• 1
  • 58
  • 59
  • 60(current)
  • 61
  • 62
  • 146
OpenELEC Testbuilds for RaspberryPi (Kodi 17.0)
Hi,
have wrote somethings about AC3 transcoding in this thread.
I'm not able with recent build to play AAC 5.1 in DolbyDigital.

well, next thing - Groundhog Day Rofl in this build i select "play in DTS" -> no, plays in DD5.1 Angel
(2016-01-30, 21:41)popcornmix Wrote:
(2016-01-30, 21:40)Quinn2015 Wrote: Ok,i bought a license key and everything is fine now.
Do you still want to know the last working build,without the license key?
Maybe out of interest?

Yes. I'm not aware of any recent change that should have made things worse, so it would be good to understand why.


I have pinned down the last working build. It´s #105b.
I could watch the dvd without the MPEG-2 license key.
From #106, software decode is messed up.

Log from working #105b

http://xbmclogs.com/pqru5ncd7

i hope it helps.
(2016-01-31, 14:53)Quinn2015 Wrote: I have pinned down the last working build. It´s #105b.
I could watch the dvd without the MPEG-2 license key.
From #106, software decode is messed up.

Thanks.
Two plausible commits:
drop obsolete libmpeg2 (PR:8687, 3 commits, 138 files changed)
VideoPlayer: fix and cleanup ffmpeg sw deinterlacing (PR:8773, 1 commit, 4 files changed)

Now I think the first should be harmless. libmpeg2 was only used for menus, not the main feature, and that had been disabled in a previous PR. This PR was just removing dead code.
So, it's possibly deinterlace related. You can try disabling deinterlace and see if it helps (although deinterlace is quite expensive anyway so it probably will help).

When I originally supported deinterlace with software decode I found Yadif 2x was too hard, and had to limit it to Yadif 1x (i.e. 30fps output rather than 60fps).
Possibly that commit has changes the default deinterlace to Yadif 2x.
I had a problem with a disappeared file on build #0127 and I can't pin down why that happened.
I watched a movie yesterday and stopped after about 30 minutes and switched everything off. Today, when I wanted to continue watching the movie, the file was gone.

Setup:
RPi2 with OpenElec Build #0127
External 3.5' HDD, Seagate 3GB formatted under OSX with exFAT, connected with a MicroUSB 2.0 cable
AVC MKV File, about 33 GB

I'm powering the RPi2 from the USB port of my TV since the external HDD has its own power supply. The whole setup is run on a switchable power strip, so I tend just so switch everything off in one go without powering stuff down. That has never been a problem before.

I might have accessed the HDD from my MacBook with Pathfinder while the movie was running, I'm not sure. But I definitely didn't delete any files. Weirdly, while connected to the RPi2 and accessing it from Pathfinder, the 33GB of space from this movie was still occupied. So I connected the HDD directly to my MacBook, but then the space was free.
I know that KODI doesn't delete files, so might this be a bug or did I do something wrong?
(2016-01-31, 16:21)cecay Wrote: I know that KODI doesn't delete files, so might this be a bug or did I do something wrong?

Yes.

Quote:I'm powering the RPi2 from the USB port of my TV since the external HDD has its own power supply. The whole setup is run on a switchable power strip, so I tend just so switch everything off in one go without powering stuff down. That has never been a problem before.

Do you just switch off a windows PC from the mains without shutting down?
If so you are playing russian roulette. Yes, you often get away with it, but if a file is being written (perhaps just to update file access times)
when you switch off there is no knowing what the state will be when you power on again. Corruption is always a possibility.

You must shut down from the kodi menu before powering off. Or power the Pi (and hard disk) from a separate plug that is always left on.
Hi guys,

i've been runing Milhouse-builds for quite some time now.

Currently i am running in trouble with Live TV. Especially stutter/lagging of Live TV playback, Playing Viedos from the same source works without any problems.
I am quite sure, that the problem is somewhere on the Pi. when connecting a PC instead of the Pi playback is fine, also whatching the same channel on the Pi and a laptop. The Pi is lagging, where the stream on the laptop is working fine.

My setup is based on a dedicated TV/Data-server and a Pi 2 as mediaplyer/client.
TV stream is served by a TVHeadend or VDR --> hts or vnsi on the Pi. (both setups with stuttering/lagging (VDR was installed to check if hts or TVHeadend is the problem)).

