Quoting Nicolas George (2021-03-09 10:48:50) > Anton Khirnov (12021-03-08): > > The argument for internal side data is literally that it's not good > > enough to be made public. So the option of making it private is used as > > an excuse to avoid designing the format properly. > > This reasoning would apply to using private_ref as well. So I understand > your objection is against any solution that avoids properly designing > the structure. I can easily agree with that.
Correct, I object to using private_ref even more strongly than to private side data. I would prefer private_ref to be removed entirely. All this said, it would be acceptable to make this new side data private (or declared unstable) for a clearly limited period of time, e.g. 3 months or until the next major bump or so. As long as it's not internal indefinitely. > > > Lavfi is enough of a black box as it is, no need to make it even more so > > with internal side data formats. > > Please, there nothing of a black box about libavfilter. This is not even my claim, just something I've seen other people state repeatedly. People find the tight coupling between filters and graphs to be limiting. -- Anton Khirnov _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
