Friday, 2019-10-25

Mister_Magisterwoohoo i'm back from ban :P06:32
*** Mister_Magister is now known as Guest2522906:35
*** Mister_Magister_ is now known as Mister_Magister06:50
jusaPSA: if someone is using pulseaudio-modules-droid-hidl please note that I split the package to two separate packages, audiosystem-passthrough ( and pulseaudio-modules-droid-hidl ( with latter depending on the former.09:06
Mister_Magisterjusa: hi btw got problem when calling on different device. on loudspeaker mode person can hear himself09:09
Mister_Magisterofc i should give u logs righte09:09
T42<Martensite %lastname%> (Document)
T42<IAmPony288> Any1 have pixel 3 ? Some1 made post asking for port for 350$09:52
Thaodanjusa: is this a replacement for audiofingerlue?10:34
jusaThaodan: yes10:46
jusadoes the same thing, just a bit differently10:47
ThaodanBut better?10:47
ThaodanI hope nothing changed for android 1010:47
jusawell, simpler. easier to build than audioflingerglue10:48
jusaso if adaptation already works with audioflingerglue no point in changing but new adaptations should be easier10:48
jusaand for QC devices with android >= 9 most likely one needs to use the new implementation anyway10:48
malI built the audiosystem-passthrough on nemo:devel:hw:common
malMister_Magister: which device was it that needed in droidmedia? fxtec?11:28
Mister_Magistermal: i have no idea11:31
malMister_Magister: you are the one who talked about that in channel logs11:32
Mister_Magisteri have no memory of that11:33
Mister_Magistercamera is working fine for us and i didn't fixed it so… yeah…11:34
mal14:44 < Mister_Magister> abranson: can i add to minimedia
Mister_Magisterthere was something broken with build11:38
maljust wondering which device11:39
Mister_Magisterah fxtec yes11:39
Mister_Magisterso hybris16 base11:39
malis the device repo for that somewhere?11:39
Mister_Magisteri mean it is on fxtec servers :P11:40
malI was just curious if there were any qti/camera related defines in device repo that could be used to determine if that additional lib is needed11:40
Mister_Magistermal: like BOARD_QTI_CAMERA_32BIT_ONLY?11:42
Mister_Magistercan't see anything else11:43
Thaodanjusa: is the droid side of audiofingerglue still needed?12:36
jusaThaodan: either use pulseaudio-modules-droid-glue+audioflingerglue OR use pulseaudio-modules-droid-hidl+audiosystem-passthrough12:52
maljusa: and which one to use depends on which android base is used if I understood correctly12:54
Thaodanand  which device is targeted12:54
jusayes, only qcom devices need those for voice call to function15:54
Thaodanjusa does it want to reuse the cross compiler that is used during kernel builds?15:58
jusaThaodan: it's built with normal sdk compiler16:01
ThaodanCROSSCOMPILE is a variable used for building the kernel with a cosscompiler that why I asked16:02
Thaodanwhich is why it tried to build using my cross compiler that I used for the kernel build16:05
T42<DylanVanAssche> mal: Have you already done something for the QMI modem? Just to avoid to do the same work.16:32
T42<DylanVanAssche> I gave the modem a spin again, seems that SMS works fine (cmd line), calling is broken though.16:32
mal@DylanVanAssche not yet, sorry16:32
T42<DylanVanAssche> mal: No need to be sorry :) Just checking in to make sure that we don't do the same work twice as with the V4L stuff ;)16:33
mal@DylanVanAssche are you going to that the qmi work?16:41
T42<DylanVanAssche> mal: Maybe in the next couple of days. The Pine devices are ported except for LIMA & modem (SMS, calling, data)16:42
T42<DylanVanAssche> GPS is also in the modem, but we got that working already in the past16:42
mal@DylanVanAssche just to be sure you verified the latest version of that iio code on device?16:44
T42<DylanVanAssche> mal: It's working fine :) Waiting for review on git.sailfishos.org16:44
mal@DylanVanAssche ok, I started to think about the define naming again but maybe that is too much nitpicking, one of those if for input (from iio) and other is for output from sensorfw16:45
maljust wondering if there can be confusion because of that and the naming16:46
T42<DylanVanAssche> mal: I think it's clear enough by prefixing it with PROXIMITY_. You know that's about the proximity sensor.16:47
T42<DylanVanAssche> We can improve it if the code can be cleaned up?16:48
T42<DylanVanAssche> when we clean up the code*16:48
mal@DylanVanAssche merged and tagged17:02
