On Wed, May 17, 2017 at 2:25 PM, Nicolas George <geo...@nsup.org> wrote: > L'octidi 28 floréal, an CCXXV, Muhammad Faiz a écrit : >> Actually, one can argue that because we've already missed 3.3.1, we >> should wait for 3.3.2. But I can argue that master branch also has >> users, and keeping the crash unresolved at master branch is not good >> idea at all. > > The complete fix is almost in shape, it needs only a few adjustments. If > you are not happy with the time I can invest in it, feel free to take it > over. But pushing something that will conflict with it is not ok.
Of course, pushing the partial fix won't conflict with the complete fix. The partial fix aligns audio data unconditionally, but the complete fix aligns audio data when needed. > > (The biggest blocker, apparently, is the lack of test for the alignment > of linesize, which is not in the partial fix either. Ha ha ha.) Ha ha ha. Of course, that's why I prefer the partial fix. Because it is simple. We assume that at first data and linesize are always aligned, so we only need to check data[0] alignment. If data[0] is aligned, then data[n] and linesize are also aligned. On the contrary, the new API must be carefully designed. It must handle all corner cases. So, that's not a trivial task. (I see more problems in linesize handling). Thank's. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
