libera/#maemo-leste/ Thursday, 2021-11-11

mighty17[m]freemangordon: weston 9 is fine? :P07:08
freemangordonwhat do you mean "fine"?07:11
mighty17[m]well will it work with your mesa?07:13
mighty17[m]freemangordon: weston 9 works with your mesa :o07:16
freemangordonah07:18
freemangordongood07:18
freemangordonso, there is some issue with sway?07:18
mighty17[m]sway? i've been using phosh xD07:19
freemangordonah, yeah07:19
mighty17[m]weston logs https://paste.debian.net/1219109/07:20
mighty17[m]freemangordon: actually, its using llvm07:21
mighty17[m]or how do i confirm that?07:21
mighty17[m]s/1219109/1219111/07:27
mighty17[m]freemangordon: looks like only weston works07:55
mighty17[m]can you give phosh a try?07:55
mighty17[m]or or I'll use xc-racer's mesa but apply your commits on top?08:00
mighty17[m]i suppose i dont need to leste changes as well08:28
freemangordonmighty17[m]: ok, I can give ti a try, but, care to link to a repo to use?08:37
freemangordonor, it should be pre-build?08:37
freemangordon"E: Unable to locate package phosh"08:38
mighty17[m]https://gitlab.gnome.org/World/Phosh/phosh source code08:39
mighty17[m]freemangordon: It's in debian08:40
freemangordonit is not08:40
freemangordonat least not in devuan we use08:40
mighty17[m]In sid and bullseye08:40
freemangordongood, but we are on older distro, iiuc08:41
freemangordonanyway, will look at it when I have time08:42
mighty17[m]Sure, till then I'll try getting your commits on xc racer's tree08:43
mighty17[m]Idk which are important tho, probably not the rebases and leste ones for me08:43
freemangordonI think it will be easier to merge  xc racer's commits08:44
freemangordonbecause my changes are based not his tree08:44
freemangordonso there should be only few commits missing08:44
mighty17[m]I know, but what if some commit breaks stuff for me, that's why I'd probably cherry pick08:51
mighty17[m]freemangordon: basically all commits by you since `Oct 29, 2021`09:46
uvoshi10:14
uvoslooks like i got baned again on libera chat10:15
uvos<bencoh> funnily enough, pvr_dri points to nouveau_vieux.so btw" that cant be right10:17
uvosbuild system must be confused somehow10:17
Wizzupuvos: ugh @ ban10:17
bencohuvos: well, feel free to check http://muarf.org/~bencoh/maemo/leste/mesa-pvr-ti/11:08
bencohmay I ask why the Status tables on the n900/droid4 pages were renamed to "old status"?11:27
bencohis there a new per-device status table?11:27
Wizzupthe old status can go basically11:28
Wizzupthe new status is on the top right of the page11:28
bencohthe "Software features"?11:29
Wizzupyeah11:29
Wizzupprobably the old status can go13:56
uvosnew sphone: storage plugin api implemented via rtcom and used for call history (not sms/messages yet) by gui, various small improvements14:51
uvosit uses local-uid as /sphone/$(BACKEND)/%(LINE_ID)14:53
uvosit absoulty needs to know the backend a call/msg was made on14:53
uvosbut im open to alternative routes for storing this14:54
uvosnot sure what the HEADERS are for/should contain so sphone ignores this, Wizzup if you figure this out pleas do tell14:54
dsc__inside the VM, with Dorian, when you click this button: https://i.imgur.com/ED3wWiP.png14:54
Wizzupwhat is line-id?14:55
dsc__screen will go all-black, and only option is to reset14:55
dsc__afterwards; every time you open Dorian, it will go all-black14:55
Wizzupdsc__: hm sounds like your VM has trouble with rotated screens14:55
dsc__unless you remove /home/user/.config/Pipacs/Dorian.conf14:55
dsc__ah14:55
Wizzupdsc__: let me try it in my vm momentarily14:55
uvosWizzup: sphones internal name for phone number or user name or whatever the $(BACKEND) needs to connect to this person.14:56
Wizzupwhat phone number is that? the person that calls you?14:56
uvossame as remote-uid i rtcom (and sphone sets it here to)14:56
uvosyes the id of the remote party14:56
Wizzupthat's wrong14:56
Wizzupsqlite> select distinct local_uid from events;14:56
Wizzupring/tel/ring14:56
Wizzupsofiasip/sip/merlijnwajer_40ekiga_2enet014:56
Wizzupspirit/skype/wizzy_2e_2e014:56
uvosin backend dependant form14:56
Wizzupremote_uid is the other party14:57
Wizzuplocal_uid is just you14:57
uvosok14:57
