[FFmpeg-devel] [PATCH]Move MLP channel layout setting from parser to decoder

John Stebbins stebbins at jetheaddev.com
Wed Apr 13 16:55:53 CEST 2011

On 04/13/2011 01:17 AM, Carl Eugen Hoyos wrote:
> John Stebbins<stebbins<at>  jetheaddev.com>  writes:
>>> This allows the mlp channel re-order patch to also work with other
>>> applications than ffmpeg/ffplay.
>> Is there an order of operations problem here? channel_layout gets set before
>> channels in read_major_sync.  If channels hasn't been set yet, doesn't the
>> test for av_get_channel_layout_nb_channels(channel_layout) != channels fail
>> the first time through?
> It does not trigger for me (I only tested one sample), but I will change it to
> if (channels&&  av_get_channel_layout_nb_channels ...
> Any other comments?
> Thank you, Carl Eugen
I was thinking the test needed to be moved to after where channels gets set.  I don't entirely understand under what 
conditions that test would be triggered, but it seems like in cases where it would be triggered, the current position of 
that test would cause it to trigger on the *next* call to read_access_unit after the number of channels change.  So for 
one packet, the value would be incorrect.

More information about the ffmpeg-devel mailing list