:: #mythtv

Daily chat history

Current users (40):

Anssi, brfransen, ChanServ, davic, dym, ghoti, gigem, gregbert, gregl, hampton, ijc, jpabq, jpharvey, jya, knowledgejunkie, kwizart, libsci, mad_enz, markspieth, MitchCapper, mkbloke, MythBuild, MythLogBot, MythNotifyBot, M_kveef, ooshlablu, Panic, peper03, peterbennett, poptix, pppingme, sdrik, sphery_, taylorr, tgm4883, tonsofpcs, tris-, Warped, xris, _charly_
Friday, October 19th, 2018, 01:58 UTC
[01:58:13] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has joined #mythtv
[01:58:13] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[02:40:24] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has quit (Quit: Konversation terminated!)
[03:23:23] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 252 seconds)
[06:12:56] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has joined #mythtv
[06:12:56] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[08:01:02] SteveGoodey (SteveGoodey!~steve@2a00:23c5:7da3:4501:7a24:afff:fe9d:c233) has joined #mythtv
[08:01:02] Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey
[08:13:20] stuarta: peterbennett: yes, NVDEC is the replacement for VDPAU
[08:18:16] stuarta: peterbennett: but as kwizart said the VDPAU interface is being implemented by various OSS backends, so we should continue to support it
[08:18:53] SteveGoodey: Hmm, that's a bummer. Although fixes/29 PPA is showing the same update as MythTV commits it's not appearing on my box. Still getting ubuntu updates though.
[08:28:33] SteveGoodey (SteveGoodey!~steve@2a00:23c5:7da3:4501:7a24:afff:fe9d:c233) has quit (Quit: Konversation terminated!)
[08:32:32] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has joined #mythtv
[08:43:30] bas-t (bas-t! has joined #mythtv
[08:43:31] warpme_ (warpme_! has joined #mythtv
[08:44:53] bas-t (bas-t! has quit (Quit: Leaving)
[08:57:14] warpme_: morning all!
[08:57:33] warpme_: guys – in need second pair of eyes.
[08:58:10] warpme_: I?m trying to fix #13265
[08:58:10] ** MythLogBot **
[09:04:14] bas-t (bas-t! has joined #mythtv
[09:05:34] bas-t (bas-t! has quit (Client Quit)
[09:05:54] warpme_: inspecting related code is typo in code. Namelly: ed85b14 add in all signalmonitors bool _release_stream as 3rd params. But in signalmonitor.cpp line226 it is declared as 4th param. the same in signalmonitor.h line 121. This may explain loss of SN/BER in OSD as now signalmonitor flags are wrongly set to _release_stream variable value....
[09:06:29] warpme_: inspecting related code is typo in code -> inspecting related code probably there is typo in code
[10:43:03] Warped (Warped!~Warped@unaffiliated/warped) has quit (Quit: ChatZilla 0.9.92-rdmsoft [XULRunner 35.0.1/20150122214805])
[11:01:02] Warped (Warped!~Warped@unaffiliated/warped) has joined #mythtv
[11:15:53] MythBuild: build #971 of master-fedora-armv7hl is complete: Failure [failed unit test core] Build details are at . . . l/builds/971 blamelist: David Hampton < >
[11:32:41] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has quit (Ping timeout: 250 seconds)
[11:38:54] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has joined #mythtv
[11:42:50] ramshadow (ramshadow! has quit (Remote host closed the connection)
[12:21:10] warpme_ (warpme_! has quit (Quit: warpme_)
[12:31:56] warpme_ (warpme_! has joined #mythtv
[12:44:26] hampton: warpme_: I see the same. DTVSignalMonitor::DTVSignalMonitor creates a SignalMonitor object with the arguments in the wrong order and the compiler doesn't catch it.
[12:46:23] warpme_: hampton: thx!. Interesting that compiler is silent on this. I verified – fixing code fixes #13265. Good :-)
[12:46:23] ** MythLogBot **
[12:49:03] warpme_: I have Q related to Qt. In tv_play.cpp I want to use QtConcurrent. I add #include "<QtConcurrent>" in tv_play.cpp but receivig tv_play.cpp:40:10: fatal error: '<QtConcurrent>' file not found. Probably I?m missing something very obovious...
[12:52:42] warpme_: maybe I need to add Concurrent in at L155?
[12:53:11] hampton: You probably need to add /usr/include/qt5/QtConcurrent to the include path, or include QtConcurrent/QtConcurrent.
[12:56:12] warpme_: right. include QtConcurrent/QtConcurrent works. thx!
