| Wednesday, March 6th, 2019, 00:10 UTC | ||
| [00:10:39] | natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 252 seconds) | |
| [01:46:10] | amessina (amessina!~amessina@unaffiliated/amessina) has quit (Quit: Konversation terminated!) | |
| [02:57:59] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has joined #mythtv | |
| [02:57:59] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [03:29:42] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has quit (Quit: Konversation terminated!) | |
| [05:51:01] | ElmerFudd (ElmerFudd!~le@jagular.dk) has quit (Ping timeout: 268 seconds) | |
| [05:51:29] | ElmerFudd (ElmerFudd!~le@jagular.dk) has joined #mythtv | |
| [06:06:53] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has joined #mythtv | |
| [06:06:53] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [06:38:01] | markspieth22 (markspieth22!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [06:38:01] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth22 | |
| [06:41:02] | markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 244 seconds) | |
| [06:56:32] | markspieth22 (markspieth22!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 244 seconds) | |
| [07:03:36] | sdrik (sdrik!~sdrik@sdrik.gabriello.fr) has joined #mythtv | |
| [07:22:19] | SteveGoodey (SteveGoodey!~steve@host86-145-19-224.range86-145.btcentralplus.com) has joined #mythtv | |
| [07:22:19] | Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey | |
| [07:53:35] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has quit (Quit: Konversation terminated!) | |
| [07:53:53] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [08:11:31] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [08:11:47] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [08:22:17] | xris (xris!~xris@mythtv/developer/xris) has joined #mythtv | |
| [08:22:17] | Mode for #mythtv by ChanServ!ChanServ@services. : +v xris | |
| [08:35:55] | SteveGoodey (SteveGoodey!~steve@host86-145-19-224.range86-145.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [08:50:35] | willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has joined #mythtv | |
| [09:00:11] | markk_: | peterbennett; did you have to do anything special to handle mediacodec returning deinterlaced frames? i.e. is there any special handling of the increased frame rate. I can't see anything. trying to figure out why interlaced playback is so wonky with mediacodec surface rendering. |
| [09:00:52] | markk_: | probably just highlights other issues |
| [09:03:50] | markk_: | stuarta: I can't find away to get vaapi direct rendering with wayland without egl. Of the available VA displays – only DRM and wayland will open successfully. DRM needs EGL to map the vaSurface->EGLImage->GL texture. With a wayland vaDisplay, in theory the GLXPixmap support works but it fails as vaPutSurface is not implemented for the Wayland backend. |
| [09:04:43] | markk_: | I've trawled through the other players – kodi, mpv, libqtav – and none of them have a way of doing it. |
| [09:05:46] | markk_: | we could in theory force EGL at the start of mythfronted (i.e. before the GUI application is started). Maybe use a setting? would be preferrable to forcing users to set an environment variable |
| [09:07:05] | markk_: | would be nice if we had a better way of detecting when wayland is running as well. the current check is likely to be error prone.# |
| [09:07:11] | Merlin83b (Merlin83b!~Daniel@office.34sp.com) has joined #mythtv | |
| [09:16:03] | stuarta: | markk_: is there a compatibility layer involved in your vaapi testing, or do you have intel hardware, so using the full vaapi? |
| [09:16:27] | stuarta: | markk_: the next best way is the vaapi2 way |
| [09:16:41] | stuarta: | at least that way the decode is offloaded |
| [09:17:41] | stuarta: | markk_: any tests i can do for you do identify if vaPutSurface is implemented in the f29 backend? |
| [09:24:55] | markk_: | stuarta: using intel with libva2. vaputsurface doesn't appear in the main git repository for wayland |
| [09:32:53] | stuarta: | so near, yet so far |
| [09:33:42] | stuarta: | so if we find it's wayland, we need to use the /dev/dri/renderD128 device when opening the decoder with ffmpeg, and suffer the copy |
| [09:40:05] | markk_: | yes – but we need to advertise the fact that using egl is a much better option... |
| [09:43:21] | stuarta: | the problem is more user perception, if they happen to chose a distro that defaults to wayland, and have vaapi, it should "just work" without much more than selecting vaapi |
| [09:44:55] | stuarta: | which is hard |
| [09:49:35] | stuarta: | markk_: guess the question then is, how do we make egl easier to consume? |
| [09:52:36] | stuarta: | interesting https://android.googlesource.com/platform/har . . . ce_wayland.c |
| [10:09:14] | markk_: | why would android have libva? |
| [10:10:30] | stuarta: | dunno |
