[debutanker]

@nexusle
---> https://forum.kodi.tv/showthread.php?tid...pid2652851

[Milhouse]

I'd updated today from 0922 to the latest build. After this, I haven't any labeling anymore in Kodi UI

Please read at least a few posts since the last build before posting because, you know, someone might already have mentioned that.

Edit: I even mention the label issue is in the summarised changelog!

[pyrodex]

@pyrodex

Run:

Code:

texturecache.py Jd tvshows halt @extrajson.tvshows.episode=playcount | pastebinit

and paste the link.

I'm fairly sure that Kodi is treating Season 4 Episode 6 as watched, because the playcount is not zero.

I can reproduce what you are seeing if I set the playcount of a partially watched episode to 1:

Code:

texturecache.py set episode 20334 playcount 1

where 20334 is the episode id of the partially watched episode (the episode id for S04E6 in your case).

If you set the playcount to 0 for episode S04E06 then you should see it now listed in the Unwatched filter. Or you can use the episode context menu and "Mark as unwatched" which should achieve the same effect.

This Wiki page explains when Kodi marks a video as watched.

Did you mention you only see this with these builds and not a Mac (or something)? If so I have no idea why that would be, unless the other devices are not sharing the same database. You can query the playcount on those other machines to confirm their playcount is zero as the filter is making decisions based only on the metadata in the video library.

Sorry finished E06 so did this with E07.


texturecache dump - http://sprunge.us/UQFi

I also reset it to playcount 0 and can confirm it comes back in the unwatched filter but the resume marker stays and the timecode in the DB to resume from.

My MAC machine is using the same DB but just checked on it and a Windows box all running the latest nightlies and they exhibit the same problem.

I also read the Wiki article and don't have any special watched settings. I was aware of the default values and have experienced Kodi marking an episode watched in the past when I was like 93% and had to stop for some reason, this is why I put a soft toggle unwatched/watched button on my Harmony remote for this case.

[Milhouse]

I also reset it to playcount 0 and can confirm it comes back in the unwatched filter but the resume marker stays and the timecode in the DB to resume from.

Playcount (which determines if something is considered watched) is unrelated to resume point.

You can clear the resume point using "Reset resume position" on the context menu, which has no effect on the current playcount. A partially watched episode can be marked as unwatched - resetting the playcount to zero - but this will not remove the current resume point.

My MAC machine is using the same DB but just checked on it and a Windows box all running the latest nightlies and they exhibit the same problem.

That's to be expected, as there's nothing specific to these builds that is likely to be contributing to your problem (I think I said this a while ago!) however I'm not entirely clear on what the problem is, unless you are saying the playcount is being bumped without reaching 90%? Maybe it's an addon that is doing it?

I also read the Wiki article and don't have any special watched settings. I was aware of the default values and have experienced Kodi marking an episode watched in the past when I was like 93% and had to stop for some reason, this is why I put a soft toggle unwatched/watched button on my Harmony remote for this case.

I believe the default is 90%, so that behaviour would sound like it is working as intended. At what percentage do you experience the playcount being bumped?

[Milhouse]

@pyrodex btw, if you want to view your resume point metadata:

Code:

texturecache.py Jd tvshows halt @extrajson.tvshows.episode=playcount,resume

[Milhouse]

@pyrodex: Also...

If you run

Code:

texturecache.py monitor

then when playback ends, if you reached the critical "watched" percentage, then you will see the new playcount reported:

Code:

2017-10-07 20:51:59.916479: Player.OnStop        : {"item": {"type": "episode", "id": 20334}, "end": true} [The Grand Tour (2016) S01E04 (Enviro-mental)]
2017-10-07 20:52:00.295036: VideoLibrary.OnUpdate: {"item": {"type": "episode", "id": 20334}, "playcount": 1} [The Grand Tour (2016) S01E04 (Enviro-mental)]
2017-10-07 20:52:00.326056: VideoLibrary.OnUpdate: {"type": "episode", "id": 20334} [The Grand Tour (2016) S01E04 (Enviro-mental)]

