:: #mythtv

Daily chat history

Current users (57):

aloril, amessina, Anssi, blahdodo, brfransen, bumblebeebat12, ChanServ, clever, davic, DevMythBuild9, dym, eharris, ElmerFudd, enyc, fjear, frobnic, gary_buhrmaster, ghoti, gregl, GreyFoxx, Hydr0p0nX, ijc, jheizer, jpabq, jpharvey_, jya, kalamaja, knowledgejunkie, libsci, mad_enz, Maliuta_, markspieth, mark___, MitchCapper, mkbloke, MythBuild, MythLogBot, MythNotifyBot_, nephyrin, ooshlablu, Panic, peper03, poptix, pppingme, ramshadow, rhpot1991, RokLobsta, sdrik, sphery, stuarta, taylorr, tonsofpcs, tris, Warped, warpme_, zbot, _charly_
Friday, January 4th, 2019, 00:45 UTC
[00:45:55] jheizer (jheizer!~jon@2601:246:8200:498c:dfe:d493:3c6f:6a7b) has joined #mythtv
[00:45:55] Mode for #mythtv by ChanServ!ChanServ@services. : +v jheizer
[00:49:27] mark___ (mark___! has quit (Ping timeout: 246 seconds)
[00:50:48] mark___ (mark___! has joined #mythtv
[00:55:31] natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 246 seconds)
[01:34:07] hampton (hampton!~dhampton@mythtv/developer/hampton) has quit (Ping timeout: 240 seconds)
[01:36:47] amessina (amessina!~amessina@unaffiliated/amessina) has quit (Remote host closed the connection)
[01:48:08] hampton (hampton! has joined #mythtv
[01:48:14] hampton (hampton! has quit (Changing host)
[01:48:15] hampton (hampton!~dhampton@mythtv/developer/hampton) has joined #mythtv
[01:48:15] Mode for #mythtv by ChanServ!ChanServ@services. : +v hampton
[05:04:39] RokLobsta (RokLobsta! has quit (Remote host closed the connection)
[07:12:04] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has joined #mythtv
[07:12:05] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[07:57:22] Tobbe5178 (Tobbe5178!~asdf@2001:2002:51eb:d24e:847a:c28e:4a54:829c) has joined #mythtv
[08:19:04] Tobbe5178 (Tobbe5178!~asdf@2001:2002:51eb:d24e:847a:c28e:4a54:829c) has quit (Read error: Connection reset by peer)
[08:19:55] Tobbe5178 (Tobbe5178!~asdf@2001:2002:51eb:d24e:847a:c28e:4a54:829c) has joined #mythtv
[08:49:19] stuarta: mark___: as gigem said, plan is to cut the branch on 7th (monday), and do the release the following monday (14th)
[08:53:41] Merlin83b (Merlin83b!~Daniel@2a00:1ee0:3:1337:ec33:6ec8:9955:17b7) has joined #mythtv
[09:09:11] SteveGoodey (SteveGoodey! has joined #mythtv
[09:09:11] Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey
[09:24:05] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has quit (Quit: Konversation terminated!)
[11:04:53] MythBuild: build #1200 of master-debian-jessie-64bit is complete: Failure [failed compile core] Build details are at . . . /builds/1200 blamelist: Roger Siddons < >
[11:08:27] nephyrin (nephyrin!~neph@2601:600:817f:f46a:e41d:6f5c:e3be:b8ad) has quit (Ping timeout: 250 seconds)
[11:09:45] nephyrin (nephyrin!~neph@2601:600:817f:f46a:e41d:6f5c:e3be:b8ad) has joined #mythtv
[11:13:25] MythBuild: build #1159 of master-debian-jessie-32bit is complete: Failure [failed compile core] Build details are at . . . /builds/1159 blamelist: Roger Siddons < >
[11:41:04] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has joined #mythtv
[11:41:04] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[11:49:32] stuarta: wish i knew why we don't get branch info from the ubuntu packages
[11:56:27] RokLobsta (RokLobsta! has joined #mythtv
[12:22:21] warpme_ (warpme_! has joined #mythtv
[12:29:58] warpme_ (warpme_! has quit (Quit: warpme_)
[12:35:01] stuarta: trac-- stop breaking on pretty much every webhook from github
[12:39:58] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[12:40:34] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[12:44:57] RokLobsta (RokLobsta! has quit (Ping timeout: 268 seconds)
[12:48:25] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[12:48:54] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[12:50:42] warpme_ (warpme_! has joined #mythtv
[12:51:34] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[12:52:06] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[12:53:31] stuarta: with the new builder i've swapped the doxygen run to nightly only (@3:15am GMT) and cppcheck to every 6 hours if required
[13:00:31] mark__ (mark__! has joined #mythtv
[13:04:14] mark___ (mark___! has quit (Ping timeout: 258 seconds)
[13:05:27] stuarta: mark__: is there no way of getting the FFmpeg api to deliver decoded frames straight to the output stage?
