[virtuallysafe]

I'd actually suggest that anyone else with duplicate actors does NOT manually fix the problem, in case there is an automated solution that requires testing. I don't particularly fancy going back to v89 just to test this...

I'd tend to agree, it wasn't much fun. That said, I'm happy to roll back to v90 and test if a fix is released (I luckily took a backup prior to manually removing the duplicates).

[menakite]

For the records, no issues here upgrading to v91 - but my video DB has been wiped and rebuilt recently (misconfiguration in service.watchdog and a NAS going offline...).

Thanks for getting the updated build out Milhouse, Merry Christmas to you and yours. Thanks Menakite for fixing the scrolling issue, hope to see it in xbian soon as well.

Hey - updated patch has been pushed to Git, but the package needs rebuilding (by mk01). I guess that'll happen in early January.

[Milhouse]

New OpenELEC I****** build: #1224
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 3.18.1 #1 PREEMPT Wed Dec 24 21:04:06 GMT 2014 armv6l GNU/Linux

# vcgencmd version
Dec 19 2014 18:40:20
Copyright (c) 2012 Broadcom
version 5abd572e2ed1811283443387af09377b95501c50 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20141224210250-r19859-gbaeddfd [Build #1224]

Based on tip of OpenELEC master (baeddfd1, changelog) and tip of XBMC master (e0e7f826, changelog) with the following modifications:

• Includes newclock4 patches
  
• Excludes the OpenELEC fernetmenta patches due to conflicts with newclock4
  
• Excludes the OpenELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-3.18.y
  
• Excludes the OpenELEC kodi-001-helix_rpb_backports patch in favour of sourcing these and possibly more recent patches directly from newclock4 branch
  
• Default setting for "Show RSS Feed" changed to disabled (new installs only) [patch details]
  
• Disabled "Total Duration" in Confluence (see build #0221 for details)
  
• Adapted service.openelec.settings to take advantage of PR:5217 [patch details]
  
• Includes latest libnfs master (9c1d3848)
  
• Includes latest libcec master (9f000e43)
  
• Includes latest xbmc-pvr-addons master (89ba5bb6)
  
• Includes latest xbmc-addon-xvdr master (328fa653)
  
• Includes additional ffmpeg codecs/muxers enabled for testing/benchmarking purposes (see patch)
  
• Includes animated splash support (patch)
  
• Includes commits from libcec-2.2.0 (popcornmix)
  
• Exclude kodi-999.99-fix-3692.patch: Superseded by PR6005
  
• Exclude lirc_rpi.conf: No longer required, let Device Tree handle lirc_rpi (add "dtoverlay=lirc-rpi" to config.txt)
  
• Include patch: Enable kernel device-tree support (details)
  
• Include patch: Add device tree kernel trailer
  
• Include patch: Add Linux 3.18.y options
  
• Include patch: Add Linux 3.18.y compatible cfg80211_rx_mgmt
  
• Include patch: Add *.dtb files to firmware
  
• Include PR:3657: [rfc] kodi crashlogs
  
• Include PR:5747: [database] - add setting for hiding watched movies/episodes in recently added lists
  
• Include PR:5752: [jsonrpc] Allow retrieving empty TV shows via VideoLibrary.GetTVShows.
  
• Include PR:5958: Fix state not saved after #5842 if playing from "Recently added".
  

Build Highlights:

1. filesystem: add COverrideFile/COverrideDirectory to reduce code duplication
   

Build Details:

1. OpenELEC:
   • zlib: move source to our mirror (baeddfd1)
     
2. XBMC:
   • filesystem: add COverrideFile/COverrideDirectory to reduce code duplication (PR:5599, 4 commits, 16 files changed)
     
   • AESinkPULSE: Use const ref when finding the channel map (PR:6016, 1 commit, 1 file changed)
     
   • [droid] enable gnutls support in ffmpeg (PR:5979, 4 commits, 6 files changed)
     
   • upnp: add xbmc:uniqueidentifier (PR:6014, 3 commits, 7 files changed)
     
   • [art] fix up dfaacca54d4a8f78ee02a041d4d1648aa0b403a2 (PR:6017, 1 commit, 1 file changed)
     
   • cppcheck performance fixes in android/ (PR:5995, 1 commit, 2 files changed)
     
   • VAAPI: Enable VC1 by default - not much interlaced content there (PR:6018, 1 commit, 1 file changed)
     

[menakite]

I was wondering about:

Code:

g_windowManager.GetTopMostModalDialogID() == WINDOW_DIALOG_BUSY

although I think it wasn't always behaving as I expected.

