Cleanup of DVD startup to improve playback reliability

[peper03@…]

The startup code for DVD playback jumps around somewhat trying to find a playable title, which is then used to create audio/video decoders before playback is restarted at firstplay.

This doesn't always work as it can leave the DVDs VM in an inconsistent state and, depending on the DVD, looks ugly as playback starts and then restarts or appears to jump somewhere else after a couple of seconds. Because the VM can be left in an inconsistent state, menu navigation/selection issues can arise, which can make it difficult to start playback.

The attached patch attempts to clean up the startup code. The DVDRingBuffer no longer tries to find a playable title in OpenFile?, but rather just lets the VM get on with playback. AVFormatDecoder no longer processes the initial frames more than is required to create decoders. Once this has been achieved, any cached data is flushed before jumping back to 'firstplay' (and cleanly resetting the VM). By not displaying the initial frames used to determine the decoders and flushing the cached data, the ugly jump at the start of some DVDs is no longer visible.

[Richard <peper03@…>]

In 414de880b9d65263849e2339df9e083df0301033/mythtv:

Simplify DVD startup to improve playback reliability

The startup code for DVD playback jumps around somewhat trying to find
a playable title, which is then used to create audio/video decoders
before playback is restarted at firstplay.

This doesn't always work as it can leave the DVDs VM in an
inconsistent state and, depending on the DVD, looks ugly as playback
starts and then restarts or appears to jump somewhere else after a
couple of seconds. Because the VM can be left in an inconsistent
state, menu navigation/selection issues can arise, which can make it
difficult to start playback.

The attached patch attempts to clean up the startup code. The
DVDRingBuffer no longer tries to find a playable title in OpenFile??,
but rather just lets the VM get on with playback. AVFormatDecoder no
longer processes the initial frames more than is required to create
decoders. Once this has been achieved, any cached data is flushed
before jumping back to 'firstplay' (and cleanly resetting the VM). By
not displaying the initial frames used to determine the decoders and
flushing the cached data, the ugly jump at the start of some DVDs is
no longer visible.

Fixes #11288

Signed-off-by: Stuart Morgan <smorgan@…>

[peper03@…]

Same patch just adapted to apply cleanly to the current 0.26-fixes

[Richard <peper03@…>]

In c1929686fd424b51c9725100e2e326525c879b0d/mythtv:

Simplify DVD startup to improve playback reliability

The startup code for DVD playback jumps around somewhat trying to find
a playable title, which is then used to create audio/video decoders
before playback is restarted at firstplay.

This doesn't always work as it can leave the DVDs VM in an
inconsistent state and, depending on the DVD, looks ugly as playback
starts and then restarts or appears to jump somewhere else after a
couple of seconds. Because the VM can be left in an inconsistent
state, menu navigation/selection issues can arise, which can make it
difficult to start playback.

The attached patch attempts to clean up the startup code. The
DVDRingBuffer no longer tries to find a playable title in OpenFile???,
but rather just lets the VM get on with playback. AVFormatDecoder no
longer processes the initial frames more than is required to create
decoders. Once this has been achieved, any cached data is flushed
before jumping back to 'firstplay' (and cleanly resetting the VM). By
not displaying the initial frames used to determine the decoders and
flushing the cached data, the ugly jump at the start of some DVDs is
no longer visible.

Fixes #11288

Backport from master [414de880b]

Signed-off-by: Stuart Morgan <smorgan@…>

[Richard <peper03@…>]

In 58035465336d74892630ca6e5d213a2510ed82b2/mythtv:

Update angle information and chapter times that was indirectly
completely disabled by the fix for #11288

Fixes #11372, Refs #11288

Signed-off-by: Stuart Morgan <smorgan@…>