| dsc_ | btw our /tmp isnt really tmp right | 01:46 |
|---|---|---|
| dsc_ | as in, in-mem | 01:46 |
| dsc_ | i need a place to write logs to but dont want to abuse the SD card | 01:46 |
| Wizzup | well, normally you'd log anything warning/error level to syslog and then that will take care of it | 01:54 |
| dsc_ | ok | 01:54 |
| Wizzup | everything else should probably just be some flag to log it at a higher debug lvl and log it to sd card | 01:54 |
| branon | sd cards are meant to be abused | 01:55 |
| gnarface | dsc_: logs aren't usually very high traffic for a personal, single-owner device, unless something is really, really wrong. the thing you want to avoid putting on there is swap. | 03:18 |
| dsc_ | gnarface: I was wondering because some distros that target embedded like IIRC armbian write logs to ramdisk then .gz them in bulk to disk periodically | 04:11 |
| dsc_ | but I forgot the specifics | 04:12 |
| uvos | on leste logs are pretty high traffic | 08:38 |
| uvos | because of high verbosity and because some leste componants are of pretty poor quality and glib logs alot of errors | 08:38 |
| Wizzup | dsc_: syslog-ng can periodically gz things yes | 09:41 |
| Wizzup | probably our other syslog can too | 09:41 |
| Wizzup | uvos: yes right now it's pretty high :D | 09:41 |
| arno11 | Wizzup: weird (again), ringtone stopped working again and sphone stopped showing caller name (but 'unknown' instead). sphone still shows caller name in 'recent' | 13:26 |
| arno11 | btw the ringtone stuff is an old random issue and probably not related to recent vcm/tp stuff ofc | 13:28 |
| arno11 | seems sphone related only (sip call ringtone works with twinkle and other sounds are not affected). | 13:37 |
| arno11 | uvos: ^^ | 13:38 |
| Wizzup | building new converations that makes it possible to open chats from contacts again | 13:55 |
| Wizzup | dsc_: apparently the last conversation where I fixed the protocolIstelephone makes the UI no longer show up... lol | 15:14 |
| Wizzup | what the.. | 15:14 |
| uvos | arno11: not showing caller name for calles placed with vcm is expected for now | 15:14 |
| uvos | arno11: sphone simply uses gst to play the ringtione, there is no special ringtone volume but sphone dose respect the rintone bit in profiled so ensure this bit is not set | 15:16 |
| uvos | arno11: to check if gst istelf is working you can open sphone -c options and try playing the ringtone there | 15:16 |
| uvos | arno11: also note that the ringtone selection in leste is compleatly broken, it hardcodes a bunch of ringtones that are not avaialble at all and has no bearing on actiual available rintone files | 15:17 |
| uvos | only selecting your own file or never touching it at all after install works | 15:17 |
| uvos | using sphones own slection mechanism via sphone -c options works too ofc | 15:21 |
| uvos | but the settings applet is broken in various ways | 15:21 |
| arno11 | uvos: ok ty | 15:21 |
| arno11 | yes indeed (for settings applet) | 15:22 |
| arno11 | i remember now having troubles with gst few months ago and reinstalling the 2 involved pkgs made ringtone working again iirc | 15:23 |
| uvos | no combinaton of pkgs can make it really work | 15:23 |
| Wizzup | isn't canberra a better way to do play these things? | 15:24 |
| uvos | the whole thing simply needs to be reworked, it needs to select ringtones based on the files in the rintone folder | 15:24 |
| Wizzup | or is that not for repetitive things? | 15:24 |
| uvos | not based on an arbitary hardcoded list | 15:24 |
| uvos | Wizzup: the whole concept of canberra would not work unless you teach it to obay libprofile | 15:26 |
| uvos | Wizzup: or better if you remove the slection of sound files from libprofile and put them into canberra | 15:26 |
| Wizzup | ringtone selections works good for me on fremantle, so maybe we're missing some bits | 15:28 |
| uvos | Wizzup: there is no missing bits | 15:28 |
| uvos | its simply broken by design | 15:28 |
| uvos | it only works on freemantle because all hardcoded assumptions are true there | 15:28 |