[13:05:47] stuarta: without a GPU -> CPU round trip?
[13:06:26] stuarta: and fyi, i did look at extending vaapicontext so it understood wether it was running under wayland or xorg
[13:06:48] stuarta: tbh there is some weird modelling internally there
[13:07:52] stuarta: there's a MythDisplay, which should be the one which holds the state of play about the display, but it ends up only being used by certain bits of code, and MythXDisplay is the boss, but it's getting called multiple times, and you also get the same multiple calls to vainitialize
[13:08:11] stuarta: to say it's horrible is an understatement
[13:10:45] mark__: stuarta: much of the display stuff in VAAPIContext was simply Daniel and myself trying to work around VAAPI instabilities. Pretty sure most of that can go. and once converted to new API, VAAPIContext will be minimal.
[13:11:08] stuarta: ok
[13:11:40] mark__: the FFmpeg default is to have frames in GPU memory
[13:12:12] stuarta: does it make more sense to have MythDisplay be the boss, which then works out whether it's talking to wayland or xorg and uses {MythXDisplay, MythWlDisplay} as appropriate
[13:12:52] stuarta: i was sure with vaapi you could just pump the decoded frames out to the display?
[13:13:34] stuarta: and unless i miss understood the ffmpeg api, you could manage all of that through ffmpeg
[13:14:21] mark__: stuarta: not sure about MythDisplay- would have to remind myself. At the end of the day – you need to figure out what the underlying display is and create a VAAPI display appropriately.
[13:15:56] stuarta: agreed
[13:16:09] mark___ (mark___! has joined #mythtv
[13:16:35] stuarta: things were originally coded when there was only X, so there are a lot of assumptions that it's running X
[13:17:28] stuarta: MythDisplay class is newer, and uses more of the Qt5 functionality, but is only a "helper" for some other bits of code
[13:19:03] stuarta: iirc
[13:19:09] mark__ (mark__! has quit (Ping timeout: 246 seconds)
[13:19:10] stuarta: been ages since i dug around in this
[13:19:42] mark___: stuarta: decoded VAAPI frames are in GPU memory. and memory is just memory – but we just need the different APIs to talk to each other. If we want OpenGL output – we need to start with a GLX display and at display time convert the VAAPI pointer to a GLX pointer.
[13:19:55] mark___: Same goes for EGL.
[13:22:42] mark___: I think the only interop that can display frames directly is va/X11 – but then you're back to the old issue of how to draw OSD etc on top of an X11 drawable
[13:24:56] warpme_ (warpme_! has quit (Quit: warpme_)
[13:42:50] stuarta: hmmm, i would love if we could do va/opengl and lose the x11 entirely
[13:46:07] stuarta: which you would doing vaapi via wayland
[13:46:19] stuarta: the lose X11 bit anyway
[13:50:01] mark___: stuarta: VA now has a newish zero copy approach – which I'm struggling to get my head around. Seems to be based mostly on DRM/PRIME.
[13:59:37] mark___: anyone familiar with 'Driver not loaded' error from Qt. mysql driver is found but refuses to work.
[13:59:45] mark___: loaded library "/home/pi/qtbase/plugins/sqldrivers/" – but then fails...
[14:01:20] stuarta: the pi can be a bit of a pain
[14:01:38] stuarta: the broadcom specific bits from the original release, being slowly replaced by fully opensource bits
[14:02:05] stuarta: i suspect we will end up with broadcom bits for OpenMAX and mesa for the free opengl bits
[14:02:09] stuarta: *opengles
[14:02:20] mark___: stuarta: I've got fullscreen mythfrontend up with eglfs – but can't get past the database issue
[14:02:54] stuarta: you shouldn't need to compile it yourself
[14:03:09] stuarta: the ansible playbooks will install everything you need
[14:03:18] stuarta:
[14:04:26] stuarta: # find /usr/lib/ -name
[14:04:26] stuarta: /usr/lib/arm-linux-gnueabihf/qt5/plugins/sqldrivers/
[14:04:32] stuarta: # dpkg -S /usr/lib/arm-linux-gnueabihf/qt5/plugins/sqldrivers/
[14:04:32] stuarta: libqt5sql5-mysql:armhf: /usr/lib/arm-linux-gnueabihf/qt5/plugins/sqldrivers/
[14:04:51] stuarta: that's on my rpi
[14:05:57] mark___: stuarta: but I assume that doesn't build Broadcom eglfs support?
