No subject


Mon Nov 3 13:03:41 PST 2008


info.c, dsp.c, etc is that virtually everything is currently dependent on a
set of ogg_references (with ogg_buffers) being present in ogg_stream_state
to move around the pages of the stream. Is this actually the case?

I'm still trying to get a good impression of how all the framing work is
connected to the stream buffers and how we could best move around the stream
without buffering (and without disrupting the sync and stream states).

Thoughts?
-----Nate

------=_Part_102632_19819309.1227145399214
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Monty et al,<br>In reference to the discussion here:<br><a href="http://lists.xiph.org/pipermail/tremor/2008-October/001486.html" target="_blank">http://lists.xiph.org/pipermail/tremor/2008-October/001486.html</a><br><br>And the problem being tracked here:<br>

<a href="http://bugs.slimdevices.com/show_bug.cgi?id=7857" target="_blank">http://bugs.slimdevices.com/show_bug.cgi?id=7857</a><br><br>I&#39;ve begun to research methods for the &#39;speculative buffering&#39; of streams that Monty was talking about in late October. Specifically right now I&#39;m focused on implementing a method to eschew buffering the comment header, however I&#39;m trying to make sure that anything I do can be easily implemented in the future for other bitstream operations. <br>

<br>So far my plan has been to implement an array of &quot;buffer control switches&quot; on the OggVorbis_File struct. It may make sense to have a few statically-defined tokens (ex: OGG_NOBUF_COMMENTS) that can be used (the full ideal implementation, from what I understand of Monty&#39;s previous comments, should be able to block buffering on only certain streams or portions of a stream inside any given vorbis file). <br>
<br>From what I&#39;ve (superficially) noted from browsing vorbisfile.c, framing.c, info.c, dsp.c, etc is that virtually everything is currently dependent on a set of ogg_references (with ogg_buffers) being present in ogg_stream_state to move around the pages of the stream. Is this actually the case? <br>
<br>I&#39;m still trying to get a good impression of how all the framing work is connected to the stream buffers and how we could best move around the stream without buffering (and without disrupting the sync and stream states).<br>
<br>Thoughts?<br>-----Nate<br>

------=_Part_102632_19819309.1227145399214--


More information about the Tremor mailing list