Looks fine here, however:

Code:

m_pPlayer->IsPlaying() && !m_pPlayer->IsCaching()

seems better as DialogBusy is massively used elsewhere (e.g. slow plugins - not all skins, like Amber, have a visual feedback).

It becomes true here:

Code:

2014-12-25 01:43:40.757 kodi.bin[7951:18084591] Debug Print: ------ Window Init (DialogBusy.xml) ------
2014-12-25 01:43:40.758 kodi.bin[7951:18084591] Debug Print: Window DialogBusy.xml was already loaded
2014-12-25 01:43:40.758 kodi.bin[7951:18084591] Debug Print: Alloc resources: 0.01m
2014-12-25 01:43:40.760 kodi.bin[7951:18084591] Debug Print: ScreenSaverDisablesAutoScrolling: playing !caching

and then false after:

Code:

2014-12-25 01:43:43.699 kodi.bin[7951:18084591] Debug Print: ------ Window Deinit (DialogBusy.xml) ------
2014-12-25 01:43:43.699 kodi.bin[7951:18084591] Debug Print: PlayFile: OpenFile succeed, play state 2
2014-12-25 01:43:43.700 kodi.bin[7951:18084591] Debug Print: OnPlayBackStarted: play state was 2, starting 0
2014-12-25 01:43:43.701 kodi.bin[7951:18084591] Debug Print: ScreenSaverDisablesAutoScrolling: playing !caching
2014-12-25 01:43:43.705 kodi.bin[7951:18094813] Debug Print: CDVDPlayer::SetCaching - caching state 0

I have to say that I'm not sure what this "caching state" is all about - it's not the file cache (could it return true while playing so that if you press "i" and the skin shows the plot is stops scrolling when it shouldn't?).

Edit: in fact it does. This works better:

Code:

m_pPlayer->IsPlaying() && g_windowManager.IsWindowActive(WINDOW_DIALOG_BUSY)

I honestly can't tell if it makes any difference and how noticeable that would be (has little to no effect at all on OS X).

[Milhouse]

New OpenELEC I****** build: #1225
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 3.18.1 #1 PREEMPT Thu Dec 25 22:57:47 GMT 2014 armv6l GNU/Linux

# vcgencmd version
Dec 19 2014 18:40:20
Copyright (c) 2012 Broadcom
version 5abd572e2ed1811283443387af09377b95501c50 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20141225225635-r19862-g5e301b4 [Build #1225]

Based on tip of OpenELEC master (5e301b42, changelog) and tip of XBMC master (3cf804cb, changelog) with the following modifications:

• Includes newclock4 patches
  
• Excludes the OpenELEC fernetmenta patches due to conflicts with newclock4
  
• Excludes the OpenELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-3.18.y
  
• Excludes the OpenELEC kodi-001-helix_rpb_backports patch in favour of sourcing these and possibly more recent patches directly from newclock4 branch
  
• Default setting for "Show RSS Feed" changed to disabled (new installs only) [patch details]
  
• Disabled "Total Duration" in Confluence (see build #0221 for details)
  
• Adapted service.openelec.settings to take advantage of PR:5217 [patch details]
  
• Includes latest libnfs master (9c1d3848)
  
• Includes latest libcec master (9f000e43)
  
• Includes latest xbmc-pvr-addons master (89ba5bb6)
  
• Includes latest xbmc-addon-xvdr master (328fa653)
  
• Includes additional ffmpeg codecs/muxers enabled for testing/benchmarking purposes (see patch)
  
• Includes animated splash support (patch)
  
• Includes commits from libcec-2.2.0 (popcornmix)
  
• Exclude kodi-999.99-fix-3692.patch: Superseded by PR6005
  
• Exclude lirc_rpi.conf: No longer required, let Device Tree handle lirc_rpi (add "dtoverlay=lirc-rpi" to config.txt)
  
• Include patch: Enable kernel device-tree support (details)
  
• Include patch: Add device tree kernel trailer
  
• Include patch: Update to Linux 3.18.1
  
• Include patch: Add Linux 3.18.y options
  
• Include patch: Add Linux 3.18.y compatible cfg80211_rx_mgmt
  
• Include patch: Add *.dtb files to firmware
  
• Include PR:3657: [rfc] kodi crashlogs
  
• Include PR:5747: [database] - add setting for hiding watched movies/episodes in recently added lists
  
• Include PR:5752: [jsonrpc] Allow retrieving empty TV shows via VideoLibrary.GetTVShows.
  
• Include PR:5958: Fix state not saved after #5842 if playing from "Recently added".
  