[14:07:30] nephyrin (nephyrin!~neph@2601:600:817f:f46a:e41d:6f5c:e3be:b8ad) has quit (Ping timeout: 268 seconds)
[14:09:45] nephyrin (nephyrin!~neph@2601:600:817f:f46a:e41d:6f5c:e3be:b8ad) has joined #mythtv
[14:10:53] stuarta: dunno. wish i knew how to find out
[14:11:30] stuarta: warpme builds his own qt on top of archlinux to provide his minimyth distro
[14:11:46] stuarta: there is a reason for that. next time he pops in he can tell us
[14:12:50] stuarta: qmake -query doesn't tell me much
[14:14:21] stuarta: mark___: when it fails, is it segfaulting?
[14:16:29] mark___: stuarta: no – just fails to initialise the database and then pops up the database setup/search screen
[14:19:58] stuarta: i'd check the stupid things like, can mysql be accessed from that frontend with that username / pass
[14:22:16] mark___: stuarta: good point
[14:24:11] stuarta: AFAICT from . . . debian/rules eglfs is only disabled on freebsd debian
[14:26:46] stuarta: looking at there are several eglfs integrations built as part of the build
[14:30:52] warpme_ (warpme_! has joined #mythtv
[14:32:49] stuarta: warpme_: hello. remind me again why it is you build your own Qt5 for minimyth?
[14:33:17] warpme_: hi stuarta!
[14:33:59] ** stuarta contemplates officially switching to the new buildbot as the main build master **
[14:34:39] mark___: stuarta: . . . B/qmake.conf
[14:35:09] mark___: and more specifically the line that reads: EGLFS_DEVICE_INTEGRATION = eglfs_brcm
[14:35:13] stuarta: so qt-5.11 isn't in stretch
[14:35:41] ** stuarta switches to 5.7 branch **
[14:35:54] stuarta: . . . ake.conf#L43
[14:35:57] stuarta: \o/ yay
[14:36:16] stuarta: so that's the default on stretch on rpi
[14:36:35] stuarta: +should be
[14:40:06] warpme_: well – i?m doing this as infact I?m cross-builiding myself whole os (only bison make flex gawk mercurial subversion texinfo gcc patch and wget packages are used from host os; rest is built). This may sound crazy – but this perfectly pays-off i.e. in situations were glibc guys in 2.28 removed bunch of functions/headers and building of 20+ key packages fails without quick solution. Goal is to have target os build without single dependency on
[14:40:06] warpme_: os (is is dependent only on build os). regarding qt: on arm platforms imho gles is much more resources efficient than desktop gl – so I?m building gt & myth with gles
[14:40:58] warpme_: gt->qt
[14:41:16] stuarta: warpme_: so mark___ is looking at opengl / opengles
[14:42:49] warpme_: ant this is FANTASTIC. If he will look later in video mapping to GLESv3 textures – then mayba someday myth will compete with kodi on all those 20$ SBC doing 4k@60p HEVEC...
[14:43:57] warpme_: sorry ofr typos – can?t type in confortable env
[14:44:03] stuarta: np
[14:46:46] stuarta: warpme_: okay, i thought you were building it to enable some specific support that wasn't in the pre built qt packages
[14:49:55] hampton: warpme_: Any idea where I would find the QuickTime framework. My mythtv build fails in libs/libmythtv because it can't be found.
[14:50:13] stuarta: hampton: that osx?
[14:50:19] gary_buhrmaster: stuarta: If you need any encouragement, please do upgrade to the new buildbot master infrastructure.
[14:50:27] hampton: yup, osx
[14:50:51] stuarta: gary_buhrmaster: i'm just gunno upgrade the base os to f28 (yay, security updates again) and that gets us a new trac and new buildbot
[14:52:52] stuarta: debian oh debian, why do you revert back to desktop opengl on arm64's ???
[14:52:59] gary_buhrmaster: F28 and not F29? Something specific holding you back?