uvosso where shal i shove the backend?14:57
Wizzupthis should be apparement from fremantle db14:58
Wizzupapparent14:58
Wizzuptypically it goes in remote_uid, but I cannot comment on the exact form14:58
Wizzupit looks like for tp-ring the remote_uid is just the number that calls you, for sip it's the sip:foo number that calls you14:58
Wizzupand for irc etc it's the irc nick that chats (not sure how that works in groups atm)14:59
Wizzupall of this can be set up on fremantle with ease fwiw14:59
uvoswell problem is this is all very ill defined14:59
Wizzup(and yeah I've been overloaded with real life work so didn't get much further yet :-))14:59
uvosseams like they just have backends randomly insert different strings14:59
Wizzupfrom my notes:14:59
Wizzup    local_uid TEXT, -- plugin local uid ('account' on a plugin/protocol)14:59
Wizzup    local_name TEXT, -- seems to own irc nickname, etc <SelfHandle> for telepathy-ring14:59
uvosso local_uid should just be sphone/something then?15:00
Wizzup    remote_uid TEXT,15:00
Wizzup    channel TEXT,  -- channel for an event15:00
Wizzupyes, but this must also be registered as service I think15:00
uvosno15:00
uvosservcies are seperate15:00
Wizzupright, yeah15:00
Wizzupjust checked with sql15:00
uvosso how about  local_uid: sphone/sphone remote_uid: %(BACKEND):(%LINE_ID), local_name: unix username15:01
Wizzupwell, I gave you an example for telepathy-ring, the remote_uid is just the phone number15:02
WizzupI don't see the point of the unix username, but for telepathy-ring the local_name is not too relevant anyway15:02
Wizzup(in empathy it's also '<SelfHandle>')15:02
uvosWizzup: right except the example is ill defined for modem calls its just the number and for sip is the protcoll (sip) and then the number15:04
uvossphone dosent/ dosent want to treat ofono calls as special somehow15:05
WizzupI don't understand15:06
uvosif i place just the number into remote-uid and have an entry for +494548712545715:07
uvoshow is sphone supposed to know this was a telegram call and calling back on this event shal use that backend instead of ofono15:08
Wizzupthe local_uid indicates the plugin used15:08
Wizzupthere's also the event type15:08
Wizzupthe event type is normal call, or sip call, etc, afaik15:08
Wizzupso that way you could know15:09
uvosno15:09
uvosthose dont exist in the plugins we have atleast15:09
uvosyou could define those sure15:09
Wizzup'those'?15:09
uvosmore event types15:09
uvos"the local_uid indicates the plugin used" no it dosent15:09
Wizzupsqlite> select * from eventtypes;15:09
Wizzup1|RTCOM_EL_EVENTTYPE_CALL|1|Call15:09
Wizzup2|RTCOM_EL_EVENTTYPE_CALL_MISSED|1|Missed call15:09
Wizzup3|RTCOM_EL_EVENTTYPE_CALL_VOICEMAIL|1|Voicemail message15:09
uvosits just a phone number15:09
Wizzupuvos: no, local_uid is not a phone number15:09
Wizzuplocal_uid is ring/tel/ring for example15:09
Wizzupremote_uid is the phone number15:10
uvoser right15:10
uvosso have local_uid as sphone/$(BACKEND)15:10
uvosand remote_uid as just the line_id15:11
Wizzupthat's how fremantle does it it looks like15:11
Wizzupthere might be more to it and in particular the chat types and stuff I haven't grasped fully15:11
uvoswell thats not terribly important rn15:11
uvosfor sphone at least15:12
Wizzupexample for sip calls:15:12
Wizzup33771|1|1|1629194398|1629194396|0|0|1|0|0|0|sofiasip/sip/sip_2exs4all_2enl0||sip:+3161xxxxxxx@sip.xs4all.nl|||15:12
uvosanyhow the functions to translate sphones internal reprisentation -> rtcom and back are here in this plugin: https://github.com/maemo-leste/sphone/blob/f1881b09c414835f5ed061c2b7a751ee4c14932b/src/modules/store-rtcom.c#L12915:14
uvosrtcom_el_iter_get_values is fun15:15
uvosbecause the values of the entrys just randomly change (not defined anywhere) names15:15
WizzupI don't understand what you mean15:17
uvosim just harping on the api15:17
uvosyou RTCOM_EL_EVENT_SET_FIELD(ev, event_type,  g_strdup("RTCOM_EL_EVENTTYPE_CALL_MISSED"));15:18
uvos(with RTCOM_EL_EVENTTYPE_CALL_MISSED being a magic string)15:18
Wizzuplike I said, those need to be defined in some headers15:18
Wizzupparts of this were closed source, and we probably lack those headers15:18
uvosthen you  cant get event_type15:18