| [10:16:16] | stuarta: | nice API? |
| [10:30:46] | jpharvey (jpharvey!~jpharvey@95.149.108.241) has quit (Ping timeout: 246 seconds) | |
| [11:54:10] | markk_: | markspieth, gigem, peterbennett: been testing a 32bit build on my phone (quallcomm adreno). mediacodec surface is much better than the shield. still far from perfect. but no scaling issues with non square pixel sources. still dropping frames because the surface update callback is sometimes late. seems to be very codec dependant. |
| [12:40:38] | markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [12:40:38] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth2 | |
| [12:41:50] | markspieth: | peterbennett: set it to 50 and it seems pretty good. very occasional jumps but very small ones. A considerable improvement to before. |
| [12:42:07] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Ping timeout: 240 seconds) | |
| [12:46:13] | blahdodo (blahdodo!~blahdodo@69.172.190.157) has quit (Quit: Bye bye) | |
| [12:46:36] | blahdodo (blahdodo!~blahdodo@69.172.190.157) has joined #mythtv | |
| [12:47:43] | stuarta: | interesting. jan newmarch has done a book on programming wayland clients https://jan.newmarch.name/Wayland/index.html |
| [13:15:29] | SteveGoodey (SteveGoodey!~steve@host86-145-19-224.range86-145.btcentralplus.com) has joined #mythtv | |
| [13:15:30] | Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey | |
| [13:23:16] | KdW (KdW!~klaas@D93E7846.cm-28.dynamic.ziggo.nl) has joined #mythtv | |
| [13:26:54] | peterbennett: | markspieth: Try setting to 10, which is the default. |
| [13:28:01] | peterbennett: | markspieth: I have some time back made a change that uses a bigger number when it is out by more than 200 so setting a low value like 10 should provide the best experience. |
| [13:36:36] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [13:39:04] | peterbennett: | markk_: There are a number of changes to support shield mediacodec frame doubling dieinterlace. |
| [13:39:48] | peterbennett: | markk_: There is a bug recently discovered with the frame doubling – try the patch in #13419 |
| [13:39:48] | ** MythLogBot https://code.mythtv.org/trac/ticket/13419 ** | |
| [14:28:05] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Ping timeout: 255 seconds) | |
| [14:45:51] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [14:46:36] | stuarta: | markk_: can we not just use egl when we find wayland in use? |
| [14:47:17] | stuarta: | tbh, i've not seen a single desktop system running wayland that doesn't support egl |
| [14:47:43] | stuarta: | or am i confusing egl with opengles ? |
| [14:48:57] | stuarta: | ah, egl is underneath opengles |
| [14:52:22] | markk_: | stuarta: I think most drivers now support egl. it's a bit of a chicken and egg problem. You have to force egl before the QGuiApplication is created. but the user may have no interest in vaapi – e.g. I don't know yet whether vdpau works with egl? |
| [14:52:37] | stuarta: | vdpau i can test |
| [14:53:08] | markk_: | that said, I've not even checked whether setting the environment variable will work after the app has started – but I don't see why not. |
| [14:53:40] | stuarta: | markk_: what happens if we attempt to create egl, and it fails, can we then fallback to what we currently do? |
| [14:53:59] | markk_: | there is so much special casing in Mythfrontend::main.cpp – really need to move it out into some platform classes |
| [14:54:24] | stuarta: | there is certainly plenty of cleanup todo. |
| [14:54:45] | markk_: | stuarta: yes – probably. I honestly don't expect there to be an issue with forcing egl. In theory it should perform better as well... |
| [14:55:05] | stuarta: | should also be available on embedded platforms? like rpi? |
| [14:56:29] | markk_: | yes – android, pi etc all have egl. I'll test what happens with osx – which has ancient gl drivers |
| [14:58:07] | markk_: | god knows about windows:) |
| [14:59:59] | markk_: | peterbennett: patch applied. can you point me to where mediacodec double rate is being handled – I just want to make sure I'm not missing anything. |
| [15:06:03] | stuarta: | markk_: right now, i think windows as a platform is irrelevant |
| [15:06:15] | dragonian: | ^^^ |
| [15:06:17] | stuarta: | if anybody gives a shit about it, they can resurrect it |
| [15:08:02] | markk_: | stuarta: I will give it a go at some point. the problem isn't getting it to run – it's getting the bloody thing to compile:) all the libraries etc |
| [15:08:15] | stuarta: | hahahahahaha |
| [17:14:28] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [17:14:41] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [17:16:11] | markk_: | peterbennett: nevermind – all sorted now. |
| [17:17:00] | markk_: | mediacodec surface is now as smooth as a baby's behind:) and it doesn't crash on exit. just don't try any seeking. or stream changes... |
