On Wed, Mar 4, 2020 at 10:47 AM Carl Eugen Hoyos <[email protected]> wrote:
> Am Mi., 4. März 2020 um 18:57 Uhr schrieb James Almer <[email protected]>: > > > > On 3/4/2020 1:56 PM, Carl Eugen Hoyos wrote: > > > Am Mi., 4. März 2020 um 17:53 Uhr schrieb Sebastian Hubbard > > > <[email protected]>: > > >> > > >> I knew this was an ABI change and probably wasn't going to be > > >> acceptable, but I decided to share it for consideration anyway. > > >> There's special code in av_probe_input_format3 to skip over id3 tags > > >> for the purposes of format detection, but Apple uses id3 tags to > > >> signal information for HLS audio only tracks. Maybe this change > > >> shouldn't be made, but if it could ever be helpful for another format > > >> prober to know about leading id3 tags this is the only way I could > > >> imagine it. > > > > > > If the additional fields are necessary (I don't know), they could be > > > added to the end of the struct where they do not break abi. > > > > > > Carl Eugen > > > > sizeof(AVProbeData) is part of the ABI because its doxy doesn't strictly > > state otherwise. This means we can't add new fields to it, even at the > > end, until a major bump. Else we'll be breaking ABI. > > Thank you both for the explanation! > > Carl Eugen > _______________________________________________ > ffmpeg-devel mailing list > [email protected] > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel > > To unsubscribe, visit link above, or email > [email protected] with subject "unsubscribe". Hi everyone. What is the recommended path forward here? To get past the ABI break I think this patch needs: * libavformat minor version number bump (and reset of micro version) * soname version bump Is there any other changes needed? -- Nikolas Bowe | SWE | [email protected] | 408-565-5137 <(408)%20565-5137> _______________________________________________ ffmpeg-devel mailing list [email protected] https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