[13:02:52] hampton: The other (probably better) route would be to add 'concurrent' to the 'QT +=' lines in the appropriate *.pro files.
[13:15:46] hampton: The compiler must be implicitly converting bool <-> uint64.
[13:22:19] stuarta: certainly fits ;-)
[13:24:51] hampton: Are you fixing 13265? You can force the compiler to not do implicit conversion by adding an overload with the parameters in the wrong order. I.E. SignalMonitor(int, ChannelBase *, bool, uint64_t) = delete;
[13:28:28] warpme_: hampton: yes. I fixed #13265 by fixing params list in signalmonitor.cpp and signalmonitor.h. pls see:
[13:28:28] ** MythLogBot **
[13:39:01] warpme_ (warpme_! has quit (Quit: warpme_)
[13:47:47] hampton: I was going to add an author line to the commit, but I need an email and he's gone. :-(
[13:49:16] stuarta: ah that's easy
[13:50:15] stuarta: git log | grep -i warpme
[13:50:32] hampton: Thanks.
[13:51:21] MythBuild: build #972 of master-fedora-armv7hl is complete: Success [build successful] Build details are at . . . l/builds/972
[13:54:41] stuarta: hampton: the slower builders such as the rpi's sometime fail the unit tests because they take too long to run the relevant step
[13:55:16] hampton: Right. This failed because it was expecting a response in less that some amount of time.
[13:55:35] stuarta: yep. i guess the fix would be to increase the timeout
[13:55:52] stuarta: dunno if that invalidates the concept of the test in question
[13:59:51] hampton: The test fires off a one second timer into the background, and then checks to see if creating the timer took less than 1/2 second. I could double both of those values and still have a valid test.
[14:00:22] hampton: err, a one second sleep
[14:00:57] stuarta: :)
[14:02:54] hampton: I need to test my tv_play.cpp fix for "a temporary bound to x only persists until the constructor exits." The latest clang has promoted this from a warning to an error. :-(
[14:03:47] stuarta: tv_play.cpp is a beast
[14:17:08] gregbert (gregbert!~gregbert@unaffiliated/gregbert) has quit (Ping timeout: 276 seconds)
[14:38:55] peterbennett (peterbennett!~Peter_Ben@mythtv/developer/peterbennett) has left #mythtv ()
[14:39:54] gregbert (gregbert!~gregbert@unaffiliated/gregbert) has joined #mythtv
[14:41:44] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has quit (Quit: - Chat comfortably. Anywhere.)
[14:42:40] peterbennett (peterbennett!~Peter_Ben@mythtv/developer/peterbennett) has joined #mythtv
[14:42:40] Mode for #mythtv by ChanServ!ChanServ@services. : +v peterbennett
[14:45:54] peterbennett: In the latest Raspberry pi raspbian the combination of openmax video and opengles osd no longer works, so I think it is time to get that raspberry pi experimental opengl support working instead
[14:46:08] stuarta: long term that's the way forward
[14:54:03] jpabq (jpabq!~quassel@mythtv/developer/jpabq) has joined #mythtv
[14:54:03] Mode for #mythtv by ChanServ!ChanServ@services. : +v jpabq
[14:54:38] warpme_ (warpme_! has joined #mythtv
[15:03:01] hampton: Hmm. I had to resend e60f71bd0f, but even after that trac didn't close the bug.
[15:07:52] stuarta: trac is being terrible at the moment
[15:08:04] stuarta: multiple resends before it picks it up
[15:08:09] stuarta: anything up to 10
[15:08:36] hampton: I resent one from last night and one from this morning. I think it took on the third try for both.
[15:13:05] stuarta: it's just painful
[15:13:35] stuarta: i'm going to have to write a commit wrangler, to handle the github webhook, and then beat trac up
[15:13:54] stuarta: tbh. i think the problems come from the way trac does thing synchronously
[15:14:07] stuarta: so the bigger todo, is add async webhook processing to trac.