| [17:17:50] | markk_: | anyone who wants to test – you need the devel/render-2019 branch for both packaging and the main source repo |
| [17:18:08] | markk_: | would be interesting to hear how it performs on firestick |
| [17:42:15] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Ping timeout: 252 seconds) | |
| [17:48:49] | natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has joined #mythtv | |
| [17:48:49] | Mode for #mythtv by ChanServ!ChanServ@services. : +v natanojl | |
| [17:54:35] | Merlin83b (Merlin83b!~Daniel@office.34sp.com) has quit (Quit: Leaving) | |
| [18:49:59] | gigem: | mark_: What happens if I try to seek? |
| [18:52:16] | willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has quit (Quit: Do your hobbies) | |
| [18:58:44] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has joined #mythtv | |
| [18:58:44] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [19:08:14] | gigem: | mark_: It doesn't look like you've pushed the devel/render-2019 branch for the packaging repo. |
| [19:10:41] | peterbennett: | gigem: I found that branch in packaging ... |
| [19:10:54] | peterbennett: | Building for fire stick now |
| [19:11:22] | gigem: | Hmm, git branch -r didn't show it for me. I'll look again after I finish building latest master. |
| [19:18:35] | gigem: | markspieth: On a related note, how would you (and others) feel about putting packaging/android in the main repo somewhere? Dealing with two repos per branch is a royal pita. |
| [19:21:06] | peterbennett: | I don't really like that idea |
| [19:24:38] | gigem: | Why? FWIW, I was polite in my statement. I absolutely abhor having the packaging repo separate. |
| [19:25:07] | markspieth2: | gigem: I thought platform specific build and packaging was separate. then you should move the rest too. modified libs arent really part of myth code but the stuff in android package source is. a tree in main repo with this stuff. I just followed the win build schema in already in place |
| [19:26:06] | markspieth2: | not sure though |
| [19:27:13] | gigem: | We don't have the libraries in the package repo — only patches. That's a common practice to include patches for external code. |
| [19:28:44] | markspieth2: | libs are normally provided but the host builder. cross builders provide host equivalent infrastructure. not part of myth core. |
| [19:28:49] | gigem: | My belief was that packaging was separate to keep the package maintainers from having commit access to the main repo. That's a bogus reason as far as I'm concerned. |
| [19:31:11] | markspieth2: | its a policy decision on architecture. if its merged do it for all. mythplugins was merged too at one point. |
| [19:31:31] | gigem: | I know I'm biased, but I strongly feel the android support for mythtv is already important and continue to grow in importance and should be considered as a first class component. |
| [19:33:52] | markspieth2: | so is deb and rpm packaging artefacts. |
| [19:34:27] | peterbennett: | It would mess me up because I do a git clean -Xfd on mythtv. That would wipeout all teh libraries |
| [19:35:36] | markspieth2: | that is manageable though. place the cross build support elsewhere maybe |
| [19:36:11] | peterbennett: | Maybe the library build and the various work directories could be outside the git directory |
| [19:38:10] | markspieth2: | i also like concurrent builds, host and android in separate build trees, which is why I implemented out of tree build dir for android. |
| [19:39:34] | markspieth2: | difficult in myth with the externals |
| [19:41:35] | peterbennett: | markk_: Playback on fire stick HD (not 4K) is much improved but still not perfect. There are lots of "MediaCodecInterop: Timed out waiting for frame update" messages and the action is somewhat jerky. I think maybe it is losing frames when thet time-out occurs. Can I increase the timeout? |
| [19:42:56] | peterbennett: | markk_: Also there seems to be no deinterlacing support. The fire stick devices do not deinterlace – currently on fire stick 4lk we are using the opengl deinterlace methods. |
| [19:45:01] | peterbennett: | markk_: Actually very few frames being dropped, before 50% or more frames were dropped, now it is reporting one frame dropped every second or so, much better |
| [19:51:46] | peterbennett: | markk_: Lots of frames dropped with 50fps content – probably to be expected with the base fire stick. |
| [19:52:22] | peterbennett: | markk_: I accidentally hit skip forward – system hangs :( |
| [19:59:02] | gigem: | One of the changes on my TODO list (that I don't expect to get to) is allow the android libraries to be placed in a shared area. |
| [19:59:10] | Warped (Warped!~Warped@unaffiliated/warped) has quit (Read error: Connection reset by peer) | |
| [20:00:06] | Warped (Warped!~Warped@unaffiliated/warped) has joined #mythtv | |
| [20:01:42] | gigem: | Okay, I give up. Where are you guys getting the packaging render branch? git branch -r doesn't list it for me. My git remote is git@github.com :MythTV/packaging.git. What is yours? |