Build Highlights:

1. cppcheck performance fixes in cores/playercorefactory
   

Build Details:

1. OpenELEC:
   • Intel: Backport drm/i915: Disable PSMI sleep messages on all rings around (PR:3716, 2 commits, 4 files changed)
     
2. XBMC:
   • cppcheck performance fixes in cores/playercorefactory (PR:6002, 1 commit, 1 file changed)
     

[expandables]

Hi is it possible to add the animated splash patch in xbmc gotham?

[Milhouse]

Hi is it possible to add the animated splash patch in xbmc gotham?

http://forum.kodi.tv/showthread.php?tid=...pid1861348

[bboo]

How is airplay working with ios 8? is it any better than in previous versions?

[Milhouse]

New OpenELEC I****** build: #1226
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 3.18.1 #1 PREEMPT Fri Dec 26 21:03:30 GMT 2014 armv6l GNU/Linux

# vcgencmd version
Dec 19 2014 18:40:20
Copyright (c) 2012 Broadcom
version 5abd572e2ed1811283443387af09377b95501c50 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20141226210155-r19865-g240acec [Build #1226]

Based on tip of OpenELEC master (240acecc, changelog) and tip of XBMC master (bcaa8f2f, changelog) with the following modifications:

• Includes newclock4 patches
  
• Excludes the OpenELEC fernetmenta patches due to conflicts with newclock4
  
• Excludes the OpenELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-3.18.y
  
• Excludes the OpenELEC kodi-001-helix_rpb_backports patch in favour of sourcing these and possibly more recent patches directly from newclock4 branch
  
• Default setting for "Show RSS Feed" changed to disabled (new installs only) [patch details]
  
• Disabled "Total Duration" in Confluence (see build #0221 for details)
  
• Adapted service.openelec.settings to take advantage of PR:5217 [patch details]
  
• Includes latest libnfs master (9c1d3848)
  
• Includes latest libcec master (9f000e43)
  
• Includes latest xbmc-pvr-addons master (89ba5bb6)
  
• Includes latest xbmc-addon-xvdr master (328fa653)
  
• Includes additional ffmpeg codecs/muxers enabled for testing/benchmarking purposes (see patch)
  
• Includes animated splash support (patch)
  
• Includes commits from libcec-2.2.0 (popcornmix)
  
• Exclude kodi-999.99-fix-3692.patch: Superseded by PR6005
  
• Exclude lirc_rpi.conf: No longer required, let Device Tree handle lirc_rpi (add "dtoverlay=lirc-rpi" to config.txt)
  
• Include patch: Enable kernel device-tree support (details)
  
• Include patch: Add device tree kernel trailer
  
• Include patch: Update to Linux 3.18.1
  
• Include patch: Add Linux 3.18.y options
  
• Include patch: Add Linux 3.18.y compatible cfg80211_rx_mgmt
  
• Include patch: Add *.dtb files to firmware
  
• Include PR:3657: [rfc] kodi crashlogs
  
• Include PR:5747: [database] - add setting for hiding watched movies/episodes in recently added lists
  
• Include PR:5752: [jsonrpc] Allow retrieving empty TV shows via VideoLibrary.GetTVShows.
  
• Include PR:5958: Fix state not saved after #5842 if playing from "Recently added".
  

Build Highlights:

1. gettext and libpng updates
   

Build Details:

1. OpenELEC:
   • gettext: update to gettext-0.19.4 (6dc65b89)
     
   • libpng: update to libpng-1.6.16 (f392b987)
     
2. XBMC:
   • binary addons: replace prebuilt dependencies with proper builds (PR:6013, 6 commits, 20 files changed)
     
   • cppcheck performance fixes in cores/dllloader/ (PR:5998, 1 commit, 1 file changed)
     
   • cppcheck performance fixes in cores/AudioEngine (PR:5997, 1 commit, 3 files changed)
     
   • cppcheck fixes in cores/dvdplayer/ (PR:5999, 2 commits, 6 files changed)
     
   • [droid] More mediacodec fixes (PR:6022, 2 commits, 2 files changed)
     

[virtuallysafe]

I'd actually suggest that anyone else with duplicate actors does NOT manually fix the problem, in case there is an automated solution that requires testing. I don't particularly fancy going back to v89 just to test this...

Further to this, I have been trying to think of potential sources of the dodgy database and recalled some time ago I performed a full export of my DB to separate NFO files and re-imported to a new DB (only using the 'Local Information Only' scraper). To validate whether this could have introduced the problem, I performed the same process again on KodiSetup-20141220-4453376-master (v90 DB) and tried running KodiSetup-20141225-3cf804c-master (v91 DB).

