaybe the whole keyboard thing should be rewritten, let's say on one of my devices i use 'onboard' which works with xorg and types in any input field.13:16
norayrthat would break old maemo things, but i think we can give up old maemo work of input at all. having 2 different backends for the keyboard means people need to be taught that we have 2 different backends.13:17
norayrthey don't understand why by tapping they get one backend and by pressing a key they get another backend, and why the kbd behaves differently.13:17
norayrthis is not consistent and makes people go away from maemo.13:17
norayrthe input is one of the most common interactions between humans and machines.13:18
norayrit should be very convenient.13:18
norayrand believe me i cannot convince anyone, not only my mother or wife, but any of my nerdy friends to do setxkbmap each time they change the kbd layout.13:19
norayrso what i think to do, i can create a map with available languages, and xkb maps. and at the onscreen kbd language change, do a corresponding setxkbmap automatically.13:19
norayri even researched how to do that in C, programmatically, via functions.13:19
norayrbut that won't help if the user needs to type comma, or semicolon inside armenian or russian or whatever text.13:20
norayrthey need to switch to english, type a comma, and then back. :/13:20
norayrand still that would mean using the second backend. one of two, and educating the users, again. nobody wants this kind of 'education' since everybody intuitively understand that computers are made to increase the efficiency of working, to make life easier, not harder.13:21
norayrso maybe quick fix is a map and doing setxkbmap programmatically, and real fix is rewriting the whole kbd stuff, and deciding to not have 2 backends, and educating users, but one backend. and removing hidlonization part of input from gtk at qt.13:23
norayrthe input is what makes my hands to search something else than maemo device. it is a torture to switch between languages even for a comma.13:24
norayrand i am ready to be 'educated' and use it in this complicated way.13:24
norayrthis is the commit i am talking about: https://github.com/maemo-leste/hildon-input-method/commit/5dcba543c36c12ac70288a7b652310e275eebee713:28
norayrnorayr == inky13:28
norayrthe file i generated is hildon-im-xcode-keysyms.h13:29
norayrbtw that's how i did it: https://github.com/norayr/gen-header-for-hildon-input-method13:29
norayrbut that doesn't matter, that would deviate from the main discussion.13:30
Wizzupnorayr: it might be good to capture this in an issue13:33
norayrthank you, i'll try.13:33
Wizzupuvos: this phone says xt910 in android but xt912 on the back...13:37
Wizzupall three of these actually13:37
buZzi think i'm getting a Cosmo Communicator on loan in a while14:00
buZztouchscreen glass is cracked i think, but works14:01
uvosWizzup: thats a bit strange, but all spiders run essentaly the same android firmware so its not suprizing that works14:44
Wizzupuvos: they were sold as xt910 though14:47
Wizzupso I suppose they are all xt912, I don't know what xt910 says on the back14:47
uvosnot sure what you mean on the back14:47
siceloheh, here's a fun one - at least on my droid 4, i can't type numbers when entering a new contact in addressbook using hwkb14:47
uvosmy xt910 dosent say anything on the back14:47
siceloin the cellular number field, that is14:49
Wizzupuvos: yeah just to identify it I mean14:51
Wizzupsicelo: hm, please file a bug :)14:51
Wizzupthis might be relatively easy to fix, but sounds odd, I think I did this without problems14:51
uvosnorayr: we will allways need 2 backends14:52
uvosunless we improve the x11 backend (with atspi maybe?) to the point we dont need the toolkit specific ones14:52
uvosotherwise input into stuff like xterm, random nontoolkit/ niche toolking applicaitons wont work14:53
norayrthat's what i am saying: improving x11 backend as an ulimate goal. short term but bad solution is creating a map (maemo input language to xkb map). and doing setxkbmap automatically.14:54
norayrwhich won't solve changing kbd laaayout when typing commas but it would short term make everything much more usable.14:55
uvoscreating a him xkb map or having him create ones dinamicly is the only way it will work14:55
uvossure it solves typing commas etc14:55
uvosa map can have 248 chars14:55
uvosthats enough for 2 languages and special chars14:56
norayrno no i believe it doesn't. i have to do setxkbmap us to be able to type commas.:/14:56
uvosbecause the arminan map dosent have the right code14:56
norayrcannot write ascii with unicode layout.14:56
uvosthats not how this works14:56
uvosits neither acsii nor unicode14:57
norayrwhich one has?14:57
uvosits xkeysims14:57
uvoshim just needs to create/have a map that is switches to when it opens that contains all the chars it can type in the user selected languages14:57
uvosthat contains a xkeycode for eatch xkeysym it wants to type14:57
norayri think no xkb map has ascii symbols14:57
uvosagain its not acsii at all14:58
uvosbut yes you have to create a map (dynamicly or otherwise) just for him14:58
uvosand not use one of the default ones14:58
norayror use existing map as a shooooort term solution.14:58
norayrby mapping him's am layout to setxkbmap am14:59
norayrand him's ru rayout to setxkbmap ru14:59
uvosno, because as you correctly state the xkb maps dont line up with the keys him wants to type14:59
norayrthat won't solve commas but at reast ih is something.14:59
norayrleast14:59