If an episode (or movie) is stopped before it has reached the critical "watched" percentage then the playcount will not be updated - the following is output when stopping playback at about 50%:

Code:

2017-10-07 20:53:21.257224: Player.OnStop        : {"item": {"type": "episode", "id": 20334}, "end": false} [The Grand Tour (2016) S01E04 (Enviro-mental)]
2017-10-07 20:53:21.680547: VideoLibrary.OnUpdate: {"item": {"type": "episode", "id": 20334}} [The Grand Tour (2016) S01E04 (Enviro-mental)]
2017-10-07 20:53:21.738589: VideoLibrary.OnUpdate: {"type": "episode", "id": 20334} [The Grand Tour (2016) S01E04 (Enviro-mental)]

[pyrodex]

I also reset it to playcount 0 and can confirm it comes back in the unwatched filter but the resume marker stays and the timecode in the DB to resume from.

Playcount (which determines if something is considered watched) is unrelated to resume point.

You can clear the resume point using "Reset resume position" on the context menu, which has no effect on the current playcount. A partially watched episode can be marked as unwatched - resetting the playcount to zero - but this will not remove the current resume point.

My MAC machine is using the same DB but just checked on it and a Windows box all running the latest nightlies and they exhibit the same problem.

That's to be expected, as there's nothing specific to these builds that is likely to be contributing to your problem (I think I said this a while ago!) however I'm not entirely clear on what the problem is, unless you are saying the playcount is being bumped without reaching 90%? Maybe it's an addon that is doing it?

I also read the Wiki article and don't have any special watched settings. I was aware of the default values and have experienced Kodi marking an episode watched in the past when I was like 93% and had to stop for some reason, this is why I put a soft toggle unwatched/watched button on my Harmony remote for this case.

I believe the default is 90%, so that behaviour would sound like it is working as intended. At what percentage do you experience the playcount being bumped?

So on a machine with #1003 (Rolled back due to labels for this test) I created a new kodi directory, added my sources, and my advanced settings (MySQL configuration) and didn't see the issue. I am seeing playcount bumped prior to 90% and with a FRESH kodi directory this doesn't occur. 

Here are the addons I had added but not sure why any of these would be doing it...

Code:

game.controller.default
metadata.album.universal
metadata.artists.universal
metadata.common.fanart.tv
metadata.common.imdb.com
metadata.common.musicbrainz.org
metadata.common.theaudiodb.com
metadata.common.themoviedb.org
metadata.themoviedb.org
metadata.tvdb.com
packages
plugin.video.discovery-channels
plugin.video.diy
plugin.video.youtube
pvr.mythtv
script.module.beautifulsoup4
script.module.html5lib
script.module.mechanize
script.module.requests
script.module.simplejson
script.module.six
script.module.t1mlib
weather.yahoo

[Milhouse]

New LibreELEC.tv Leia build #1007: RPi / RPi2
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 4.13.5 #1 Sat Oct 7 21:04:25 BST 2017 armv6l GNU/Linux

# vcgencmd version
Oct  6 2017 13:59:43
Copyright (c) 2012 Broadcom
version bde38f02b236b6e1bd16eb0f9b44b35f5a85cb30 (clean) (release)

# lsb_release
LibreELEC (Milhouse): devel-20171007210313-#1007-ge21ac23 [Build #1007]

# Kodi version
(18.0-ALPHA1 Git:bedc63f). Platform: Linux ARM 32-bit

Based on tip of LibreELEC.tv master (e21ac23, changelog) and tip of XBMC master (bedc63f, changelog) with the following modifications:

• Includes newclock5 patches
  
• Excludes the LibreELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-4.13.y
  
• Includes latest bcm2835-driver master (2669578, ahead +1)
  
• Includes latest kodi-platform master (36fb493)
  
• Includes latest libcec master (8adc786, ahead +26)
  
• Includes latest libnfs master (2911afe, ahead +103)
  
• Includes latest p8-platform master (a822e19)
  
