| sicelo | with upower, status-battery, and mce done, the only remaining stuff will be: (1) shipping a modified UPower.conf [via leste-config] for devices that need it. I think N900 and Droid4 at least. I doubt devices like PP need it | 00:01 |
|---|---|---|
| sicelo | (2) shipping, via leste-config, an mce-ini file with the new mce keys. Droid 4 probably doesn't need these, since it can just use the CriticalVoltage key. For N900, I do recommend using the new options and not CriticalVoltage. The new keys are sure to allow the calibration trigger to be hit before shutdown begins | 00:04 |
| Wizzup | that sounds good :) | 00:13 |
| sicelo | gn for now. if there is need to clarify anything in connection with the foregoing, you can just ping | 00:18 |
| Wizzup | gn, will do | 00:25 |
| Wizzup | sicelo: I merged some of the PRs, will make releases tomorrow | 01:36 |
| Wizzup | mkfx: well so armel will require an additional builder, and we'll have to rebuild all our packages, deal with any compiler warnings, other flags, missing asm that nokia had for arm, etc | 01:37 |
| freemangordon | Wizzup: I think we're hitting some ICE | 07:55 |
| freemangordon | ugh | 08:12 |
| freemangordon | Wizzup: tests are using /lib/arm-linux-gnueabihf/librtcom-eventlogger.so.1 :D | 08:12 |
| freemangordon | scratch that | 08:26 |
| freemangordon | Wizzup: this start to look as an issue in sqlite, but need some more time to confirm | 08:43 |
| freemangordon | Wizzup: https://paste.debian.net/1368822/ | 09:21 |
| freemangordon | have to stop for now, no idea when I will be able to continue | 09:22 |
| freemangordon | but at first glance it looks to be either ICE or sqlite3 bug | 09:23 |
| freemangordon | BTW, you said you have a test program, could you share it? | 09:45 |
| Wizzup | freemangordon: the test program is in /root/ - it's mostly just the failing tests | 10:37 |
| Wizzup | I think you're at the point (or maybe past) where I got stuck as well- trying to figure out the printf problem in sqlite | 10:37 |
| freemangordon | yeah | 11:24 |
| Wizzup | freemangordon: I guess would get either older/newer compiler or older/newer sqlite and re-test | 13:07 |
| Wizzup | I guess we could* | 13:07 |
| sicelo | Wizzup: just to take you back to voicecallmanager ... I understand correctly that sphone was switched to qt mainloop because vcm is qt as well? | 13:51 |
| sicelo | my d4 battery is toast, so I'm daily'ing the N900 now. outgoing calls are OK, but incoming is quite bad as far as UI goes. I hit answer, and UI doesn't immediately reflect this, and still vibrates. However, call has already been picked. Anyway, I blame my SD card for this | 13:54 |
| Wizzup | yes, sphone has to be qtloop | 13:54 |
| Wizzup | I don't think vmg has anything to do with vibration though | 13:54 |
| uvos__ | for sure sphone+glibloop+ofono is mutch lighter than sphone+qtloop+vcm | 13:59 |
| sicelo | yes. what I want to ask is - is it possible to answer a vcm/telepathy call directly via dbus (like it's possible with ofono)? | 13:59 |
| uvos__ | if you dont need vcm you should probubly use ofono on n900 | 13:59 |
| uvos__ | sicelo: yes its possible | 14:00 |
| uvos__ | someone would have to write a native sphone telepathy sphone module | 14:00 |
| uvos__ | Wizzup mainly chose to implement vcm because its was less work to do so | 14:01 |
| Wizzup | sicelo: telepathy is dbus, so yes, but you'll have to replicate many many manhours that we don't have for not a big percieved gain | 14:01 |
| uvos__ | regarding vibration, stoping vibration happens after the backend reports the call was picked up locally | 14:03 |