deathmistmal: can you check my PR #165 for hybris-boot from late July this year? I've updated cheeseburger (OnePlus 5) mountpoints for treble & hybris >=15.1 bases (and just a moment ago added dumpling a.k.a OnePlus 5T to use same mountpoints) which both are confirmed to work on mine and others devices :)17:27
maldeathmist: ok, does the device really have such complicates partition structure?17:27
deathmistyep, that's all the partitions listed under "ls -l /dev/block/bootdevice/by-name/", I've thought about cleaning it up too17:28
rodrigosolarihelp me17:29
maldeathmist: sorry to bug you but could squash the commits in that pull request to one17:31
maldeathmist: or actually maybe two is ok, at least remove the merge commit17:32
deathmistmal: how would I do that properly? I've tried it on a test branch and succeeded in fully removing the commits from the branch with an interactive rebase and dropping the commits I merged from master earlier, but wouldn't this also remove them from the upstream master repo? I've not done something like this before so I just want to make sure I don't17:48
deathmistmess things up :)17:48
maldeathmist: "git rebase -i HEAD~3" and then change the two latter commits from pick to squash and then once you save the edit then it will ask for new commit message for the combined commit, I think one commit is enough in this case17:50
deathmistsure, but what about the merge commit(s) from upstream/master? should I drop them or?17:51
malif you do the squash like I said it will also be gone, or should be17:52
malwait a moment, I will clone that and test17:54
maldeathmist: hmm, looks like your way was a bit ugly, trying to see how to fix it17:55
maldeathmist: usually rebasing is preferred over merging17:55
deathmistyeah I accidentally merged from upstream and realized later that it was the PR branch :/17:56
maldeathmist: run "git rebase upstream/master"17:56
malthat will make the merge commit go away17:57
deathmistmal: I tested on another branch and it seems to be exactly what we need! now when I push to "origin cheeseburger-treble" instead tho it wants me to pull, should I just force push?18:03
malgit push --force18:04
deathmistaand it's gone! I'm learning slowly to master git as well :D18:05
malgit can be confusing sometimes18:05
vknechtsaw that recently, might help some day: :-)18:05
malnice :D18:07
deathmistmal: I assume the PR is now in a good state and we don't need to merge the treble update and dumpling commit into one? :p18:15
maldeathmist: merged18:24
deathmistmal: awesome, thanks! another thing I just remembered is droidmedia PR #11 (conditional AudioPolicyService startup), which my and many others devices have needed for a long while was approved by abranson a few days ago but still not merged yet18:53
Mister_Magisterdeathmist: i use that website a lot :P19:00
maldeathmist: I forgot to merge that droidmedia PR19:11
maldeathmist: droidmedia is also merged and tagged now19:17
Thaodantalking about droidmedia: is MiniSurfaceFlihnger an abstract class?19:44
ThaodanI'm getting this error while building libminisf:  allocating an object of abstract class type 'MiniSurfaceFlinger'19:46
Thaodan        return sm->addService(String16(SERVICE::getServiceName()), new SERVICE(), allowIsolated,19:46
malThaodan: you mean for android 10? you need to add suitable services file, you can find the old ones in services folder19:53
malin case you haven't done it already, just copy the android 9 version and then adapt to API changes19:53
Thaodanmal: found my error missed to implement some new services19:54
Thaodaneh methods19:54
malyes, usually some methods needs to be added or modified19:55
Thaodanis there a reason argument names missing or this there some kind of generator? Its just C&P and removing the argument names adds extra work.19:55
malthis is the first time in a long time when I'm not the one adapting to new android base19:55
Thaodanis that a good thing? :D19:56
malwell in a way it might have been faster if I did it since I have quite much experience19:56
malThaodan: what do you mean? the file was made manually19:56
Thaodanfeels like a deep dive through SFOSs inner workings.19:56
malalways just the previous android version service file is taken as the base and modified where needed19:57
ThaodanI means generating this file since its pretty much just taking the header and adding something like return 019:57
malThaodan: mostly yes but it has many classes there19:58
maland some are not just return 0, most are, easier to copy than make a script19:58