[15:14:30] stuarta: right, i'm off
[15:14:35] stuarta (stuarta!~stuarta@mythtv/developer/stuarta) has quit (Quit: holiday)
[15:25:58] warpme_ (warpme_! has quit (Quit: warpme_)
[16:45:30] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has quit (Quit: Do your hobbies)
[16:46:07] warpme_ (warpme_! has joined #mythtv
[16:53:24] natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has joined #mythtv
[16:53:24] Mode for #mythtv by ChanServ!ChanServ@services. : +v natanojl
[16:59:38] warpme_ (warpme_! has quit (Quit: warpme_)
[17:20:28] warpme_ (warpme_! has joined #mythtv
[17:22:07] warpme_: hampton: FYI: recent master commits are breaking macOS compile:
[17:25:42] warpme_: peterbennet: FYI: lastly I was testing minimyth2 arm7 and aarch64 builds on rpi2/rpi3 with current mesa + mainline 4.18 kernel using OpenGL painter. To my surprice all OpenGL UI and OpenGL playback works really nice. what is missing is video HW decode. I?m dreaming that somebody will provide VC4 VDPAU someday
[17:31:23] warpme_: but I see light in tunel with virgl where You can have 2 GL provides in the same screen. I need to investigate archlinux arm. It uses virgl and I suspect it can offer UI GL provided by mesa and videoplayback by broadcom OMX blobs (this needs to be verified!!!)
[17:35:28] hampton: That paste doesn't make sense to me. I don't see clock_nanosleep anywhere in mythplayer.cpp, or anywhere in the MythTV sources for that matter.
[17:35:34] hampton: All my last change to that file did was replace old C include files with their supposed C++ equivalent.
[17:36:55] warpme_: hampton: wait a sec: maybe this is issue of #13334 I?m working with peterbennet
[17:36:55] ** MythLogBot **
[17:37:29] warpme_: now building macOS without #13334
[17:37:29] ** MythLogBot **
[17:37:43] hampton: Perhaps one of the "equivalent" C++ headers on MacOS didn't include time.h? Try adding "#include <time.h>" to the top of the file and see if it helps.
[17:38:18] warpme_: sure. give me sec. need to waith till current build ends
[17:56:11] warpme_: hampton: indeed – macOS compile issue was by #13334. Unfortunatelly patch for #13334 already has "#include <time.h>"
[17:56:11] ** MythLogBot **
[17:56:11] ** MythLogBot **
[18:03:34] peterbennett: Patch in #13334 fails because time.h problem, caused by other commits
[18:03:34] ** MythLogBot **
[18:04:06] peterbennett: I will attach a new patch for #13334 that works.
[18:04:06] ** MythLogBot **
[18:09:54] peterbennett: warpme_: The latest raspbian breaks my openmax vidoe output so I am planning to look at opengl for raspberry pi with openmax decode.
[18:10:55] peterbennett: warpme_: See #1331
[18:10:55] ** MythLogBot **
[18:11:03] peterbennett: sorry #13331
[18:11:03] ** MythLogBot **
[18:12:38] peterbennett: warpme_: What is VC4 VDPAU?
[18:14:04] peterbennett: warpme_: The AVSYNC2 is working with 2017 version of raspbian, but I found yesterday it was failing with 2018 version of raspbian.
[18:15:12] warpme_: peterbennett: i was referring to software driver exposing VC4 hw (videocore4 video decoder) via VDPAU API (exactly like nvidia blobs does for nvidia hw and mesa vdpau for ati hw)
[18:16:24] peterbennett: warpme_: If ffmpeg supports it, it could be easy, otherwise not so easy.
[18:18:05] warpme_: IIRC there is work for allwinner hw (cedrus project). But honestly – I don?t believe Eben will provide such thing for rpi. If there will be vdpau for rpi – then I suspect it will be rather for rpi4 with vc6
[18:20:07] warpme_: peterbenett: indeed ffmpeg route is interesting. IIRC ffmpeg can use OpenMX IL
[18:20:58] peterbennett: warpme_: Unfortunately it is only for encoder not decoder
[18:21:16] warpme_: right. maybe this . . . acceleration
[18:21:42] warpme_: right also for encoder
[18:22:00] peterbennett: Why does anybody want to encode on raspberry pi?
[18:23:26] peterbennett: I already checked it here where it shows OpenMAX for encoding only
[18:25:14] warpme_: i think for saving bw from attached hd camera. after all raw bw from hd camera easily can exceed 100mbps of rpi LAN
[18:25:48] peterbennett: yes that makes sense
[18:30:33] warpme_: peterbennet: i think good demo what is possible with open software + max exploiting of rpi hw is archlinux arm. i remember when lastly I was checking how myth is working on archlinux arm – I was able to install vanilla archlinux on rpi3, download myth packages and got fully working myth. When I checked video stack I discover that GL is provided by mesa (!!!) (not by brcm blobs). that was surprice. investigation shows me that this was because
[18:30:33] warpme_: nice things like virgl
[18:32:11] warpme_: i?m really tempted to switch minimyth to virgl and play a little to get mesa opengl + omx il for video decode
[18:37:46] warpme_: peterbennet: did you investigate who is GL provider for Qt in latest raspbian (this with broken openmax)? I suspect it is mesa – not brcm blobs
[18:37:59] peterbennett: That sounds interesting – I am planning to try with the experimental OpenGL in raspbian
[18:40:02] peterbennett: There is what is called "experimental" opengl that you have to activate with a config change – that switches the opengl from the fullscreen opengles to a xwindow based opengl which I think is the one that qt is built with.