Wizzup?15:18
uvosbecause its event-type-id now and magicly became a int15:18
uvosbut if you add start_time it gets translated to start-time15:19
uvosand stays an int15:19
Wizzupof course the database uses ints15:19
Wizzupyou can probably set the "event-type-id" properly15:19
Wizzupproperty*15:19
Wizzup(I haven't checked, just a guess)15:19
uvosWizzup: right but the api just randomly translates some strings to ints and not others15:20
uvosand your suposed to just now15:20
uvosit also translates the field names15:20
uvosand your suposed to just know that the input field name free_text becomes free-text when you need the field again15:20
uvosits a mess15:20
WizzupI never used RTCOM_EL_EVENT_SET_FIELD15:20
WizzupI used the gobject properties15:20
Wizzupwhich seemed like how they want it to be used15:21
Wizzupbut again I cannot attack or defend the API15:21
uvosi kind of doubt that they have that macro and use it in the unit test beacuse they dont want it used but ok15:23
Wizzupuvos: the unit test might not be the best example of public api15:24
Wizzupdid you look at code of rtcom-eventlogger-ui as well?15:24
WizzupI don't know if it is any different15:24
uvosits N/A since it dosent add any events15:26
WizzupI think it had code to add events actually15:26
Wizzupwhich is commented in the main()15:26
Wizzupin the example15:26
Wizzupsee log_some_stuff15:27
uvosRTCOM_EL_EVENT_SET_FIELD15:27
Wizzuplook that uses the string api15:27
uvosso yeah15:27
Wizzuplooks like*15:27
Wizzupin any case what I still want to clear up how we can efficiently separate out different protocols15:27
WizzupI'd like to split sip/telegram/signal/ring/xmpp or at least filter by them15:28
Wizzupbut in the current plugins xmpp/telegram/signal are all the same rtcom plugin15:28
Wizzupand the same event types15:28
Wizzupso I guess there's just local_uid for that15:28
Wizzupbut the local_uid is a string and we'd have to match a part of the string or something, which is yuck15:28
uvosyeah right15:29
Wizzupsince they all use service_id RTCOM_EL_SERVICE_CHAT15:29
uvosthats what sphone dose rn15:29
uvosie match sphone/ofono15:29
uvosand call with ofono15:29
WizzupI don't want to do that really15:29
uvosright15:29
uvosalso same problem with RTCOM_EL_SERVICE_CALL15:29
Wizzupbut I will worry about this once the telepathy logging works15:29
Wizzupwrt call, you mean sip vs ring?15:29
uvosi assumed RTCOM_EL_SERVICE_CHAT was for messages15:30
Wizzupoh, good point15:30
uvosor telegram whatsapp whatever15:30
WizzupI'd have to check skype calls15:30
uvosnot just "sip"15:30
WizzupI can check that actually15:30
Wizzupsec15:30
Wizzupsqlite> select distinct event_type_id from events where local_uid='spirit/skype/wizzy_2e_2e0';15:31
Wizzup115:31
Wizzupand:15:31
Wizzupsqlite> select * from eventtypes where id=1;15:31
Wizzup415:31
Wizzup615:31
Wizzup1|RTCOM_EL_EVENTTYPE_CALL|1|Call15:31
Wizzupsqlite> select * from eventtypes where id=4;15:31
Wizzup4|RTCOM_EL_EVENTTYPE_CHAT_MESSAGE|2|Normal message15:31
Wizzupsqlite> select * from eventtypes where id=6;15:31
Wizzup6|RTCOM_EL_EVENTTYPE_CHAT_ACTION|2|Action message15:31
Wizzuplooks like you're correct15:31
Wizzupuvos: got any idea on how to solve above problem?15:48
Wizzupapart from string-matching local_uid15:48
Wizzupwe could make it different services with shared event types15:48
Wizzupalthough I am not sure if that breaks the model somehow15:48
Wizzupuvos: potentially flags.... somehow15:49
Wizzupor we could change the database model.15:49
Wizzupand provide a migration tool15:49
uvosdownside to different services is that you need a plugin for it15:50
uvosso everything needs to register a plugin for dubious benefit15:50
uvossure we could add a packend field that gets filled with a uuid15:50
Wizzupwell I don't see why all chat really fits in one plugin though15:50
uvosno idea what the flags are for15:50
Wizzupby that logic I think sms and chat should be the same plugin15:51
Wizzupuvos: https://dpaste.com/3PLPNGE3H15:51
uvosthe question is what is the point of the plugins even15:51
Wizzupthey offer helper functions for plugin specific code15:51
uvos"the plugins exist to offer helper functions to the plugins"15:52