• Includes latest addons: inputstream.adaptive (dde3921), inputstream.rtmp (26260c9), peripheral.joystick (cc14287), pvr.argustv (23cc0e8), pvr.demo (122dedd), pvr.dvblink (816f90c), pvr.dvbviewer (9765663, +1), pvr.filmon (7a9af25, +1), pvr.hdhomerun (ddfe2cb, +1), pvr.hts (0530971, +15), pvr.iptvsimple (ac4f4bd), pvr.mediaportal.tvserver (fa1069d, +1), pvr.mythtv (8965048), pvr.nextpvr (5b7caa2, +1), pvr.njoy (cc1cb56), pvr.octonet (e9b4c05), pvr.pctv (549f690), pvr.stalker (e15edca), pvr.teleboy (1cf8e96, +2), pvr.vbox (56052da, +1), pvr.vdr.vnsi (f3f80d5), pvr.vuplus (c53478d, +1), pvr.wmc (02edba6, +1), pvr.zattoo (73009cd)
  
• Include [env] compare: dmidecode: initial package
  
• Include [env] patch: RPi/RPi2: enable Broadcom WiFi debugging (see details)
  
• Include [env] patch: libcec: don't link non-existant libtinfo
  
• Include [env] patch: remove PKG_ADDON_VERSION in order to apply PR:2072
  
• Include [env] patch: rev hack for kodi
  
• Include [env] patch: tvheadend42: fix makefile patch
  
• Include [env] patch: media_build: disable due to conflicts
  
• Include [env] patch: Add experimental splash video for RPi
  
• Include [env] patch: Bump included addon versions to prevent online updates
  
• Include [env] patch: HACK: Disable multiple PVR addons during migration. Always enable inputstream.* and os.*
  
• Include [env] patch: Add kodi binary addons (pvr, adsp, inputstream, vfs, other)
  
• Include [env] patch: pvr addons: disable broken addons
  
• Include [env] PR:1776: Don't cause system to hang on halt/reboot when net-booting
  
• Include [env] PR:1791: nss/nspr: include in main image [rfc]
  
• Include [env] PR:1835: add support for optional kernel modules in image and from addons
  
• Include [env] PR:1978: linux: update to linux-4.13.4
  
• Include [env] PR:2002: linux: include HBR and ELD fixes for Intel
  
• Include [env] PR:2048: mesa: update to 17.2.2
  
• Include [env] PR:2061: added EFI 32bit support via GRUB2
  
• Include [env] PR:2072: buildsystem: unset all PKG_* variables before sourcing a new package
  
• Include [env] PR:2075: xorg-server: update to xorg-server-1.19.4
  
• Include [pkg] patch: skin.estuary: reduce system info font size (kodi)
  
• Include [pkg] PR:234: implement intensity compensation for VC-1 decoding for Ivybridge and better (intel-vaapi-driver)
  
• Include [pkg] PR:260: Fix Seeking with VC1 after 6702b01ee3598017f05374bfc7956493f790361c (intel-vaapi-driver)
  

Build Highlights:

1. Fix missing labels
   

Build Details:

1. LibreELEC.tv:
   • Fix perl/JSON.pm dependency on OpenSUSE and CentOS (PR:2032, 1 commit, 1 file changed)
     
2. XBMC:
   • Controller dialog: Resize layouts (PR:12884, 2 commits, 3 files changed)
     
   • GLES: fix accidentally changed code for fonts (PR:12890, 1 commit, 1 file changed)
     
   • AESinkAudioTrack: Verify ENCODING_IEC61937 before using it (PR:12862, 3 commits, 2 files changed)
     
3. pvr.hts:
   • Fix windows build (missing ssize_t type definition) after #334 (PR:335, 1 commit, 3 files changed)
     
   • Fix windows build (missing ssize_t type definition), take 2. (PR:336, 1 commit, 2 files changed)
     
   • Fix windows build (missing channel_type_t definition). (PR:337, 1 commit, 1 file changed)
     
4. Additional commits/pull requests/changes not yet merged upstream:
   • Added: [env] patch: remove PKG_ADDON_VERSION in order to apply PR:2072
     
   • Added: [env] PR:2072: buildsystem: unset all PKG_* variables before sourcing a new package
     
   • Added: [env] PR:2075: xorg-server: update to xorg-server-1.19.4
     

[Milhouse]

Here are the addons I had added but not sure why any of these would be doing it...

