• 1
  • 103
  • 104
  • 105(current)
  • 106
  • 107
  • 168
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)
(2015-10-10, 18:29)Milhouse Wrote:
(2015-10-10, 11:58)Basje Wrote: It seems there is some inconsistency in the Python API between these builds and the official Kodi Nightly code? For instance: the xbmc.PLAYER_CORE_DVDPLAYER is no longer available here, but is in other nightly builds (Windows for instance).

I understand that they are no longer of any use, but removing them like this really breaks Python scripts (like Retrospect which I develop). Would it be an option to log the usage of these deprecated Python code as Warnings in the Kodi log and only remove them in a next version? This was done with other Python API changes.

Because in the current situations it is rather difficult to have an add-on that is compatible with the Jarvis builds and these builds.

These builds are based on VideoPlayer, which is the long term replacement for DVDPlayer.

You could try PLAYER_CORE_VideoPlayer in place of PLAYER_CORE_DVDPLAYER.

All I can suggest is that you feedback to the addon developers and perhaps they can add a solution that works with both DVDPlayer and VideoPlayer-based builds, as this is something they're going to have to deal with eventually.
Well, I am the developer Big Grin, but may I suggest to not break compatibility with less experimental builds and just ignore whatever PLAYER_CORE is specified, instead of removing it from the Python API.

From an Add-on developers point of view, I already have a hard time keeping compatibility between the current Kodi, the one before and XBMC4Xbox, so I don't want to make stuff more complex and support different kind of experimental builds.
The Python API is generated at compile time from the available Kodi players, so replacing DVDPlayer with VideoPlayer in Kodi automatically has the same effect on the Python API.

I don't think there's a better solution other than for addons to accept that DVDPlayer will at some point be replaced by VideoPlayer, and to code accordingly. You're welcome to post on FernetMenta's github which is where these changes are all originating from, and ask if he has any better ideas.
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.
(2015-10-10, 19:58)Milhouse Wrote: The Python API is generated at compile time from the available Kodi players, so replacing DVDPlayer with VideoPlayer in Kodi automatically has the same effect on the Python API.

I don't think there's a better solution other than for addons to accept that DVDPlayer will at some point be replaced by VideoPlayer, and to code accordingly. You're welcome to post on FernetMenta's github which is where these changes are all originating from, and ask if he has any better ideas.

Fair enough. I will try to catch this in my add-on code. Thanks for the quick reply.
ping popcornmix:

1009 fails for me in rpi2. 1008 did not crash, 1007 seemed most stable of the three.

1009 crashed on startup the first start:
kodi.old.log http://sprunge.us/IDCN
kodi_crashlog_20151010112259.log http://sprunge.us/UXMS

Also, after that it craches whenever I click "Tv Shows"
kodi.old.log http://sprunge.us/NMZh
kodi_crashlog_20151010113503.log http://sprunge.us/hScV

Also, I thought I didnt have any overclock, because "OpenELEC RPi Config" says my overclock is disabled. In that kodi.old.log I see something about the addon not being symlinked.

Here is my config.txt http://sprunge.us/YJKX

Clearly overclock is not disabled:
arm_freq=1000
core_freq=500
sdram_freq=500
- over_voltage=6

EDIT:
Disabled all overclocks and omx acceleration, still same crash.
kodi.old.log http://sprunge.us/FjYV
kodi_crashlog_20151010115339.log http://sprunge.us/GPcV

dmesg shows nothing about the crash (just a kodi crash, not rebooting, I dont even lose ssh): http://sprunge.us/aDRO

EDIT #2:
Same problem with #1010
New OpenELEC Jarvis build #1010: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.10 #1 Sat Oct 10 21:04:30 BST 2015 armv6l GNU/Linux

# vcgencmd version
Oct 10 2015 00:23:33
Copyright (c) 2012 Broadcom
version 8231aaab61ad018b24c2f16f9072048d507463bf (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20151010210341-#1010-g33cf42b [Build #1010]

