On Thu, Mar 8, 2012 at 12:27 PM, Kieran Kunhya <[email protected]> wrote:
> On Thu, Mar 8, 2012 at 8:09 PM, Alex Converse <[email protected]> wrote:
>> 2012/3/8 Måns Rullgård <[email protected]>:
>>> Alex Converse <[email protected]> writes:
>>>
>>>> On Thu, Mar 8, 2012 at 11:43 AM, Ronald S. Bultje <[email protected]> 
>>>> wrote:
>>>>> Hi,
>>>>>
>>>>> On Thu, Mar 8, 2012 at 11:41 AM, Alex Converse <[email protected]> 
>>>>> wrote:
>>>>>> From: Carl-Eugen Hoyos <[email protected]>
>>>>>>
>>>>>> Signed-off-by: Alex Converse <[email protected]>
>>>>>> ---
>>>>>>  libavformat/mpegts.c |    1 +
>>>>>>  1 files changed, 1 insertions(+), 0 deletions(-)
>>>>>
>>>>> LGTM.
>>>>>
>>>>
>>>> Thinking about this more I'm not really sure if it's correct at all.
>>>> Yes it fixes the particular sample but I think we should be honoring
>>>> the registration descriptor instead.
>>>>
>>>> Any thoughts?
>>>
>>> The registration descriptor should always be used.  If there is none,
>>> there's no telling what streams with private IDs contain.
>>>
>>
>> So why do we explicitly ignore it?
>
> Private data formats aren't handled correctly right now -
> STREAM_TYPE_PRIVATE_DATA only refers to 0x6 which is what 302M
> specifies.
> There's nothing stopping anyone from changing (or keeping the same)
> their 302M bitstream and sticking another stream_type from 0x80->0xff.
>
> So basically it's a bug. Either you could let a descriptor trump
> stream_type for everything or just make it trump stream_type for all
> types of private data.

Ok, that seems more clear. I've sent a new patch.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to