I have no idea, either.

Are you seeing playcount bumped when viewing is stopped? Or is it bumped at some other time? You could monitor the MySQL query log to see when the playcount is updated, or repeatedly query the library using texturecache.py Jd (monitor will only capture the new playcount if it is updated using the standard Kodi mechanism so it may not report the correct value if it is being updated "behind the scenes" by an addon). Maybe grep those addons to see if any mention playcount in their source code.

[pyrodex]

Here are the addons I had added but not sure why any of these would be doing it...

I have no idea, either.

Are you seeing playcount bumped when viewing is stopped? Or is it bumped at some other time? You could monitor the MySQL query log to see when the playcount is updated, or repeatedly query the library using texturecache.py Jd (monitor will only capture the new playcount if it is updated using the standard Kodi mechanism so it may not report the correct value if it is being updated "behind the scenes" by an addon). Maybe grep those addons to see if any mention playcount in their source code.

I am at a loss of words here.

Running #1007 clean kodi directory minus my source.xml, advancedsettings.xml (MySQL), and remote keymap.xml with ONLY the yahoo.weather addon setup and this happens.

Code:

2017-10-07 19:27:58.170244: Playlist.OnAdd       : {"item": {"type": "episode", "id": 1362}, "playlistid": 1, "position": 0} [The Deuce S01E04 (I See Money)]
2017-10-07 19:27:58.297272: Player.OnPlay        : {"item": {"type": "episode", "id": 1362}, "player": {"playerid": 1, "speed": 1}} [The Deuce S01E04 (I See Money)]
2017-10-07 19:28:00.217827: Player.OnSeek        : {"item": {"type": "episode", "id": 1362}, "player": {"playerid": 1, "seekoffset": {"hours": 0, "seconds": 0, "minutes": 10, "milliseconds": 0}, "speed": 1, "time": {"hours": 0, "seconds": 1, "minutes": 10, "milliseconds": 343}}} [The Deuce S01E04 (I See Money)]
2017-10-07 19:28:05.486830: Player.OnStop        : {"item": {"type": "episode", "id": 1362}, "end": false} [The Deuce S01E04 (I See Money)]
2017-10-07 19:28:05.799131: VideoLibrary.OnUpdate: {"item": {"type": "episode", "id": 1362}, "playcount": 1} [The Deuce S01E04 (I See Money)]
2017-10-07 19:28:05.827542: VideoLibrary.OnUpdate: {"type": "episode", "id": 1362} [The Deuce S01E04 (I See Money)]

I started S01E04 FRESH, never watched nothing, and then skipped ahead 10 minutes to trigger a resume point. I stopped the playback and you can see it clearly set playcount to 1.

Debug log - http://sprunge.us/ZMCE
Texture info - http://sprunge.us/LQfD

Short of me reimagine the entire SD card I started with a fresh Kodi running today's episode with minimal addons.

UPDATE: I was curious if the SAME machine with the same KODI settings as mentioned above would exhibit the problem with an NFS source and LOCALDB so I was able to try it out and not reproduce the issue. This led me to thinking maybe it was something with MySQL so I attempted Kodi to use a new MySQL name instead of the default and basically have a new Database. I then tested the same test case with a new MySQL DB of a custom name and didn't have the problem. Reverted back to my MySQL DB and had the problem, so I wiped the DB from the MySQL server and had Kodi recreate it with no issues. I didn't lose much since I purge my watched shows but not sure if this was just something maybe carried over recently but I am still at a loss for words.