| [20:02:16] | peterbennett: | git branch -r lists it |
| [20:02:45] | peterbennett: | origin git@github.com :MythTV/packaging.git (fetch) |
| [20:03:07] | peterbennett: | Did you do "git pull" ? |
| [20:04:49] | gigem: | Ah, now it's there. Why is the pull needed? I thought the whole point of the -r is to go and check the origin. Stupid git! :( |
| [20:05:43] | peterbennett: | no – you need to do git fetch or pull to get it |
| [20:09:00] | peterbennett: | mark_: I believe shield mediacodec is unusual in that it does deinterlacing automatically – most android devices will need a separate deinterlace. |
| [20:17:08] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has joined #mythtv | |
| [20:21:50] | markk_: | peterbennett: thanks for testing. yes – it tends to lock up when seeking:) re deinterlacing – unfortunately that can't be done with surface rendering. The texture returned to us is RGBA – too late for use to do anything other than display it:( no colourspace control, filtering etc. As I understand it, interlaced content just isn't a 'target' for most android devices. |
| [20:24:18] | peterbennett: | markk_: Probably not even "one field" is possible? |
| [20:27:58] | markk_: | hrm – onefield might work. will need a new shader though. I'll put that on the todo list |
| [20:28:30] | gigem: | peterbennett: I thought Amazon added deinterlacing support late last year. I'm pretty sure that's what the HDHomeRun app users reported. |
| [20:32:08] | peterbennett: | gigem: If they did, ffmpeg did not expose that functionality |
| [20:41:31] | gigem: | peterbennett: https://forum.silicondust.com/forum/viewtopic.php?t=70123 and https://forum.silicondust.com/forum/viewtopic . . . mp;start=120 . The latter includes something about a new fix coming in q2. |
| [20:43:50] | toddejohnson (toddejohnson!~toddejohn@50.196.222.100.tjsquared.com) has quit (Quit: WeeChat 2.3) | |
| [20:51:29] | markk_ (markk_!~mark@host109-149-186-157.range109-149.btcentralplus.com) has quit (Ping timeout: 255 seconds) | |
| [20:53:05] | peterbennett: | gigem: AFAIK those refer to the fact that on fire stick 4K the system crashes if it finds interlaced content. I believe they will call it fixed when it no longer crashes, but that may not mean it will be deinterlaced. |
| [20:54:02] | peterbennett: | gigem: On the fire stick non-4k the interlaced content works without crashing but is still interlaced. |
| [20:55:40] | peterbennett: | markk_: one problem may be that you do not know if the content is interlaced coming out of mediacodec so you don't know to deinterlace it. |
| [20:59:32] | KdW (KdW!~klaas@D93E7846.cm-28.dynamic.ziggo.nl) has quit (Remote host closed the connection) | |
| [21:16:02] | gigem: | peterbennett: I don't think you read it all correctly. |
| [21:27:35] | markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [21:31:10] | markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [21:31:10] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth2 | |
| [21:44:31] | peterbennett: | gigem: TL;DR |
| [21:44:39] | peterbennett: | I admit it |
| [21:46:07] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:e6b7:3488:f9bc:a00f) has quit (Quit: Konversation terminated!) | |
| [21:49:19] | gigem: | peterbennett: :) If I have time (and remember), I'll run the HDHR app on my stick4k. I almost never use it but got one last Christmas just to try and have as a backup to my shields and chromebox. |
| [21:59:42] | SteveGoodey (SteveGoodey!~steve@host86-145-19-224.range86-145.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [22:02:24] | peterbennett: | gigem: I am using the 4k stick every evening for watching and it is working well, using software decode for mpeg2 and mdeiacodec for the rest (h264). |
| [22:32:19] | gigem: | peterbennett: I know you use yours. It can't cut it for me in normal use. I timestretch too much and also have some interlaced, h.264 content. |
| [22:42:59] | ooshlablu (ooshlablu!~ooshlablu@2601:18d:4600:5f31:d1b4:b6a1:1a4c:7266) has quit (Remote host closed the connection) | |
| [22:43:43] | ooshlablu (ooshlablu!~ooshlablu@98.217.124.239) has joined #mythtv | |
| [23:13:01] | peterbennett (peterbennett!~pi@mythtv/developer/peterbennett) has quit (Quit: WeeChat 1.6) | |
| [23:14:49] | peterbennett (peterbennett!~pi@2601:183:101:1dd4:e6c8:b90c:eb45:8cbe) has joined #mythtv | |
| [23:14:49] | peterbennett (peterbennett!~pi@2601:183:101:1dd4:e6c8:b90c:eb45:8cbe) has quit (Changing host) | |
| [23:14:49] | peterbennett (peterbennett!~pi@mythtv/developer/peterbennett) has joined #mythtv | |
| [23:14:50] | Mode for #mythtv by ChanServ!ChanServ@services. : +v peterbennett | |
| [23:21:36] | natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 252 seconds) | |
IRC Logs collected by
BeirdoBot.
Please use the above link to report any bugs.