sicelo and sphone (abook?) doesn't show caller's name if saved in E.164 format. maybe operator only sends local part of number? it should be somehow possible for sphone/abook to match the contact nonetheless15:08
* sicelo is using his work sim on droid 4 for a while15:08
uvossphone uses eds functions to lookup the phone number in the address book15:09
uvosif eds supports converting between regions then it should work if eds just dose a strcmp then thats how it works.15:10
uvosso dunno15:10
uvoseds has specific functions to match phone numbers so i would guess it dose understand at least 00 + and local notation15:11
moparisthebestWizzup: pinephone here, is that code anywhere yet?15:42
Wizzupyup, although some of the work in progress branches need some cleaning up before dsc_ can/will merge them15:43
moparisthebestnorayr: latest Dino with calls is gtk415:44
moparisthebestI tried gajim which worked except I couldn't focus the input box to type messages :(15:46
Wizzupso the way to go in the near future is absolutely going to be conversations, but I'm surprised you can't focus with gajim15:59
Wizzupit's gtk2, right?15:59
uvoshim pretty mutch breaks any application that focuses fields itself16:00
uvosthis affects pdigen too16:00
moparisthebestI'm guessing gajim is gtk3 but I'm not sure right now, I'll check16:02
moparisthebestIt's interesting I can hit tab and a name will be put in the input box, and I can hit enter to send it which works, but I can't focus or type in it16:03
moparisthebestI do all my phone calls over XMPP too so that'll be hard to go without16:05
freemangordonsicelo: I don't know what sphone uses to show caller's name, but I can assure you abook has fuzzy number matching16:11
freemangordonwhether it works, is another story though, I am not sure it was tested thoroughly after REing16:12
freemangordonuvos: what 'eds functions' are used?16:12
freemangordonsicelo: http://maemo.org/api_refs/5.0/beta/libosso-abook/libosso-abook-osso-abook-util.html#osso-abook-query-phone-number16:14
freemangordonhttps://github.com/maemo-leste/osso-abook/blob/a731589ac51e2059b667ab8d8c84b41b0befab03/lib/osso-abook-aggregator.c#L232716:22
uvosfreemangordon: https://github.com/maemo-leste/sphone/blob/eacc386bf976e999a0c2a0d0a536cebe5b7f788d/src/modules/contacts-evolution.c#L5716:35
uvosthis is the query used16:35
uvosE_BOOK_QUERY_EQUALS_SHORT_PHONE_NUMBER internally uses e_phone_number_compare_strings16:35
uvosso yes fuzzy matching16:36
uvosnot sure why abook reinvents the wheel here16:36
freemangordonIIUC because: 1. it normalizes the phone number before doing the matching. 2. It uses last 7 digits to do the match.17:55
freemangordonI would always trust Nokia when it comes to something telephony17:55
uvosfreemangordon: this is exactly what eds dose uuic18:34
uvosthe last 7 digits is the SHORT of the query type18:34
uvosso iiuc these functions are the same18:35
uvosi would not trust nokia in anything18:35
freemangordonwhy is that? what they did to you?18:38
uvosi mean nokia implementations of things in maemo are honestly pretty uniformly bad18:39
siceloheh. anyway, just to be sure - d4 needed some hack for call audio. is that in leste kernel or not?18:39
freemangordonnot abook18:39
uvosmaybe not abook, i havent looked at it18:39
freemangordona side note - on chimaera chromium tries to use GPU rendering and fails strangely18:39
freemangordonon d4 that is18:39
uvoswell so dose qwebengine18:40
uvosso that might be related18:40
uvosit failes to compile a shader18:40
freemangordonPVR:(Error): LoadCompilerModule: Couldn't load library libglslcompiler.so [0, ]18:40
freemangordonyes18:40
freemangordondo we have binaries with hacks removed?18:40
uvosti git18:40
freemangordonumm...18:41
uvoshmm?18:41
freemangordonwhat is ti git here?18:42
freemangordonhttps://github.com/maemo-leste/sgx-ddk-um/tree/maemo/chimaera18:42
freemangordon"This branch is up to date with master."18:42
freemangordonso, the binaries we use on chimaera are hacked ones18:43
freemangordonI would bet that's the reason for the failures18:43
freemangordonhave to run now, will check that tomorrow18:43
uvosnot sure why the hacks would cause that18:43
uvosits just weakening a symbol18:44
freemangordonwell...18:44
uvoswell what?18:44
freemangordonone of the libs that was 'weakened' is libglslcompiler.so18:44
freemangordoniirc18:44
uvossure18:44
uvosbut i mean how would that cause an isue i mean18:44
uvossure we should remove it anyhow18:44
freemangordonno idea18:44
freemangordonyes, will try tomorrow18:44
freemangordonttyl18:44
uvosti git is upstream18:45
uvoshttps://git.ti.com/cgit/graphics/omap5-sgx-ddk-um-linux/refs/heads18:46
freemangordonI know, but that's not what we distribute18:46
uvossure no18:46
uvosim awnsering the question "what is ti git here?"18:47
freemangordonok18:47
siceloWizzup: i would like to upstream spinal's patch - https://github.com/maemo-leste/n9xx-linux/commit/fbc4ce7a84e59215914a8981afe918002b191493, unless you can do it19:13
siceloif i do it, please remind me how to send someone's patch so it has proper attribution. you sent my lp5523 patch and it does credit me in the git history19:14
uvossicelo: git am?20:24
uvosor while genrateing the email?20:24
uvosgit format-patch has an option there20:24
Wizzupsicelo: please do, maybe Authored-by or Suggested-by or so?20:25
sicelocool, will do. i wonder if spinal can still be contacted via mail maybe. perhaps i should start with that20:28
siceloheh, he's just responded to the gh issue i opened against charge-mode :-p22:06
siceloemail sent22:15
sicelo(to spinal, that is)22:22

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