[Milhouse]

New LibreELEC.tv Leia build #1008: RPi / RPi2
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 4.13.5 #1 Sun Oct 8 21:06:14 BST 2017 armv6l GNU/Linux

# vcgencmd version
Oct  6 2017 13:59:43
Copyright (c) 2012 Broadcom
version bde38f02b236b6e1bd16eb0f9b44b35f5a85cb30 (clean) (release)

# lsb_release
LibreELEC (Milhouse): devel-20171008210504-#1008-gadf3f57 [Build #1008]

# Kodi version
(18.0-ALPHA1 Git:623d2ce). Platform: Linux ARM 32-bit

Based on tip of LibreELEC.tv master (adf3f57, changelog) and tip of XBMC master (623d2ce, changelog) with the following modifications:

• Includes newclock5 patches
  
• Excludes the LibreELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-4.13.y
  
• Includes latest bcm2835-driver master (2669578, ahead +1)
  
• Includes latest kodi-platform master (36fb493)
  
• Includes latest libcec master (8adc786, ahead +26)
  
• Includes latest libnfs master (2911afe, ahead +103)
  
• Includes latest p8-platform master (a822e19)
  
• Includes latest addons: inputstream.adaptive (dde3921), inputstream.rtmp (26260c9), peripheral.joystick (cc14287), pvr.argustv (23cc0e8), pvr.demo (122dedd), pvr.dvblink (816f90c), pvr.dvbviewer (9765663, +1), pvr.filmon (7a9af25, +1), pvr.hdhomerun (ddfe2cb, +1), pvr.hts (0530971, +15), pvr.iptvsimple (ac4f4bd), pvr.mediaportal.tvserver (fa1069d, +1), pvr.mythtv (8965048), pvr.nextpvr (5b7caa2, +1), pvr.njoy (cc1cb56), pvr.octonet (e9b4c05), pvr.pctv (549f690), pvr.stalker (e15edca), pvr.teleboy (1cf8e96, +2), pvr.vbox (56052da, +1), pvr.vdr.vnsi (f3f80d5), pvr.vuplus (c53478d, +1), pvr.wmc (02edba6, +1), pvr.zattoo (72651ae, +2)
  
• Include [env] compare: dmidecode: initial package
  
• Include [env] patch: RPi/RPi2: enable Broadcom WiFi debugging (see details)
  
• Include [env] patch: libcec: don't link non-existant libtinfo
  
• Include [env] patch: rev hack for kodi
  
• Include [env] patch: tvheadend42: fix makefile patch
  
• Include [env] patch: media_build: disable due to conflicts
  
• Include [env] patch: Add experimental splash video for RPi
  
• Include [env] patch: Bump included addon versions to prevent online updates
  
• Include [env] patch: HACK: Disable multiple PVR addons during migration. Always enable inputstream.* and os.*
  
• Include [env] patch: Add kodi binary addons (pvr, adsp, inputstream, vfs, other)
  
• Include [env] patch: pvr addons: disable broken addons
  
• Include [env] PR:1776: Don't cause system to hang on halt/reboot when net-booting
  
• Include [env] PR:1791: nss/nspr: include in main image [rfc]
  
• Include [env] PR:1835: add support for optional kernel modules in image and from addons
  
• Include [env] PR:1978: linux: update to linux-4.13.4
  
• Include [env] PR:2002: linux: include HBR and ELD fixes for Intel
  
• Include [env] PR:2061: added EFI 32bit support via GRUB2
  
• Include [env] PR:2075: xorg-server: update to xorg-server-1.19.4
  
• Include [pkg] patch: skin.estuary: reduce system info font size (kodi)
  
• Include [pkg] PR:234: implement intensity compensation for VC-1 decoding for Ivybridge and better (intel-vaapi-driver)
  
• Include [pkg] PR:260: Fix Seeking with VC1 after 6702b01ee3598017f05374bfc7956493f790361c (intel-vaapi-driver)
  

Build Highlights:

1. Minors
   

Build Details:

1. LibreELEC.tv:
   • mesa: update to 17.2.2 (PR:2048, 2 commits, 1 file changed)
     
   • buildsystem: unset all PKG_* variables before sourcing a new package (PR:2072, 1 commit, 2 files changed)
     
2. XBMC:
   • libcdio: add osx 10.13 patch (PR:12840, 1 commit, 2 files changed)
     
   • FFmpeg: Bump to 3.3.4-Leia-Alpha-1 (PR:12894, 1 commit, 1 file changed)
     
   • [dxva] add VP9 profile 0 hardware decoder (PR:12887, 2 commits, 1 file changed)
     