| Wizzup | ok, but it works fine on fremantle :) | 15:28 |
| uvos | yes if you have excatly all sound files in exactly the places where they are hardcoded to be then it works | 15:29 |
| uvos | this is not really workable ofc | 15:30 |
| Wizzup | so the ringtone not playing for arno is related to this? | 15:30 |
| Wizzup | because I'd think if it works for him some of the time, it shouldn't be related to this | 15:30 |
| uvos | if he ever tried to select a ringtone via the applet yes | 15:31 |
| uvos | because not libprofile carries a string pointing to a file that dose not exist | 15:31 |
| uvos | this is easly testable by setting the string from sphone or setting a custom file via the applet | 15:31 |
| uvos | or just looking at the path in sphone -c options | 15:31 |
| uvos | beyond that sphone can currently play the ringtone if the volume is 0, which is ofc supriseing since you would expect this to have its own seperate volume | 15:32 |
| uvos | but we dont have a ucm "ringer" role or sutch that would allow this | 15:33 |
| Wizzup | so this will make it work randomly? | 15:33 |
| Wizzup | it seems to me if it works some of the time it should just always work, or never, not some of the time | 15:33 |
| uvos | the last one could def look random | 15:33 |
| uvos | i ofc am only speculateing and giveing resons why it may work | 15:33 |
| uvos | i have also noted that using mafw/omp quite often breaks pulseaudio | 15:34 |
| uvos | this may also be the reason | 15:34 |
| Wizzup | not on n900 | 15:34 |
| Wizzup | this is (another) droid4 kernel issue | 15:34 |
| uvos | and he would have to check if pa works ingeneral in the non-ringing state | 15:34 |
| Wizzup | and he doesn't use the d4 | 15:34 |
| uvos | ok well point is there are manny potentiall reasons i can come up with | 15:35 |
| uvos | but without further testing i cant say | 15:35 |
| Wizzup | ok, I doubt it's because of the libprofile setting of a ringtone though | 15:35 |
| Wizzup | but yeah, good to rule things out | 15:35 |
| uvos | trying to play the ringtione on demand via sphone -c options is a good test | 15:35 |
| uvos | since it excercises the same code that plays the ringtone for real | 15:35 |
| arno11 | playing ringtone frome sphone -c always works btw | 15:36 |
| arno11 | but not from a call | 15:36 |
| uvos | ok theres not mutch difference between these really | 15:37 |
| Wizzup | well if the ucm is already on phone call mode I wouldn't expect it to play a sound | 15:37 |
| arno11 | indeed | 15:37 |
| uvos | one is that the libprofile silent bit is checked by the call but not by options | 15:37 |
| arno11 | but i already have a look and ucm is not already in call | 15:37 |
| Wizzup | dsc_: I'm bisecting the UI no longer showing up | 15:37 |
| uvos | the other is ucm | 15:38 |
| uvos | ok you checked that | 15:38 |
| arno11 | yep | 15:38 |
| uvos | dose playing another sound via pa work during the ringing phase | 15:38 |
| uvos | maybe the modem pa module is breaking pa for a bit | 15:38 |
| arno11 | i doublechecked again and ucm is still under hifi profile and yes playing another sound works during the ringing phase | 15:44 |
| uvos | ok | 15:44 |
| arno11 | and when ringtone works, other sounds are mixed with it | 15:44 |
| uvos | is sphone uhappy about something? | 15:45 |
| uvos | try running it in a console with -v -v logs | 15:45 |
| arno11 | ok | 15:45 |
| uvos | (you have to kill it first) | 15:45 |
| arno11 | ok | 15:46 |
| arno11 | it seems ok | 15:49 |
| arno11 | and it says gstreamer is playing the ringtone file | 15:50 |
| arno11 | but in fact not | 15:50 |
| arno11 | ucm is switching correctly | 15:51 |
| arno11 | call properties and status seem ok | 15:51 |
| uvos | i mean this means it sucesfully gets here | 15:53 |
| uvos | https://github.com/maemo-leste/sphone/blob/d7e5442aae137664f14a813c66a258b1afb2d9ad/src/modules/playback-gstreamer.c#L105 | 15:53 |
| arno11 | ah ok | 15:54 |
| uvos | and none of the subsiquent calls failed | 15:54 |
| uvos | since those return NULL on failure | 15:55 |