# vcdbg log msg 2>&1 | grep DTOK
001737.921: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (33cf42bd, changelog) and tip of XBMC master (350a8ea1, changelog) with the following modifications: Build Highlights:
  1. New firmware
  2. Disable libaacs as this seems to be responsible for BD ISO crashes
Build Details:
  1. Firmware (Oct 10):
    • firmware: arm_loader: Support DT commands in included files. See: link
    • firmware: di_adv: Need up align up height to at least 4. See: link
    • firmware: Allow loading of tuning data from a memory block
  2. OpenELEC:
    • systemd-addon-wrapper: make binary addons executable ... (PR:4374, 1 commit, 1 file changed)
  3. XBMC:
    • [xbmc]keyboard]Added init/uninit for coding tables to avoid having the baidu thread … (PR:8176, 1 commit, 10 files changed)
    • [rbp] tools: only use sudo if we have it on the system (PR:6910, 1 commit, 1 file changed)
    • Lets skins use a light font weight (PR:8182, 1 commit, 5 files changed)
    • fixed: setlocale issue other than windows (PR:7820, 1 commit, 1 file changed)
    • [gitignore] rename xbmc -> kodi (PR:8193, 1 commit, 1 file changed)
  4. newclock5:
    • New commits in this build:
      • VideoPlayer: add debug log for flushing audio (11ff8435)
      • VideoPlayer: do not flush and drain audio on stall (f096d601)
      • yuv2rgb: don't error on non neon devices - squash me (f4d741c1)
      • AE: avoid skip or delay of odd number of bytes (45b400cd)
      • AE: fix calculation of bytesToSkip to avoid cracking sounds (8ba0f2e3)
      • [renderer] Remove RenderUpdateCallBack and RenderFeaturesCallBack (d93f05a3)
      • [AE] Fix large buffertime check (8617d963)
      • [videoplayer] Allow passthrough with display clock and pll adjust (06fe567f)
    • Commits no longer in build:
      • [videoplayer] Hacky scheme to enable accelerated playback as first ff step (0d7dbf5c)
  5. Additional commits/pull requests/changes not yet merged upstream:
    • Added: [env] patch: Disable libaacs which may be crashing BD ISOs
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.
As mentioned in this post : http://forum.kodi.tv/showthread.php?tid=...pid2129583 , I am having trouble playing full SBS (3840x1080) and full TAB (1920x2160) 3D files on almost all builds. As can be seen from the picture, video is highly pixelated. These files play properly on the Nexus Player, nVIDIA SATV and other Android boxes.

http://imgur.com/svWwPsF.jpg

Debug log: http://xbmclogs.com/pvon15zoq
It was taken after playing the full SBS and full TAB versions of the same movie.
With #1010 most of the 3D isos still reboots Rpi2. Starting ssif makes no difference.
(2015-10-11, 11:03)Caramba69 Wrote: With #1010 most of the 3D isos still reboots Rpi2. Starting ssif makes no difference.

Okay I'll need a sample. I got a sample from querty1000 which used to hang on opening but now plays with latest update, but I suspect there is another failure mode.
Note: The cracking sound on startup or seeks should be fixed with latest version. If you were suffering from that please test and report.
(2015-10-11, 06:49)wesk05 Wrote: As mentioned in this post : http://forum.kodi.tv/showthread.php?tid=...pid2129583 , I am having trouble playing full SBS (3840x1080) and full TAB (1920x2160) 3D files on almost all builds. As can be seen from the picture, video is highly pixelated. These files play properly on the Nexus Player, nVIDIA SATV and other Android boxes.

http://imgur.com/svWwPsF.jpg

Debug log: http://xbmclogs.com/pvon15zoq
It was taken after playing the full SBS and full TAB versions of the same movie.

The Pi only supports resolutions up to 1080p. It also supports MVC, which is two streams of 1080p.
However it won't play 3840x1080 or 1920x2160 videos. The hardware does not support that.
(2015-10-11, 06:49)wesk05 Wrote: As mentioned in this post : http://forum.kodi.tv/showthread.php?tid=...pid2129583 , I am having trouble playing full SBS (3840x1080) and full TAB (1920x2160) 3D files on almost all builds. As can be seen from the picture, video is highly pixelated. These files play properly on the Nexus Player, nVIDIA SATV and other Android boxes.

http://imgur.com/svWwPsF.jpg

Debug log: http://xbmclogs.com/pvon15zoq
It was taken after playing the full SBS and full TAB versions of the same movie.