I have tried to record logfiles with both systems.
Build #0130b

Logfile 1 - Kodi_vnsi.log:
Errors @:
- 15:00:19 511
- 15:06:03 855

Logfile 2 - Kodi_hts.log:
Errors @:
- 15:14:56 149
- 15:24:15 708

I have put the logfile into Dropbox:
https://www.dropbox.com/s/nc6z87exsduapz...g.zip?dl=0

Probably the problem is somehow related to this:
http://forum.kodi.tv/showthread.php?tid=255816&page=2
But for me the problems occure very fast after switching to a channel.

I am still trying to find a build where playback was good, im am back to december 2015, but still no luck.

If there are more details needed please let me know.

Probably there is an easy fix, im not sure if I have messed up any settings (video playback, audiosync ...).

Thank you in advance.
(2016-01-31, 17:16)DerMarshal Wrote: I have put the logfile into Dropbox:
https://www.dropbox.com/s/nc6z87exsduapz...g.zip?dl=0

You are using Amber which isn't supported. Switch to Confluence until Amber updates to support Krypton.
You have set the audio latency to 150ms. It suggest you return it to default (50ms).
Switch "Resample quality" back to high (you are on gpu).
I'd be interested if enabling omxplayer in "video/acceleration" settings helps
I'd be interested if disabling deinterlacing helps
Do recording play well?

Assuming nothing there helps, then finding a working build would be useful.
(2016-01-25, 23:47)popcornmix Wrote: Thanks for sample. I get the same behaviour as you. I've done a little digging and it seems libaacs is only used when reading files through libbluray.
We don't currently use libbluray for 3D ISOs which is why it doesn't work.
There's no quick fix for this, but this provides another argument for integrating the ssif parsing inside libbluray.

Do you think it would be possible to simply put libaacs somewhere in between? Without libbluray?
I use mythtv backend and DVB-T2/T/S2 tuners with several RPis as frontends using pvr.mythtv.
I have been having a problem with some channels here in the UK which change audio stream between commercial breaks and programmes for a while now.

When using OMX acceleration on my RPi (B+), the audio stops approx 4 seconds before changing to the new audio type while video playback continues. With sync playback to display turned off, the same happens.
(Deinterlace is turned off due to an aspect ratio switching issue which @popcornmix said would take serious effort to fix, although I hope it's still in the pipeline somewhen during the videoplayer restructuring...)

The 4 second audio issue doesn't seem to happen when using maml acceleration (at least with #0130b and #1109, the newest and oldest builds I have lying around)
I am working around this on my main TV by turning on passthrough, but that doesn't work on my other frontends :-(

Anyone here with the knowhow fancy trying to fix OMX accelerated playback for this?
Happy to stress test/try alternative settings etc... if it helps.

See http://www.thomasinfletcher.com/testfiles/test1.mpg for a short clip which exhibits the issue.
(2016-01-31, 10:43)niwa2 Wrote: I can confirm the total duration bug.
I found a 3DMVC iso file in my collection that shows the behavior. This seems to be the only file i have that shows this behavior.
The total time increases with the play time plus about 7-10s for buffering.

Debug Log #0130b

I do not know when this bug whas introduced. It is definately present in #0129 and #0130b.

Also the delay after refresh rate change is still not working for me.

This bug was introduced in build #0123.
Everything is perfect in build #0122 for me, could you confirm this on your side ?
I noticed that on this latest build #0130b, one of my 3D BD ISO that was not working is working perfectly fine now (The Martian).
But another 3D BD ISO is still buggy on this last release (Everest).
Here is the debug log that shows the total duration bug (Everest 3D) :
http://xbmclogs.com/p3rksugnh

And the debug log where the bug is not present (the Martian 3D) :
http://xbmclogs.com/ptygzlfcd
New OpenELEC Krypton build #0131: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.4.0 #1 Sun Jan 31 22:23:10 GMT 2016 armv6l GNU/Linux

# vcgencmd version
Jan 29 2016 13:52:31
Copyright (c) 2012 Broadcom
version 75a32fd37768d64654506c3eaba453ecc3d587a6 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20160131222215-#0131-g0fe06b7 [Build #0131]

