[FFmpeg-devel] [PATCH] Clip-wrapped MXF support (new attempt)

Maksym Veremeyenko verem
Thu Apr 15 12:26:32 CEST 2010

Baptiste Coudurier ???????(??):
> Hi,
> On 04/14/2010 06:57 AM, Maksym Veremeyenko wrote:
>> *0001-revert-MXFCodecUL-struct-modification-from-r11567.patch*
>> *0002-partially-revert-essence_container-from-r11567.patch*
>> reverts some changes from r11567 that simplify wrapping detection but
>> work wrong some items.
> That may be right. Can you please add a new struct only for the 
> container uls ? All the codecs uls are not wrapping scheme related.

add new *MXFContainerUL* struct with additional wrapping parameter and 
added appropriate search by id. Please refer to updated patches:


>> *0003-make-key-output-in-RP-224.10-form.patch*
>> on debug output PRINT_KEY in a form represented in
>> /RP224v10-publication-20081215/ with dots as hex codes separator
>> *0004-add-new-essence-container-uls-and-wrapping-methods.patch*
>> adds known (or maybe possible in my case) container uls with proper wrap
>> method
> Will be added later when code actually use them.
actually it used, because of P2's audio mxf is clip-wrapped too, but not 
listed in original ids list.

Additionally its easy to search in the RP-224.10 tables for appropriate 
id to check for container and codec id when it formated in the same manner.

updated patches:


> Also, the code should match the codec first then try the container, to 
> avoid avoid huge tables.
May be in future it has a sense to create mxfdict.c with /huge/ table 
with additional codec parameters and dataset types linking to UID?

>> *0005-extend-MXFIndexTableSegment.patch*
>> extends *MXFIndexTableSegment* struct members and appropriate read 
>> function
> Logging without context by default is not really ok. Please keep it 
> dprintf for now, or extends func to take AVFormatContext as parameter.

av_log replaced with dprintf, patch attached:


Maksym Veremeyenko

More information about the ffmpeg-devel mailing list