Off topic - but out of interest - how are these movies output by the boxes that DO play them correctly? (nVidia Shield, Nexus Player)

Do they all get output as Frame Packed 1920x1080/24p (which is effectively 1920x2205/24p with 2 x 1920x1080 streams one above the other with 45 lines of blanking between the top and bottom formats) and retain their Full HD 3D resolution or are they converted to 1920x1080/24p HSBS (i.e. 960x540/24p for each eye with no blanking between each 2:1 squeezed eye feed) for output, effectively halving their resolution?

Doesn't the nVidia Shield not cope with 24p properly? So are the 3840x1080/24p movies replayed at 1920x1080/60p with 3:2 judder and halved horizontal resolution? I don't think the nVidia supports a Frame Packed Full HD 3D output does it? (The Pi and Pi 2 are unusual in doing so, outside of Intel Core i-series CPUs and dedicated nVidia and AMD GPUs?)

I don't think there is a standard HDMI output format for 3840x1080 or 1920x2160 - though there may be a 1920x1080/48p variant (alternate frame) which is subtly different from the 1920x2205/24p frame packed standard that is used for mainstream Blu-ray 3D output of 24p content. (ISTR that 3D Blu-ray also allows for 1280x720/60p 3D content to be output as 1280x1470/60p - which is a frame packed format with 30 lines of blanking between the two 1280x720 eye feeds within the frame packed frame)
(2015-10-11, 11:36)popcornmix Wrote:
(2015-10-11, 11:03)Caramba69 Wrote: With #1010 most of the 3D isos still reboots Rpi2. Starting ssif makes no difference.

Okay I'll need a sample. I got a sample from querty1000 which used to hang on opening but now plays with latest update, but I suspect there is another failure mode.

Ah! - I have caught a segfault with Coraline iso in my debug build. That gives me something to attack.

EDIT: Okay I have a fix for Coraline iso. Should be in tonight's build.
(2015-10-11, 14:20)popcornmix Wrote:
(2015-10-11, 11:36)popcornmix Wrote:
(2015-10-11, 11:03)Caramba69 Wrote: With #1010 most of the 3D isos still reboots Rpi2. Starting ssif makes no difference.

Okay I'll need a sample. I got a sample from querty1000 which used to hang on opening but now plays with latest update, but I suspect there is another failure mode.

Ah! - I have caught a segfault with Coraline iso in my debug build. That gives me something to attack.

EDIT: Okay I have a fix for Coraline iso. Should be in tonight's build.

Great. I will test it.
(2015-10-11, 12:53)popcornmix Wrote: Note: The cracking sound on startup or seeks should be fixed with latest version. If you were suffering from that please test and report.
Cracking with start & seek when using mmal acceleration is indeed fixed in latest build
(2015-10-11, 14:07)noggin Wrote: Off topic - but out of interest - how are these movies output by the boxes that DO play them correctly? (nVidia Shield, Nexus Player)

Do they all get output as Frame Packed 1920x1080/24p (which is effectively 1920x2205/24p with 2 x 1920x1080 streams one above the other with 45 lines of blanking between the top and bottom formats) and retain their Full HD 3D resolution or are they converted to 1920x1080/24p HSBS (i.e. 960x540/24p for each eye with no blanking between each 2:1 squeezed eye feed) for output, effectively halving their resolution?

Doesn't the nVidia Shield not cope with 24p properly? So are the 3840x1080/24p movies replayed at 1920x1080/60p with 3:2 judder and halved horizontal resolution? I don't think the nVidia supports a Frame Packed Full HD 3D output does it? (The Pi and Pi 2 are unusual in doing so, outside of Intel Core i-series CPUs and dedicated nVidia and AMD GPUs?)

The other players output these files as HSBS or HTAB. This just means that there is actually no benefit to ripping the movies to full SBS/TAB. It looks like full SBS/TAB files are supported only on PC (Cyberlink PowerDVD, Stereoscopic Player etc., output is framepacked 3D).

For whatever reasons, RPi2 is unable to convert these files to HSBS/HTAB. It could be because of the hardware MVC decoding support, but I would have expected full TAB to work.

nVIDIA SATV does have proper 23.976 support on OTA 2.0 firmware.
  • 1
  • 103
  • 104
  • 105(current)
  • 106
  • 107
  • 168

Logout Mark Read Team Forum Stats Members Help
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)10