[14:53:04] stuarta: (for testing / next release)
[14:53:11] stuarta: gary_buhrmaster: it's the next one along
[14:53:26] stuarta: then i can upgrade the dev instance to f29 and see what breaks again
[14:53:57] gary_buhrmaster: It's always something.
[14:53:59] warpme_ (warpme_! has quit (Quit: warpme_)
[14:54:19] stuarta: gary_buhrmaster: yeah, i *want* to go to f29, i've just not been able to test it
[14:54:35] mark___: stuarta: what does qmake -query show?
[14:54:43] stuarta: still have some shit to sort out, like beirdobot won't build on f28
[14:55:06] mark___: for me on raspbian stretch: QMAKE_SPEC:linux-g++
[14:56:07] stuarta: mark___:
[14:56:12] stuarta: same
[14:56:13] gary_buhrmaster: I think beirdobot needs to change from gethostbyname to getaddrinfo (or something like that).
[14:56:26] stuarta: gary_buhrmaster: oh it needs plenty of work
[14:56:46] stuarta: gcc8 errors on `strncat` as being horrible and insecure
[14:56:49] stuarta: fixing it requires work
[14:59:36] hampton: hmm, I think I need an older version of apple's xcode dev tools
[15:02:37] warpme_ (warpme_! has joined #mythtv
[15:12:29] gary_buhrmaster: There is always the possibility of moving from irc to something else for project communications (turning the beirdobot molehill into the slack/teams/ryver/chanty/rocket chat/etc. mountain)
[15:15:12] ** stuarta explodes **
[15:15:47] stuarta: i have a pet hate of migrating to latest shiny communication medium just because it exists
[15:25:21] peterbennett: mark___: Let me know what I can do to help with the OpenGL stuff especially for Android.
[15:25:29] warpme_ (warpme_! has quit (Quit: warpme_)
[15:26:23] blahdodo (blahdodo!~blahdodo@ has quit (Quit: Bye bye)
[15:26:42] blahdodo (blahdodo!~blahdodo@ has joined #mythtv
[15:53:24] hampton: *cough* webex *cough*
[15:54:55] ** stuarta sets fire to hampton **
[15:55:23] stuarta: had it working on my fedora machine, then something changed and it's broken again
[15:55:28] stuarta: steaming pile of java poo
[15:57:07] hampton: I didn't think I needed a <sarcasm> tag. :-)
[15:57:57] stuarta: :-0
[15:58:03] stuarta: </rant>
[15:58:10] ** stuarta chuckles **
[15:58:36] stuarta: although trac is beginning to annoy me more and more
[15:58:46] stuarta: and it's starting to look unmaintained
[15:59:27] stuarta: i'm having to monkey patch it in the ansible playbooks to ensure basic things like mysql db access works (and some others)
[15:59:44] stuarta: trac-1.3 has been under dev for years
[16:18:59] hampton: omg. changing version of xcode causes tools like got to fail. I guess I get to completely rebuild the ports.
[16:20:41] stuarta: you mean you aren't going to make it work under the latest XCode ??? ;-p
[16:21:29] hampton: LOL. I need to make it work at all first. Then I can consider the pain of migrating from Quicktime to whatever replaced it.
[16:21:58] stuarta: oh that's a whole can of worms
[16:25:22] warpme_ (warpme_!~warpme_@ has joined #mythtv
[16:47:28] mark___: peterbennett: thanks. I've been doing some performance testing. All UYVY (non-YV12). Default playback on a 1080i sample gives ~96fps. If I force it to progressive (simpler packing) that rises to ~123fps. If I use FFmpeg that rises again to ~136fps.
[16:48:11] mark___: if I disable MMX – default gives 36fps (interlaced) and 60fps (progressive). (Didn't test ffmpeg without MMX).
[16:49:30] peterbennett: mark___: Must be something different about the android setup
[16:49:38] mark___: So I'm guessing the comparable performance drop on Android is the 136 to 36 :(
[16:50:30] peterbennett: It plays fine with version before your change. I assume we need some very simplified OpenGL process for Android
[16:50:59] mark___: peterbennett: just to be clear. Before you would be getting 136fps and now 36... assuming FFmpeg has some decent NEON optimisations
[16:51:29] mark___: I think it is just the CPU side that is slowing things down.
[16:51:48] peterbennett: mark___: I don't follow
[16:52:05] gigem: stuarta: might be relevant to your Debian/arm64 comment.
