On Sat, Aug 26, 2023 at 06:53:50PM +0200, Michael Niedermayer wrote: > Fixes: NoLegacy.ape > Found-by: Matt Ashland <[email protected]> > Signed-off-by: Michael Niedermayer <[email protected]> > --- > libavcodec/apedec.c | 106 +++++++++++++++++++++++++++++++++++--------- > 1 file changed, 84 insertions(+), 22 deletions(-)
Also, the same issue should exists in the mono codepath. I have no testcase for mono, so left this unchanged for now, as i cannot test it. But if someone has a testcase, tell me It also may be needed to wait till one of the CRCs to distinguish between the 2 decoding forms. (this is how the SDK does it) Waiting for some CRC is a bit inconvenient as we return data before currently. And this quicker variant here works for all files i have We also differ from the SDK by decoding both forms in parallel until one fails. The advantage here is that if only one is decoded then a random failure cannot be distinguished from "interim mode" which would result in a bitstream error causing a switch to the wrong mode. Above is how its done ATM in git master. As more issues get submitted that may change thx [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB I do not agree with what you have to say, but I'll defend to the death your right to say it. -- Voltaire
signature.asc
Description: PGP signature
_______________________________________________ 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".
