Ups! Maybe it's time somebody corrected that statement then - the wiki seems the right place to put the correct algorithm, IMHO.<br><br>Cheers,<br>Silvia.<br><br><div class="gmail_quote">On Feb 13, 2008 10:06 AM, Ralph Giles <<a href="mailto:ralph.giles@artifex.com">ralph.giles@artifex.com</a>> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div style="">
<br><div><div class="Ih2E3d"><div>On 12-Feb-08, at 2:57 PM, Ralph Giles wrote:</div></div><blockquote type="cite"><div><div class="Ih2E3d"><div><br></div><blockquote type="cite"> <a href="http://wiki.xiph.org/index.php/GranulePosAndSeeking" target="_blank">http://wiki.xiph.org/index.php/GranulePosAndSeeking</a> </blockquote>
</div><div class="Ih2E3d"><div>I'm not sure it's helpful to keep linking to this. It's important historically, but Monty's whole thesis about "start decoding here and you're good" is WRONG because of the continued packet issues.</div>
</div></div></blockquote><div><br></div><div>To clarify, the seeking algorithm he describes is inaccurate. This message is an accurate description of the rationale behind the split/granulepos hack for encoding both the timestamp and the back pointer in the granulepos, as is now used by theora, dirac and cmml.</div>
<font color="#888888"><div><br></div><div> -r</div></font></div></div></blockquote></div><br>