[FFmpeg-devel] [PATCH 1/3] utils.c: when packets are written by s->oformat->write_packet( ), add output packet size to out_bytes.
michael at transparentpixel.com
Wed Aug 29 22:24:16 CEST 2012
In this instance, I completely agree with the notion of not wanting to break things for others in the world. I'm going to pull the proposed changes to util.c out entirely.
With respect to the outstanding items in the implementation of RTSP itself, is there a ticket for
that issue? Thanks for your insight on this front.
On Aug 14, 2012, at 2:49 PM, Nicolas George wrote:
> L'octidi 28 thermidor, an CCXX, Michael Smithng a écrit :
>> The return values (considering the changes to ffmpeg.c) are being used.
>> This is how we're getting the actual number of bytes written to the ost
>> rather than just bytes decoded.
>> The return value is still consistent with the way it's being used
>> elsewhere in the code. Previously the return value was either 0 for
>> success or negative for an error. It's still positive for success but now
>> can have a non-zero value. Since the value was checked for negative or
>> positive, this will not affect its use.
> The docs says:
> * @return 0 on success, a negative AVERROR on error.
> There are certainly some applications out in the world that check it as
> "if (ret)" and not "if (ret < 0)", such a change would break them.
>> We tried using AVPacket.size but the return values were less than useful
>> for the purpose.
> A quick look into the code seems to indicate that the problem lies in the
> implementation of the RTSP protocol, that does not update the pos field of
> the "main" protocol according to the progress of the individual streams.
> Nicolas George
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
More information about the ffmpeg-devel