On Wed, Sep 07, 2022 at 02:12:46PM +0100, Derek Buitenhuis wrote:
> On 9/6/2022 10:47 PM, Mohammad Izadi wrote:
> > +    if (side_data && side_data_size > 0)
> > +        ff_write_dynamic_hdr10_plus_to_full_itu_t_t35((AVDynamicHDRPlus*)side_data, &hdr10_plus_itu_t_t35, &hdr10_plus_itu_t_t35_size);
> 
> You can't use ff_-prefixed functions across library boundaries.
> 
> It nees to be either public (av*) or avpriv. I suspect people won't want it to
> be avpriv.
> 
> Personally, I think having serialization as a public API is useful, but YMMV. Mostly
> because I was just writing my own serialization to make use of the exported side data :P.

I agree

on a related subject, side data serialization should be moved to a common API
We have common APIs for parsers, decoder, bitstream filters but for 
parsing/decoding side data this is heading toward something less structured

Above is not a comment on this patch, the patch is fine. I just want to point
to this before we have several dozen such functions which need to be
deprecated and supported when a more structured system is introduced

thx


[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Into a blind darkness they enter who follow after the Ignorance,
they as if into a greater darkness enter who devote themselves
to the Knowledge alone. -- Isha Upanishad