Le primidi 1er vendémiaire, an CCXXIII, Michael Niedermayer a écrit : > i dont know details about the application/lib using this, this patch > was just the result of some investigation into why this other patch > was needed: > "[FFmpeg-devel] [PATCH 1/2] mux: make sure that the AV_PKT_DATA_H263_MB_INFO > side data is received by the H.263/RTP packetizer (rfc2190)"
Ok. I just had a more careful look at this one, and my reaction is "WTF?!?". It seems very very wrong and comes without any explanations. > maybe > > the idea though was that a 5 line patch would avoid some application > redesign work. And it may very well be usefull in various other > scenarios. In general I can agree with that argument, but in this particular case I believe it does not apply for two reasons: First, the application will need some redesign anyway to make use of the merged-side-data hack. In the short run, I believe the work to do that may be less than a redesign to use the API properly, but only slightly. In the longer run, the application will be better with a proper design anyway, by far. Second, a 5-lines patch is ok as long as it does not create a maintenance burden. I believe this one very much does. Most critically, if someone endeavours to kill that ugly and fragile hack of merged-side-data, having it exposed to public API like that makes the task that much harder. Regards, -- Nicolas George
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