[16:52:35] peterbennett: mark___: it was playing fine at 30 fps on android, but now only plays around 10 fps on android
[16:54:17] peterbennett: mark___: The decoding is fine, keeps up, rendering is slowed. So you get dropped frames but the action keeps up with real time.
[16:54:44] warpme_: hampton: regarding missing QuickTime:
[16:55:42] mark___: peterbennett: previously we were using the FFmpeg YV12 to UYVY conversion. That tops out at 136fps here. Now we use our own code (with proper interlaced packing) – but without SIMD support that can only handle 36fps. The 36fps 'mimics' what will be happening on Android with no NEON help. So your 30 to 10 fps drop sounds about right.
[16:57:02] hampton: warpme_: Thanks. I hope that works. I just tried running xcode 8.3 (apparently the last version with Quicktime builtin) and it fails to run with a missing symbol in appkit.
[16:57:25] peterbennett: mark___: I wonder – when it drops a frame is that dropped before or after the UYVY conversion ?
[16:57:31] mark___: peterbennett: I think I'm going to revert to FFmpeg code and add a setting to enable 'high quality chroma sampling'. Post v30 I will also try and improve both YV12 and UYVY shaders to do the sampling properly.
[16:58:35] peterbennett: mark___: OK – If you want I can test the patch before you commit, if you think that is useful.
[16:58:54] warpme_: oh – pls go with current latest xcode on current latest macos. Im using this build script: For me it builds perfectly.
[16:59:09] mark___: peterbennet: before
[16:59:38] mark___: need to head out – will pastebin the patch later
[16:59:58] peterbennett: ok
[17:00:18] stuarta: gigem: that's where i was reading about it
[17:03:58] hampton: warpme_: That appears to compile everything from scratch instead of using macports.
[17:05:27] warpme_: yes. Maybe not everything (qt is prebuild) – but rest is built. I prefer this way as it gives me pure standalone .app with install as drag-n-drop
[17:06:49] mark___ (mark___! has quit (Quit: mark___)
[17:09:21] hampton: ack. Hopefully I can at least get compiles to test my changes. I was also thinking about trying to make this system a buildbot. I'll have to see if I can make that work with this script.
[17:16:47] stuarta: hampton: fyi i'll make sure we have a dev instance of buildbot master we can play with when i do the upgrade
[17:17:08] ** stuarta probably should send an email about that **
[17:41:37] ** stuarta head off **
[17:52:39] MythBuild: build #1203 of master-debian-jessie-64bit is complete: Success [build successful] Build details are at . . . /builds/1203
[17:57:54] Merlin83b (Merlin83b!~Daniel@2a00:1ee0:3:1337:ec33:6ec8:9955:17b7) has quit (Quit: Leaving)
[18:01:00] MythBuild: build #1162 of master-debian-jessie-32bit is complete: Success [build successful] Build details are at . . . /builds/1162
[18:35:13] ** hampton sighs. Apparently jya's website is down so I can't pull a required file. **
[18:36:42] jya: hampton: which site?
[18:37:19] hampton:
[18:38:06] jya: working for me here
[18:38:33] jya: I changed domain name registrar yesterday
[18:38:37] jya: shouldn't have an impact
[18:39:18] jya: hampton:
[18:39:21] hampton: I get redirected to and get a 'service unavailable' page.
[18:40:14] hampton: interesting. works with https but not http
[18:41:27] ** hampton tweaks the build script **
[18:41:42] jya: hampton: likely an error in the apache config
[18:41:53] jya: I have the https everywhere extension installed
[18:42:57] hampton: so do i. wonder why it didn't kick in. The original problem was with curl, so that wouldn't have helped.
[18:43:15] jya: I never use plain http
[18:43:39] jya: hampton: all the java processes (that runs atlassian services) weren't started somehow
[18:43:49] ** hampton is running the osx-build-script for the first time, pulling all the dependancies **
[18:44:14] jya: hampton: i haven't touched it in almost 4 years, likely won't work anymore
[18:44:28] jya: with every new version of Qt or macOS it required extensive changes
[18:44:39] hampton: couple of broken download links so far.
[18:45:15] hampton: I started trying to build with macports, but warpme suggested this script.
[18:45:53] jya: should try with brew
[18:45:57] jya: likely more maintained
[18:46:02] jya: I don't use macport anymore either
[18:46:35] hampton: lol. ask three people, get three different answers.