[18:41:06] peterbennett: I looked at the libraries with the broken openmax and it looks like all teh same libraries, even te same lengths.
[18:48:15] warpme_: ?xwindow based opengl which I think is the one that qt is built with? – sounds to me like gl provider is mesa. You may run ldd on qt libs realted to ql to see who provides gl for qt. I believe in past – in debian world – it was almost always brcm blobs (arch very early started using mesa; that?s why I was building myself qt for arch I for eglfs only brcm is choice). Now – when mesa18 by eric Anholt fantastic job with v3d (mesa support for vc
[18:48:15] warpme_: starts to offer working gl on vc4 – dostros starting to switch gl provider to mesa
[18:50:40] warpme_: i tested mesa18.2.2 on rpi2/rpi3 and gl provided by mesa is really working well (I mean fully working UI with gl painter and gl playback). this is VERY promissing. now – if we will able to do video playback with hw access by ffmpeg – them we are in home!
[18:50:54] peterbennett: I never used arch linux, is it easy?
[18:51:55] peterbennett: Does arch linux have a package system with qt packages and others or do you have to build them yourself?
[18:52:14] warpme_: ough – for me it is BEST distro ever (sorry for marketing bs). pls look
[18:52:42] warpme_: for rpi3 pls look here
[18:55:37] warpme_: at bottom you have ?AArch64 Installation? – this is arch with vanilla latest kernel + latest mesa. this is good demo what pure FOSS can do on rpi. (I?m experimenting with pure foss on rpi by minimyth2 – where I have only free upstream and now have fully working myth with gl painter and gl playback)
[18:57:36] peterbennett: It uses pacman? So we need a new type of package than for debian.
[19:01:36] warpme_: right. If You need – I have full insfrastructure as I?m reguallry building myth for my server (which is arch based) and also building in past minimyth2 rpi edition (when minimyth2 was based on arch; now it is my system build from scratch like minimyth2 x68)
[19:02:56] warpme_: and pacman in opinion of many ppl is best package manager in linux world. I personaly – after playin with pacman – never look back on any other package manager
[19:04:06] peterbennett: Does digital audio work with archlinux (dolby pass through)? This also stopped working with the new raspbian from 2018.
[19:05:08] peterbennett: Also my AVSYNC2 goes crazy with the 2018 raspbian, but works perfectly with the 2017 raspbian. ALtogether a bad experience.
[19:05:22] warpme_: well – i don?t know as my tests was always with analog audo.
[19:06:42] peterbennett: With raspbian there was a command that created an IEC958 device in the list of audio devices and that works with digital audio. Now that IEC958 device is not created.
[19:10:40] peterbennett: I see that arch linux has separate builds for rpi1,2 and 3 so that seems better so it can take advantage of features
[19:13:08] warpme_: yes. diferentiation is on gcc options used at distro compiling level and kernel config level
[19:18:36] warpme_: arch is known by strategy with minimal modifications in upstream sources (patched usually only if compile requires mods or integration with other component needs it). For me this is most important aspect when compared to ubuntu (imho ubuntu has ambition customize sources as heavy as possible as ubuntu differentiation strategy is based on making upstream sources modifications called ?improvements?).
[19:31:40] peterbennett: warpme_: your vaapi 2x problem – I do not have this happening but I will take a look and see if there is something that could go wrong with the 1x and 2x selector of deinterlace to cause this back and forth toggle.
[19:34:06] warpme_: peterbennett: perfect. i?m on your disposal :-)
[19:35:07] warpme_: regarding arch – pls feel free to ask. I?ll be more than happy to share with you my expierience with arch (on x86 and arm)
[19:37:24] warpme_ (warpme_! has quit (Quit: warpme_)
[20:00:35] warpme_ (warpme_! has joined #mythtv
[20:19:57] warpme_ (warpme_! has quit (Quit: warpme_)
[20:29:37] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:7a84:3cff:fedf:a99) has quit (Quit: Konversation terminated!)
[20:35:53] natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 245 seconds)
[20:38:07] warpme_ (warpme_! has joined #mythtv
[21:02:17] warpme_ (warpme_! has quit (Quit: warpme_)

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