T42<adampigg> Well, that was a digital detox kinda day, accidentally left my phone at home!19:59
malThaodan: do you have those changes somewhere in your github?20:01
Thaodanmal: I got them on github20:14
ThaodanI forked the mer/sailfishos repos I needed to edit.20:14
T42<DylanVanAssche> Mal: thanks!20:57
mal@DylanVanAssche were those all for now, I'll remove sensorfw from dontbeevil repo because I added that new version to native-common21:00
T42<adampigg> Neat21:02
T42<DylanVanAssche> mal: all sensors are working now, we can use them from the common repo for now :)21:05
mal@DylanVanAssche what sensors does it have? do we need to implement compass in sensorfw using sensor fusion or something21:11
T42<DylanVanAssche> mal: Compass is already in there (magnetometer). The scaling is wrong according to the kernel ABI docs and my own measurements from the Xperia X. The PinePhone has: accelerometer, gyroscope, magnetometer, als and proximity sensor. If you count GPS also as a sensor, then also GPS21:17
mal@DylanVanAssche actual compass direction is usually calculated from magnetometer, accelerometer and gyroscope (if available)21:19
mal@DylanVanAssche if you look at some compass app that is available in store then you can see that compass is probably not seen for your device21:20
mal@DylanVanAssche I have such a virtual sensor on my endless todo list21:21
T42<DylanVanAssche> Oooh I haven't tested that yet, will do. But that's not urgent, the modem is since withoit sms/calling/data I can't use it as daily driver21:21
Thaodanmal: this are my changes to droidmedia21:35
mallooks reasonable21:48
malThaodan: some inconsistency, you still make 10 include libmediaextractor folder but don't add that as lib, is the include needed?21:51
malof course the include might not cause any problems21:52
Thaodanmal: doesn't cause problems but I removed it.22:05
Thaodanthe last error is this: frameworks/native/include/android/looper.h:47:16: error: reference to 'ALooper' is ambiguous22:08
Thaodantriggered by including in #include <sensor/Sensor.h> included in the android 10 services  file22:08
Thaodanshould it be the stagefright variant or the stagefright variant of that type?22:10
ThaodanApperantly stagefright implements its own ALooper overring the struct from looper.h23:11
malmaybe you need to include something else then23:13
MSameerwow! droidmedia is still alive after all those years? :)23:16
malMSameer: it is :)23:17
Thaodanmal: the issue is that other stagefright headers are included that may include ALooper23:17
MSameermal: seems so :)23:18
Thaodanfrom stagefright23:18
malMSameer: for newer android bases there could be optional ways to do it using direct binder communication using libgbinder23:18
ThaodanOn android 8+?23:20
MSameermal: you are the expert here. we did not have those long ago (or I did not know). but of course if there is better usable API then it'd be better23:21
ThaodanI'm not the expert here after seen the other changes the way using those interfaces via binder is less invasive23:22
ThaodanBecause the binder interfaces are public interfaces that don't change so much compared to building directly on top23:23
MSameerit is always a trade off. easier/public interfaces vs code size23:25
malMSameer: that new way came with the treble feature in android, so a stable API for many things23:25
MSameercode size loaded in ram23:25
MSameer+1 :)23:26
malMSameer: we use those binder interfaces already for things like modem, gps, sensors etc23:26
malfor android 8+ devices23:27
malfrom the main hw features audio and camera/codecs are the ones left not using that23:27
ThaodanI hope that more features get mainlined and uses less binder services if possible.23:28
MSameermal: media code is not small unfortunately. not that easy to port23:29
MSameeror not that trivial rather23:30
malMSameer: yes, that is the reason why we haven't tried that yet23:30
MSameerbut you are doing a great job :)23:30
MSameerthe problem is backward compatibility23:32
MSameerthat's if you still need to support jolla1 phone :P23:32
ThaodanFor Android <8 yes23:32
ThaodanIm interrested in seeing more changes23:33
malMSameer: yep, we need to keep droidmedia quite long time for the current devices23:46
Thaodanmal: do you have an idea?
malThaodan: not right now, it's quite late already23:48
Thaodanthats ok, I know its late. Sometimes it hard to stop when you onto something and time goes by :D23:53
malI know :)23:54
Thaodanwhat keeps you up?23:57

Generated by 2.17.1 by Marius Gedminas - find it at!