| sicelo | right I see what I'm looking for in vcm's dbus. let me make some tests | 14:03 |
| uvos__ | which means theres a syncronous trip to the backend | 14:03 |
| uvos__ | so it could be the vcm/telepathy stack being slow | 14:03 |
| uvos__ | or it could be sphone stalling for some other reason | 14:04 |
| Wizzup | right, it could also be a bug in our code | 14:04 |
| Wizzup | I don't think the dbus round trips are really what kill the performance jere | 14:04 |
| Wizzup | s/jere/here/ | 14:04 |
| uvos__ | dbus round trips can be quite slow | 14:04 |
| uvos__ | mce spends most of its time in those | 14:04 |
| uvos__ | but yeah | 14:04 |
| uvos__ | could be either | 14:05 |
| sicelo | what I want to try is pickup the call (via dbus) with sphone killed. currently testing ... | 14:05 |
| sicelo | Wizzup: the stack is ofono -> tp -> vcm? | 14:07 |
| uvos__ | jup | 14:07 |
| uvos__ | -> sphone-tp -> sphone | 14:08 |
| sicelo | ok. so yes, let me see what happens at the vcm layer. I'm aware the ucm, etc. won't be working, but that's a different concern. for now I just want to see if reacting to an incoming call is fast enough on the N900 | 14:09 |
| sicelo | unfortunately seem to have run out of call credit :-p | 14:10 |
| Wizzup | I don't have troubles on my n900 | 14:12 |
| sicelo | prolly my SD card indeed | 14:12 |
| Wizzup | and lack of nice/cgroups/etc | 14:12 |
| sicelo | also to be clear, I'm definitely not considering replacing tp/vcm. but might end up looking for a simple way to not add anything on any layer higher than these | 14:13 |
| sicelo | for testing/fun :-) | 14:14 |
| Wizzup | my aim is to reduce additional work and use products that are supported upstream (in this case sailfishos) so that we can focus on many other things that are lacking | 14:15 |
| Wizzup | your sd card is likely much slower than the emmc as well, so it's not a fair comparison then | 14:16 |
| Wizzup | (when comparing to fremantle) | 14:16 |
| Wizzup | keep in mind that vcm also gives us xmpp/sip calls | 14:16 |
| sicelo | agreed, that's why I'm not even remotely considering removing/bypassing it. I'll have a look at vcm's dbus interface | 14:20 |
| arno11 | sicelo: yeah Wizzup is right imo @sdcard and lack of nice/cgroup. one more thing is the current bug with incoming call on n900 | 15:43 |
| arno11 | issue #767 | 15:43 |
| arno11 | only on daedalus | 15:44 |
| sicelo | i saw that issue | 15:45 |
| arno11 | ok, btw boost mode and H-D renicing to -8 should help your sdcard | 15:47 |
| arno11 | *for incoming calls | 15:47 |
| arno11 | Wizzup: currently working on new picodrive (almost finished) but have few questions | 16:21 |
| Wizzup | shoot | 16:23 |
| arno11 | 1) is the buider compatible with flto (link time optimization) ? | 16:23 |
| arno11 | 2) does picodrive need a new tag for new daedalus version ? | 16:24 |
| arno11 | 2) if i remove cortex and neon optimization from configure file, does the buider automatically add them to armv7 ? | 16:26 |
| arno11 | *builder | 16:26 |
| arno11 | (in fact i'd like to make a more generic build to be able to build pico for arm64 and amd64 | 16:27 |
| arno11 | *3) | 16:28 |
| arno11 | btw local build (with armv7 optimizations) works better than ever on n900: some 32x games run @60fps ! | 16:36 |
| Wizzup | I think we should be able to lto | 16:37 |
| Wizzup | I think we have some default compiler flags in our builder for armhf, but there are ways not to add them | 16:37 |
| arno11 | ok cool, if not, that's easy to fix anyway | 16:37 |
| Wizzup | https://git.maemo.org/leste/jenkins-integration/src/branch/master/debian_glue#L166 | 16:37 |
| arno11 | *@flto | 16:37 |
| arno11 | ok @flags | 16:38 |
| Wizzup | nice @ 60fps :) | 16:38 |
| arno11 | yeah i should make a video running virtua racing for next news post, as it was impossible to do that on fremantle | 16:39 |
| sicelo | fun, `busctl --user call org.nemomobile.voicecall /calls/active org.nemomobile.voicecall.VoiceCall answer && pactl set-card-profile 0 'Voice Call'` gives me working incoming call (obviously still need to know when the call has arrived) | 16:39 |
| Wizzup | sicelo: making a sphone.sh? :D | 16:39 |
| sicelo | exactly ;-) | 16:40 |
| Wizzup | I wish we'd be able to write some of our things in rust or go, but with gtk2 this is just not doable | 16:40 |
| Wizzup | maybe if/when we move to gtk3 :) | 16:40 |
| arno11 | Wizzup: i'd like to launch a daedalus picodrive build plz :) (as a test) | 16:49 |
| Wizzup | maemo/daedalus branch? | 16:49 |
| arno11 | yes plz | 16:50 |
| Wizzup | running | 16:50 |
| arno11 | cool ty | 16:50 |
| arno11 | at least source is ok | 16:52 |
| Wizzup | I'm planning to fix this also in the near future, the complexity of running jobs | 16:52 |
| arno11 | ah indeed could be cool | 16:52 |
| arno11 | it fails @binaries, will have a look later | 16:58 |
| arno11 | libgles dep missing in control imo | 17:02 |
| sicelo | Wizzup: although i won't use/test this right away ... sip calls also come in via vcm? or those are only available at tp level? | 17:23 |
| sicelo | and, if you still recall, what does "active" call state really mean, because here's what i got on the call: https://paste.debian.net/hidden/ee140b7b/ | 17:41 |
| sicelo | i'd think active means the call has been answered, but then, (in that log) i see it changing to 'alerting', then 'active' again :-/ | 17:42 |
| Wizzup | sicelo: well, tp, and then vcm, yes | 17:51 |
| freemangordon | sicelo: install procmgrd and set all dbus processes nice to -5 with it, then see if it helps with latency | 18:22 |
| freemangordon | sicelo: example https://paste.debian.net/1368937/ | 18:23 |
| freemangordon | you may want to increase tp-ring and vcm prio as well | 18:24 |
| freemangordon | though, we shall control those through cgroups | 18:24 |
| freemangordon | sicelo: procmngd will be in repos in few minutes | 18:47 |
| sicelo | want to try a sip call on the N900 now. i add it via SIMPLE (haze), or I need to use a different connector? | 19:12 |
| freemangordon | SIMPLE does not support voice | 19:22 |
| freemangordon | sicelo: iirc, you need tp-rakia | 19:48 |
| arno11 | yeah tp-rakia for sure | 19:55 |
| sicelo | ok cool, thanks. i can configure the account from UI, or needs mc-tool? | 19:57 |
| arno11 | should work from the ui | 19:57 |
| arno11 | sip calls work from sphone/vcm but atm the best sip results i got on n900 were through twinkle | 20:00 |
| sicelo | i hope to try them on dbus :-P | 20:03 |
| arno11 | yeah could be cool :) | 20:03 |
| sicelo | i do see the sip call in vcm dbus iface... nice!!! | 20:43 |
| arno11 | cool | 21:02 |
| arno11 | Wizzup: btw i need another kick for picodrive plz (when you can ofc) | 21:03 |
| Wizzup | done | 21:20 |
| Wizzup | arno11: looks like it built | 21:42 |
| arno11 | Wizzup: yeah, cool, ty | 22:01 |
| arno11 | at the moment, that's still experimental as the emulator is really powerfull with last notaz stuff but still not really user friendly unfortunately | 22:02 |
| arno11 | cool, the pkg works fine, full speed. could be interesting to test on PP + kbd | 22:17 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!