3. pvr.zattoo:
   • Multiple improvements (452e089)
     
   • Fix race conditions while closing the addon (72651ae)
     
4. newclock5:
   • Commits no longer in build:
     • ffmpeg: Bump to 3.3.4 (3124d94d)
       

[Milhouse]

UPDATE: I was curious if the SAME machine with the same KODI settings as mentioned above would exhibit the problem with an NFS source and LOCALDB so I was able to try it out and not reproduce the issue. This led me to thinking maybe it was something with MySQL so I attempted Kodi to use a new MySQL name instead of the default and basically have a new Database. I then tested the same test case with a new MySQL DB of a custom name and didn't have the problem. Reverted back to my MySQL DB and had the problem, so I wiped the DB from the MySQL server and had Kodi recreate it with no issues. I didn't lose much since I purge my watched shows but not sure if this was just something maybe carried over recently but I am still at a loss for words.

Sounds like a corrupt database, or incorrect metadata - did you check the runtime for S01E04 to make sure it was correct and not ridiculously short? Maybe Kodi is calculating 90% of the runtime metadata field and if that field is massively incorrect then Kodi might think that 10 minutes is 90% of the run time and bump the playcount. If you are using NFOs with embedded streamdetails then check them to make sure the calculated runtime is correct.

I've just checked with 1008 and a MySQL db and Kodi is spot on when setting the playcount - an episode that is 1:01:43 (3703 seconds) long bumps the playcount when stopping after 00:55:32 (3332 seconds), which is 90%, and not before.

[andiewill]

@ Millhouse 

 DVB-C/T2 USB Stick -> Xbox One Digital TV Tuner

I have just bought this stick using it with a Rpi3 libreelec latest test build and it works fine picking up HD / Sd UK channels with no issues, when i update the image using the millhouse nightly build it does not show the channels and give an TV headend error, i believe that it may be something to do with the drivers ? can his be implemented into the millhouse build by any chance, or am i missing something ?

[Milhouse]

can his be implemented into the millhouse build by any chance, or am i missing something ?

https://forum.kodi.tv/showthread.php?tid...pid2650389

[pyrodex]

UPDATE: I was curious if the SAME machine with the same KODI settings as mentioned above would exhibit the problem with an NFS source and LOCALDB so I was able to try it out and not reproduce the issue. This led me to thinking maybe it was something with MySQL so I attempted Kodi to use a new MySQL name instead of the default and basically have a new Database. I then tested the same test case with a new MySQL DB of a custom name and didn't have the problem. Reverted back to my MySQL DB and had the problem, so I wiped the DB from the MySQL server and had Kodi recreate it with no issues. I didn't lose much since I purge my watched shows but not sure if this was just something maybe carried over recently but I am still at a loss for words.

Sounds like a corrupt database, or incorrect metadata - did you check the runtime for S01E04 to make sure it was correct and not ridiculously short? Maybe Kodi is calculating 90% of the runtime metadata field and if that field is massively incorrect then Kodi might think that 10 minutes is 90% of the run time and bump the playcount. If you are using NFOs with embedded streamdetails then check them to make sure the calculated runtime is correct.

I've just checked with 1008 and a MySQL db and Kodi is spot on when setting the playcount - an episode that is 1:01:43 (3703 seconds) long bumps the playcount when stopping after 00:55:32 (3332 seconds), which is 90%, and not before.

Not to beat a dead horse but I think I found the cause of this and it wasn't a corrupt DB issue. On a few select clients I have video cache settings to help with skipping video when background disk IO is heavy on the NAS or even on remote VPN clients (for when I travel). I used http://kodi.wiki/view/HOW-TO%3AModify_the_video_cache as the model and have Example 1 on my MAC Laptop for when I travel and Example 4 on my main RPI3s.

I think Kodi is caching the entire file, which is expected behavior, but once the cache is there it is causing a playcount to be incremented even though the show isn't over. I can reproduce this behavior on a system with a fresh Kodi directory with advanced settings for the MySQL AND caching, without caching I can't reproduce this issue.