Hi,
On 28.06.2015 16:10, Paul B Mahol wrote:
> As you may already know new asf demuxer hit the tree but did not
> replace the old one.
>
> Here I ask for samples which show that one is better than another,
> whichever that one is.
For starters, the new asf demuxer crashes, where the old one didn't:
int nchunks = pkt_len / chunk_len;
(chunk_len can be 0 here.)
Then the new one can leak memory if asf_read_header fails.
It probably should call asf_read_close on every failure path.
Additionally the new demuxer can hang in the while loop in
detect_unknown_subobject, if size is 0 in asf_read_unknown,
so that avio_skip(pb, size - 24) undoes all progress of the
pb->buf_ptr.
If you ask me, the new demuxer needs more polishing before it can
sensibly be the default/only implementation.
Best regards,
Andreas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
