It's actually the same indeed.

Looks like a severe case of analysis paralysis on my side :)

On Thu, May 2, 2013 at 10:41 AM, Anton Khirnov <[email protected]> wrote:
>
> On Tue, 30 Apr 2013 17:19:21 +0200, Jindřich Makovička <[email protected]> 
> wrote:
>> On Mon, Apr 29, 2013 at 9:10 PM, Anton Khirnov <[email protected]> wrote:
>> >
>> > On Mon, 29 Apr 2013 12:10:50 +0200, Jindřich Makovička 
>> > <[email protected]> wrote:
>> >> Hi,
>> >>
>> >> this patch adds original and copyright flags from MPEG audio headers
>> >> to AVFormatContext, so they can be displayed by avprobe.
>> >>
>> >> Could this feature be included in upstream? We encountered MPEG-PS
>> >> sources that (ab)use the "original" flag to distinguish between
>> >> original and dubbed audio, so it would come handy.
>> >>
>> >
>> > I'm not against exposing this information in principle, but I don't like 
>> > adding
>> > a format-specific field to AVCodecContext. History shows that it doesn't 
>> > lead to
>> > anything good.
>> >
>> > I see two possible ways to expose this.
>> > One is through decoder-private AVOptions, which contrary to the name can 
>> > also be
>> > used to retrieve values from opaque structs.
>>
>> I tried this approach before realizing that avctx->priv_data is
>> discarded by avcodec_close(), and unavailable after
>> avformat_find_stream_info() anyway.
>>
>> > Another is through newly added frame side data/metadata.
>>
>> Is it acceptable to simply export the 32bit MPEG header as a side data?
>>
>
> How is this better than simply looking at the first four bytes of the packet
> data?
>
> --
> Anton Khirnov



-- 
Jindřich Makovička
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to