[18:46:59] jya: with the scripts you need nothing
[18:47:11] jya: so long as you use the version it says to use
[18:47:24] jya: if you update a version, in particular Qt or mysql
[18:47:32] jya: be prepared for a lot of effort
[18:47:39] jya: and don't build Qt
[18:47:44] jya: that will take a day
[18:50:14] jya: it's no wonder doesn't work
[18:50:24] jya: there's no longer a config file for it :(
[19:07:42] hampton: osx-build-script failed building exiv2 with a "symbol not found" error.
[19:07:55] hampton: bree, here I come?
[19:07:57] hampton: brew
[19:14:27] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has quit (Ping timeout: 252 seconds)
[20:01:10] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has joined #mythtv
[20:01:11] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[20:18:44] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[20:19:17] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[20:33:19] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[20:33:47] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[20:35:11] DevMythBuild9 (DevMythBuild9! has quit (Remote host closed the connection)
[20:35:14] RokLobsta (RokLobsta! has joined #mythtv
[20:35:42] DevMythBuild9 (DevMythBuild9! has joined #mythtv
[21:44:00] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has quit (Quit: Konversation terminated!)
[22:22:51] gigem: peterbennett: Did you see my email from today on the -users list regarding transcoding with closed captions (title Ubunto 18.04 / 0.29 export to mkv)? In short, the subtitle (converted from the captions) work on Linux but not on Android.
[22:23:21] peterbennett: gigem: I did not see it
[22:25:18] peterbennett: I see it now :)
[22:26:09] amessina (amessina!~amessina@unaffiliated/amessina) has joined #mythtv
[22:26:43] peterbennett: I know the subtitles are likely a problem with mediacodec. If you set to software decode do they work?
[22:27:28] peterbennett: The subtitle problem affects other H/W decoders also, e.g. OpenMAX
[22:28:10] peterbennett: I have some ideas that do not require modifying ffmpeg but unfortunalety will likley be a lot of wor to get done.
[22:38:38] SteveGoodey (SteveGoodey! has quit (Quit: Konversation terminated!)
[22:49:01] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has quit (Quit: - Chat comfortably. Anywhere.)
[22:54:06] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has joined #mythtv
[22:54:07] Mode for #mythtv by ChanServ!ChanServ@services. : +v jpabq
[23:00:39] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has quit (Quit: - Chat comfortably. Anywhere.)
[23:03:52] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has joined #mythtv
[23:03:53] Mode for #mythtv by ChanServ!ChanServ@services. : +v jpabq
[23:11:37] mark___ (mark___! has joined #mythtv
[23:25:17] Tobbe5178 (Tobbe5178!~asdf@2001:2002:51eb:d24e:847a:c28e:4a54:829c) has quit (Read error: Connection reset by peer)
[23:48:34] gigem: peterbennett: In my test sample, these are proper, subtitle streams not closed captions muxed into the video. The two are related but completely different animals. The former are demuxed separately from the video (and audio). The latter are encoded in the video stream and must be decoded during video decoding. The subtitle streams I have work on Linux whether I used ffmpeg, vaapi or vaapi2 decoding. They
[23:48:36] gigem: do not work on Android whether I use ffmpeg or mediacodec encoding. They should work on all platforms regardless of the video decoding method.
[23:51:27] peterbennett: gigem: The subtitle streams should work
[23:53:14] gigem: Yes. That what I said, I think. Any idea why they wouldn't work like an ffmpeg configuration switch or soemthing?
[23:53:22] peterbennett: gigem: Oh – I think libass is disabled in android build.
[23:53:42] gigem: What is libass?
[23:54:11] peterbennett: gigem: Your mission should you care to accept it is to download libass and include it in the android library build :)
[23:54:24] peterbennett: libadd is a subtitle package used by MythTV
[23:54:29] peterbennett: libass
[23:54:53] peterbennett: I noticed before that the android mythbuid had it disabled.
[23:56:13] gigem: Is it a mythtv dependency and not an ffmpeg one, then? Do you know why it isn't included? Won't build on Android? I'll try to look at it later.
[23:58:26] peterbennett: 1. Correct 2. I think Mark tried not to have too many libraries needing to be built, and I sympathize with that.
[23:59:39] peterbennett: I see mheg is also disabled, that may upset the 2 people who actually use it :)
[23:59:53] peterbennett: (snide remark)

IRC Logs collected by BeirdoBot.
Please use the above link to report any bugs.