On Mon, Dec 11, 2017 at 08:37:43PM +0800, Tiejun.Peng wrote: > yes, it is. > maybe we should add every stream's status and the whole file status can't be > fail only by one of streams. > but the modification will be too much and the effect is extensive.
iam not sure i understand what you suggest but this issue can be avoided by avoiding the overlap between the error codes. simply adding a otherwise not used error code for signaling that data is truncated but everything is fine and code can continue should work. we can also use EOF as in your patch if people feel the advantages outweight the risk. I just dont want this to be done as undocumented sideeffect. If EOF return implies everything must have been fully consistently initialized then this must be documented [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB The bravest are surely those who have the clearest vision of what is before them, glory and danger alike, and yet notwithstanding go out to meet it. -- Thucydides
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list firstname.lastname@example.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel