Andreas Rheinhardt (12020-09-09): > The documentation of ff_inlink_check_available_samples() contains the > note "on EOF and error, min becomes 1". Which means that > ff_inlink_check_available_samples() can be used to peek into the error > code of this inlink by setting min so high that the not enough samples > can be available. You seem to want to forbid it absolutely that a filter > knows whether there will be more data forthcoming after the data that is > already queued, so I am surprised that you have not also closed this > loophole in your patch just now.
It can be used to do that, but a filter that uses it is doing something very wrong. But in the meantime, ff_inlink_check_available_samples() is doing exactly what it is supposed to do, i.e. predict that ff_inlink_consume_samples() will succeed. I do not want to "close loopholes" because we are not in an adversarial situation. > For this filter to work it is necessary to know whether all inputs > except input 0 are finished; it can only start when they are finished. Then it is easy: the filter should consume the samples on its input, if there are none try to acknowledge the status change: if the status has changed, the input is finished. Checking beforehand is quite unnecessary. Regards, -- Nicolas George
signature.asc
Description: PGP signature
_______________________________________________ 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".