| uvos | g_object_set and gst_bus_add_watch would complain about bein passed NULL (but this snippet dose need some better error checking) | 15:55 |
| uvos | neither dose gst report GST_MESSAGE_ERROR | 15:56 |
| uvos | so this is pretty mysterious | 15:56 |
| arno11 | yes indeed and totally 'random' | 16:01 |
| arno11 | ah now 'caller name' works again after restarting sphone (still with vcm/tp) | 16:07 |
| uvos | thats even more mysterious since this is simply imposible | 16:19 |
| arno11 | lol | 16:19 |
| Wizzup | I've seen some races where loading the evolution stuff doesn't work | 16:19 |
| Wizzup | but I thought it was maybe just my setup | 16:19 |
| Wizzup | that would probably explain it | 16:20 |
| uvos | sure but current sphone in repos is this https://github.com/maemo-leste/sphone/blob/0b40311c7ae58d8ad788acd6154471998281069e/src/modules/contacts-evolution.c#L121 | 16:20 |
| uvos | the code simply dose nothing unless the backend is ofono | 16:20 |
| Wizzup | ok, I think if you use the qtloop a few times you might run into the same race | 16:22 |
| Wizzup | I gotta go now though | 16:22 |
| arno11 | i doublechecked again and with qtloop and last vcm-fixes branch, i get caller name when the phone is 'ringing' | 16:25 |
| uvos | oh you compiled sphone yourself | 16:27 |
| uvos | well that woudl have been good to know | 16:27 |
| uvos | theres plenty of unlreased code | 16:27 |
| arno11 | yes | 16:27 |
| arno11 | but i got exactly same troubles before | 16:27 |
| uvos | yes caller name in arbirary modules is a new feature | 16:27 |
| uvos | well before you had the repo version | 16:27 |
| arno11 | yes | 16:28 |
| uvos | as i say its not implemented thair at all | 16:28 |
| uvos | so ofc that dosent work | 16:28 |
| uvos | vcm-fixes branch is broken anyhow | 16:28 |
| uvos | since sphone will fail to read the contact name of past events if saved with the commit that changes loging | 16:29 |
| arno11 | well, it works fine on n900, at least from a user pov like me | 16:29 |
| uvos | (from the database) | 16:29 |
| uvos | nope its broken | 16:29 |
| uvos | if it can lookup the contact later it gets hidden | 16:29 |
| uvos | but the contacts system is totally broken there | 16:30 |
| Wizzup | weird, it works for me in the branch, so what do you mean 'totally broken'? | 16:31 |
| uvos | this dosent mean that the issue you see is related as this points toward more that contact lookup is not working rather than contact loading from the db is not working | 16:31 |
| uvos | Wizzup: its broken it can load the contact name form the db | 16:31 |
| uvos | sphone then tries to look it up in evolution | 16:31 |
| Wizzup | it can for me | 16:31 |
| uvos | but its still broken | 16:31 |
| uvos | not it _DOSE NOT WORK_ period | 16:31 |
| Wizzup | maybe if you explain what is broken it might be more clear :) | 16:32 |
| uvos | its hidden from the user if it manages to look it up in eds later | 16:32 |
| uvos | its still broken | 16:32 |
| uvos | anyhow i dont thing its related to arno11s problem | 16:32 |
| Wizzup | ok, maybe I didn't have these commts in my testing | 16:32 |
| uvos | which as is say sounds more like eds lookup is flaky | 16:32 |
| Wizzup | yeah, like I said, I think there is/was a race with qtloop somehow, that could be it, where it would fail to load the evo module | 16:32 |
| Wizzup | but I don't recall for sure | 16:33 |
| Wizzup | and since qtloop it used for vcm , but not for the 'regular' setup, it could be related | 16:33 |
| uvos | chaning the xsession sphone script to verbose and loging that somewhere | 16:33 |
| Wizzup | I think you could probably reproduce this too | 16:33 |
| uvos | would help to see if the eds module sometimes dosent initalise for whatever reason | 16:33 |
| Wizzup | but I need a break for now :) | 16:33 |
| arno11 | uvos: just to avoid confusion: the only real problem i have is ringtone and it is ofc not related to any sphone custom branch since the issue appears on fresh install as well and for a long time now | 16:44 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!