# vcdbg log msg 2>&1 | grep DTOK
001743.901: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (0fe06b74, changelog) and tip of XBMC master (ec3ee1bb, changelog) with the following modifications: Build Highlights:
  1. Remove boost from GUIFontCache
  2. Improve handling of refresh rate changes
Build Details:
  1. OpenELEC:
    • mkpkg_binary-addons: don't checkout and create a new branch at the sa… (PR:4657, 1 commit, 1 file changed)
    • xf86-video-nvidia: update to xf86-video-nvidia-352.79 (PR:4660, 3 commits, 2 files changed)
    • connman: fix lease expiry issues (PR:4663, 1 commit, 1 file changed)
  2. XBMC:
    • [Confluence] increase textbox height in DialogConfirm.xml (PR:8994, 1 commit, 1 file changed)
    • [jenkins/rbpi] - use new toolchain for rbpi and get rid of the unused… (PR:9006, 1 commit, 1 file changed)
    • [win32] remove BOOST_PP_STRINGIZE and replace with simple stringification (PR:8990, 1 commit, 1 file changed)
    • [Confluence] cleanup non existent skin setting (PR:9004, 1 commit, 1 file changed)
    • [lang] move addon language file handing to CLocalizeStrings (PR:8819, 6 commits, 21 files changed)
  3. kodi-platform:
    • [debian] bump package version to v17 (PR:19, 1 commit, 4 files changed)
  4. pvr.argustv:
    • Add travis.yml and gitignore (PR:42, 4 commits, 4 files changed)
  5. pvr.filmon:
    • Add travis.yml and gitignore (PR:38, 4 commits, 4 files changed)
  6. pvr.hdhomerun:
    • Add travis.yml and gitignore (PR:9, 3 commits, 3 files changed)
  7. pvr.hts:
    • HTSPDemuxer: fix m_seekTime race in CHTSPDemuxer::Seek (PR:190, 1 commit, 1 file changed)
    • Remove superfluous "attempting reconnect" log message (PR:191, 1 commit, 1 file changed)
    • Add note about branches to the README (951f0c27)
  8. pvr.pctv:
    • Add travis.yml and gitignore (PR:30, 4 commits, 4 files changed)
  9. pvr.stalker:
    • Add travis.yml and gitignore (PR:43, 4 commits, 4 files changed)
  10. pvr.vbox:
    • Add support for all product families ("VB", "VJ", "VZ" etc) (PR:112, 1 commit, 1 file changed)
  11. newclock5:
    • New commits in this build:
      • Revert "ActiveAE: Enable display lost callbacks for EGL" (19e22cfa)
      • VideoPlayer: Avoid deadlock with m_StateSelection (c48d21e5)
      • VideoPlayer: Handle lost display with pause/resume (ede0de7f)
      • Revert "VideoPlayer: Handle lost display with pause/resume" (0f75e179)
      • AE: fix number of provided buffers to input stream, 86301d8f5abd081792520d69752a519e1cc666e0 already fixed GetCacheTime and added resample buffers a second time (e9f17179)
    • Updated commits in this build:
    • Commits no longer in build:
      • hack: Avoid videoplayer deadlock (8aa778c3)
  12. Additional commits/pull requests/changes not yet merged upstream:
    • Added: [pkg] PR:8996: remove boost from GUIFontCache
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
(2016-01-25, 23:54)rewsteruk Wrote: Here is the log with confluence http://sprunge.us/HHff
This did not wake the server, on first episode selected, wol then gave remove from server prompt. second episode froze the debug info on screen no wol prompt then pop up with remove from server prompt

Hi rewsteruk, presumably you still have WOL issues with the most recent builds. There's going to be a potential WOL fix in tonight's #0201 build so would be good if you can test that. Thanks.
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
(2016-01-31, 14:53)Quinn2015 Wrote:
(2016-01-30, 21:41)popcornmix Wrote:
(2016-01-30, 21:40)Quinn2015 Wrote: Ok,i bought a license key and everything is fine now.
Do you still want to know the last working build,without the license key?
Maybe out of interest?

Yes. I'm not aware of any recent change that should have made things worse, so it would be good to understand why.


I have pinned down the last working build. It´s #105b.
I could watch the dvd without the MPEG-2 license key.
From #106, software decode is messed up.
...
Sure, i've downgrade from #0131 to #105b for sometime watch the ISO file without the license key.
  • 1
  • 58
  • 59
  • 60(current)
  • 61
  • 62
  • 146

Logout Mark Read Team Forum Stats Members Help
OpenELEC Testbuilds for RaspberryPi (Kodi 17.0)6