The DB upgrade from v90 to v91 also failed, log can be found here. Once again searching the v90 DB indicates there are duplicates in the 'actors' table, however, not as many as previously.

Could this possibly be a problem with the NFO scraper?

[Milhouse]

Further to this, I have been trying to think of potential sources of the dodgy database and recalled some time ago I performed a full export of my DB to separate NFO files and re-imported to a new DB (only using the 'Local Information Only' scraper). To validate whether this could have introduced the problem, I performed the same process again on KodiSetup-20141220-4453376-master (v90 DB) and tried running KodiSetup-20141225-3cf804c-master (v91 DB).

The DB upgrade from v90 to v91 also failed, log can be found here. Once again searching the v90 DB indicates there are duplicates in the 'actors' table, however, not as many as previously.

Could this possibly be a problem with the NFO scraper?

That's very interesting (though I'm not sure what KodiSetup-20141220-4453376-master is, presumably this is one of my builds?)

I tend to use the Universal Movie and TMDB scrapers myself, but exclusively use local NFOs which should be used in preference to the internet but could be resulting in the same behaviour as the Local Information Only scraper.

Unfortunately I won't be able to confirm until Monday (travelling - I've scheduled builds for the next couple of days which may or may not work, and there won't be any release posts until I get back). What would be really useful is to narrow down the NFOs that are responsible for the duplicate actors (or at least some of the duplicate actors), which may help give a better idea of what is happening.

Edit: Ah, OK - KodiSetup-20141220-4453376-master is not one of my builds, it's x86. Also good to know that it's not something I've introduced.

[virtuallysafe]

Edit: Ah, OK - KodiSetup-20141220-4453376-master is not one of my builds, it's x86. Also good to know that it's not something I've introduced.

Apologies, yes - the builds I was testing were Windows mainline to ensure the issue is there (and not considered to be with some 'external tool that writes directly into our database' ).

I have a debug log of the fresh v90 NFO Import (and associated errors) as well as the original NFOs, however, they are perhaps a little too verbose to post so will work on cutting them up - or PM me when you have some time or are keen...


As an aside, Merry Christmas! As I'm sure anyone who follows this thread would attest, love your work Milhouse (and Popcornmix). I have been actively watching the mainline for a number of years, running nightly alpha XBMC versions on the original ATV, ATV2's and now RPi's. This whole RPi thing (after the work of you guys) is an absolute game changer. For what its worth - sincere thanks.

[gendo]

Hi milhouse & popcornmix first of all thanks for all you effort. would it possible to have a one off build of openelec 5 final(once available) + helix final + all the newclock 4 optimizations? I'm asking this as i run multiple kodi devices ( not all RPI) with a shared mysql and livetv and milhouse builds are much faster for livetv then plain openelec. but if i use latest milhouse build i will end up with 2 mysql db versions..

[popcornmix]

Hi milhouse & popcornmix first of all thanks for all you effort. would it possible to have a one off build of openelec 5 final(once available) + helix final + all the newclock 4 optimizations? I'm asking this as i run multiple kodi devices ( not all RPI) with a shared mysql and livetv and milhouse builds are much faster for livetv then plain openelec. but if i use latest milhouse build i will end up with 2 mysql db versions..

OpenELEC 5.0 final will include my "helix_rbp_backports" commits, which are the commits from newclock4 that are well tested.
It should contain (virtually) all the current optimisations and should run well.

I imagine most users who run newclock4 builds for performance reasons will find switching to official OE 5.0 will be just as good.
We'd obviously like to keep some newclock4 testers to try new features and help ensure Helix+1 will work well.

I'm sure OE 5.0 will be released very soon now (Helix going final so close to Christmas is obviously delaying the OE release).

[gendo]

Hi milhouse & popcornmix first of all thanks for all you effort. would it possible to have a one off build of openelec 5 final(once available) + helix final + all the newclock 4 optimizations? I'm asking this as i run multiple kodi devices ( not all RPI) with a shared mysql and livetv and milhouse builds are much faster for livetv then plain openelec. but if i use latest milhouse build i will end up with 2 mysql db versions..

OpenELEC 5.0 final will include my "helix_rbp_backports" commits, which are the commits from newclock4 that are well tested.
It should contain (virtually) all the current optimisations and should run well.

I imagine most users who run newclock4 builds for performance reasons will find switching to official OE 5.0 will be just as good.
We'd obviously like to keep some newclock4 testers to try new features and help ensure Helix+1 will work well.

I'm sure OE 5.0 will be released very soon now (Helix going final so close to Christmas is obviously delaying the OE release).

Thanks popcornmix even i thought so, but live tv takes much longer to start with openelec 4.73 compared to these nightly builds..