Quoting James Almer (2024-03-01 21:42:28) > On 3/1/2024 3:19 PM, Anton Khirnov wrote: > > Quoting Jan Ekström (2024-02-29 17:42:54) > >> diff --git a/libavutil/frame.h b/libavutil/frame.h > >> index 47d0096bc4..908fd4a90d 100644 > >> --- a/libavutil/frame.h > >> +++ b/libavutil/frame.h > >> @@ -1081,6 +1081,26 @@ AVFrameSideData > >> *av_frame_side_data_new(AVFrameSideData ***sd, int *nb_sd, > >> enum AVFrameSideDataType type, > >> size_t size, unsigned int flags); > >> > >> +/** > >> + * Add a new side data entry to a set from an existing AVBufferRef. > >> + * > >> + * @param sd pointer to array of side data to which to add another > >> entry. > >> + * @param nb_sd pointer to an integer containing the number of entries in > >> + * the array. > >> + * @param type type of the added side data > >> + * @param buf AVBufferRef for which a new reference will be made > > > > This seems inconsistent with av_frame_new_side_data_from_buf(), which > > takes ownership of the buffer reference passed to it. > > Is there a strong reason to do it differently here? > > I asked for it. Taking ownership of the AVBufferRef and then expecting > the caller to discard the pointer to it is awkward, and unlike any other > function using AVBufferRef we have in the tree. > > The alternative, to keep the behavior of taking ownership of the passed > on reference, is to have this function take a pointer to pointer, and > clearing it on success.
That seems preferable to me, in most cases the caller does not want to keep a reference. -- 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".