uvosheh15:52
uvosno seriously i dont know what the benefit of having the plugins is over RTCOM_EL_SERVICE_CALL just being a enum15:52
Wizzupuvos: did you look at the plugin code?15:53
uvosi did15:53
Wizzupbtw, the remotes table might also be relevant somehow, but IDK who writes to it15:53
WizzupI think it maps local_uid, remote_uid to remote_name and osso abook id15:54
uvosthe plugins really just seam like a half implemented idea that dident go anywhere15:55
uvoslike they where supposed to have wrappers for adding new events in an easy way without the insane RTCOM_EL_EVENT api15:55
uvosand sutch15:55
uvosie abstract the database more, rn rtcom-eventlogger is a farily thin wrapper around the raw sql database15:56
Wizzupwell, we can do that15:57
uvosnot sure we want tho15:57
uvossince then you need to maintain a rtcom plugin for every protocoll15:57
Wizzupwhy is that?15:58
uvosseams not sufficantly benifical15:58
Wizzupok so what about another table protocol and give events a protocol_id ?15:58
uvossure15:58
uvoshow is protocol_id generated?15:58
uvosjust have the issuer generate a uuid?15:59
Wizzupwell I would suggest it gets inserted once and doesn't change - i.e. you can't delete it15:59
Wizzupno, I would suggest the protocol table is basically (1, 'telepathy-ring'), (2, 'telepathy-morse'), etc15:59
uvosok so you want to add a function to eventlogger that inserts a new protocol16:00
Wizzupyes, probably to the plugins ;)16:00
Wizzupsince we have no other way of adding things to the db16:00
uvosie int add_some_protocol(const char *name);16:00
Wizzup(upon init)16:00
WizzupI wouldn't suggest random applications add a protocol16:01
WizzupI would suggest that's a well controlled and defined list16:01
uvoshmm16:01
uvosnot sure its great that there is then a defined list of protocols you cant change as an app developer16:02
uvoslike i implement $(SHINY_NEW_PROTOCOL)16:02
uvosand then i cant log its events16:02
Wizzupand then you get it packaged for maemo/telepathy16:02
uvoswithout submitting a pr etc16:02
Wizzupand then we add it :)16:02
uvosyeah16:02
Wizzupas a dev you can do one INSERT INTO into db the for testing probably16:02
uvosbut thats not whats gona happen in practice16:02
uvosin practice the person with $(SHINY_NEW_PROTOCOL) is gonig to just go16:03
uvosoh my protocol is kinda like xmpp so lets just log events like that instead of this headace of geting a patch upstream somewhere16:03
WizzupI think the one detail you are omitting is that when someone implements a tp plugin16:04
Wizzupthey don't need to take care of the logging16:04
Wizzupwe have a program that logs tp channels to the db16:04
uvosthe world dosent nessecarly revolve around tp16:04
Wizzupand that is abstracted no matter the protocol16:04
Wizzupwell rtcom kinda does16:04
Wizzupat least in maemo fremantle16:04
uvosnot at the moment16:04
uvosnot really16:04
WizzupI haven't discovered anything in there that is not TP16:04
Wizzupwith ~10 years of usage and events16:05
uvosrtcom-eventlogger currently dosent care mutch about where the events come from16:05
uvoslets not fix it to tp16:05
Wizzupit is not, but our UI code will be16:05
Wizzupotherwise we will need to add random plugins for every protocol16:06
Wizzupto the UI as well16:06
Wizzupwhich is not scalable16:06
uvosno you makeing the assumption that your ui and stack will be the only thing running16:06
WizzupI am assuming that that is the maemo chat integration and other things that don't integrate won't use rtcom :P16:06
uvosbut people will not like you ui and want to run $(RANDOM_MATRIX_CLIENT) instead16:06
Wizzupyeah but I don't think that would log to rtcom16:06
uvosand maybe they want to hildonize $(RANDOM_MATRIX_CLIENT)16:07
uvoswe should allow it16:07
Wizzupit could, but I don't understand why they would16:07
Wizzupsince telepathy already does it16:07
Wizzup(telepathy-tank, iiuc)16:07
Wizzupsupporting libpurple next to tp on the other hand seems interesting to eventually try16:07
uvospoint is if you dont allow a ecosystem of $(RANDOM_MATRIX_CLIENTS) ported from elsewere to exist on our OS by forceing everyting through tp16:08
uvosthey wont16:08
Wizzupthey can exist, but I am not sure why they would log to rtcom?16:08
Wizzupthey're probably happier using whatever existing logging they have16:08
uvoswhy not? better integraion16:08
uvosmaybe so, but closing the door is unwise16:08
Wizzupslightly better integration, proper integration is in conversations :P16:08
WizzupI'm not really closing the door at all I think16:09
Wizzupit's an open platform and people can do what they want16:09
uvosconversations can not and will never fit all needs16:09
WizzupI was just trying to figure out a way to make the db schema more amendable to many different tp plugins16:09
uvosand this way16:09
uvosis bad16:10
Wizzupwhat's bad about it?16:10
uvoswe just discused that16:10
WizzupI don't see it16:10
uvos"adding a proctocoll requires a patch upstream"16:10
uvosto the table16:11
WizzupI would say proper integration requires working with upstream, which is common in FOSS16:11
Wizzupdoesn't stop anyone from just doing INSERT () from their application16:11
Wizzupbut sure it could a call that does the ~5 line insert code16:11
WizzupI think it's mostly a non-issue16:11
uvoswe definatly dont want people doing sql calls directly.16:12
uvosi think its very mutch an issue16:12
uvoscompeard to what you suggest i very mutch prefer fremantle status quo even16:13
Wizzuponly tp?16:13
Wizzup(that's fremantle status quo)16:13
Wizzupeven skype went through TP on fremantle16:14
Wizzupthey just didn't let you filter on telepathy plugin (protocol) in the UI16:14
Wizzupwhich is what this is mostly about16:14
Wizzupat least that's my understanding16:14
uvoswell they have to strcmp local_uid somewhere16:15
uvosto figure out what protocol to use16:15
uvosfor reply etc16:15
WizzupI don't think they need16:16
WizzupI don't think they need to*16:16
Wizzupconversations all revolves around contacts mostly16:16
Wizzupand for calls you get the option to use sip (accounts) or regular16:16
Wizzupbut yeah there must be some way they do that16:17
Wizzupin any case I'm more worried about storing plugin specific stuff through tp to begin with16:17
uvosok16:19
uvoswell youll figure something out16:19
Wizzuphopefully :P16:20
uvosill continue to do my thing, adding conversation a thread view to sphones sms-history is next16:20
uvosoh btw the dialer16:26
uvosdosent force portrait anymore16:26
Wizzup<316:26
uvosinstead the keypad is hidden16:27
WizzupI'll try it on my n900 momentarily :-p16:27
uvoswhen in landscape16:27
Wizzupdo you know what it looks like in fremantle btw?16:27
WizzupI can make a screenshot now if you want16:27
uvosyes i do16:28
Wizzupit's basically split in half16:28
uvosi know16:28
Wizzupright half is the dialer numbers16:28
Wizzupok ok16:28
Wizzup:)16:28
uvosi honsely i think its fine rn16:28
uvoswithout the numbers16:28
WizzupI'll check it out16:28
uvoshmm maybe on n900 its less fine16:29
uvos(no number hwkeys)16:29
Wizzupit does support that16:29
WizzupI just tried it16:30
Wizzupthe input field is set to a special input mode16:30
uvossupport what?16:30
Wizzupso 's' is automatically '+'16:30
uvosyeah that16:30
Wizzupand 'w' is automatically '2'16:30
uvosok16:30
uvosok16:30
uvosbtw the hildon opening animation is really janky for sphone16:37
uvosbecause it launches and then infroms the primary instance and then exits immitaly16:37
uvosand then the window pops up16:38
uvosmodest also has this problem16:38
uvosnot sure if there is a solution to tell hildon to keep the rectangle even if the process exits16:38
Wizzupwhat should I see with modest?16:38
uvoswith both sphone and modest the rectangle/screenshot spawns then hangs while the new window appears then the new window appears16:39
uvoslooks janky16:39
Wizzuphm I am not sure if I see it with modest16:39
Wizzuplet me upgrade sphone16:39
uvosprobubly cant see it on n900 (its to slow)16:40
WizzupI see it on d4 I think16:41
Wizzupkind of a minor detail but yeah :p16:41
uvosmaybe a .desktop key that tells hildon: this launches really fast dont bother with the animation16:43
uvosis in order16:43
Wizzupthat could be a solution yeah16:44
WizzupI think the phone app on fremantle has the same problem16:44

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