[Ffmpeg-devel] Contacting developers by subversion name

Paul Richards paul.richards
Fri Dec 29 20:47:39 CET 2006


Apologies for the top post.

On 29/12/06, M?ns Rullg?rd <mru at inprovide.com> wrote:
>
> Don't top-post.
>
> "Paul Richards" <paul.richards at gmail.com> writes:
>
> > On 29/12/06, M?ns Rullg?rd <mru at inprovide.com> wrote:
> >> "Paul Richards" <paul.richards at gmail.com> writes:
> >>
> >> > Hi,
> >> > Apologies for spamming the mailing list.  Is there an easy way to
> >> > contact a developer given their subversion name?
> >> >
> >> > I am attempting to implement an output codec, and I am using other
> >> > codecs for reference (in this case oggvorbis.c).  I am currently
> >> > looking to contact "mru", or anyone familiar with oggvorbis.c.
> >>
> >> That would be me, but I'm afraid I'm no more familiar with oggvorbis.c
> >> than anyone.  I didn't write it, and I'm not the maintainer of it.
> >>
> >> Just say what you need to know, and we'll probably try to tell you.
> >
> > I'm looking from line 104 onwards.  At this part the various
> > ogg_packet structs are being packed together into the AVCodecContext's
> > extradata fields.  There is some magic happening that I don't follow:
> >
> > len = header.bytes + header_comm.bytes +  header_code.bytes;
> > avccontext->extradata_size= 64 + len + len/255;
> > p = avccontext->extradata= av_mallocz(avccontext->extradata_size);
> > p[0] = 2;
> > offset = 1;
> > offset += av_xiphlacing(&p[offset], header.bytes);
> > offset += av_xiphlacing(&p[offset], header_comm.bytes);
> > memcpy(&p[offset], header.packet, header.bytes);
> > offset += header.bytes;
> > memcpy(&p[offset], header_comm.packet, header_comm.bytes);
> > offset += header_comm.bytes;
> > memcpy(&p[offset], header_code.packet, header_code.bytes);
> > offset += header_code.bytes;
> > avccontext->extradata_size = offset;
> > avccontext->extradata= av_realloc(avccontext->extradata,
> > avccontext->extradata_size);
> >
> > Can the "64 + len + len / 255" expression be explained?  I see that
> > the buffer is later reduced in size based on the actual value for
> > offset.  Is this expression simply something which is guessed to be
> > "large enough"?
>
> That's right.
>
> > Can the "p[0] = 2" be explained?
>
> The first byte is followed by 2 xiph-coded values.
>
> > What does av_xiphlacing do?  It has no documentation that I can find.
>
> It has source code.
>
> Do yourself a favor and keep away from the xiph brain damage.  What
> are you trying to accomplish?
>


I am trying to use libtheora inside libavcodec to add theora encoding
to ffmpeg.  I'm making progress again now.  The ffmpeg / theora
mailing lists and irc channels are very helpful. :)


-- 
Paul Richards



More information about the ffmpeg-devel mailing list