[theora] Oggz 0.9.8 Released
dale at ucsc.edu
Wed Jul 16 02:31:44 PDT 2008
at least with the VLC playhead issue should be solved by
giving content length in response header. (if you wget the
file and then host it with apache sending out
content-length it has a playhead and is http seekable :)
Giving content length would be beneficial for other
reasons too like generating %downloaded and be more
friendly to caching/proxy systems ... if just cutting a
ogg file this should theoretically not be too hard to
calculate before sending out the stream no?
On Wed, 16 Jul 2008 15:58:03 +0900
"Conrad Parker" <conrad at metadecks.org> wrote:
> 2008/7/12 aphid <aphid at ucsc.edu>:
>> Conrad - thanks for the great work on this.
>> Here are some observations with release oggzchop, using
>> as test stream.
>> VLC: when playing the stream over the net, VLC's
>>playhead does not move
>> and the time index stays at 0:00:00.
>> However, when the above is saved locally (I used curl
>>-O), it behaves
>> ~normally. The start time even begins at 15:22, even
>>though it's the
>> front of the file (which is pretty cool :D) and the
>>playhead moves as
>> Quicktime via xiphQT (on intel mac): Audio and video are
>>not in sync -
>> both when streaming the file and playing it locally. If
>>I play the file
>> from the beginning (as opposed to 15m22s in) things are
>>in sync so I'm
>> assuming something is happening at a keyframe...
> thanks for testing these out. Perhaps we should make a
>wiki page to
> track which players etc. can deal with Skeleton (ie.
> properly, along the lines of:
> ... which will of course involve documenting what needs
>to be done to
> deal with them properly, which is probably atm only
>documented in the
> annodex internet-drafts.
> It's partly documented in:
> ie. it's mostly a matter of players handling the
>presentation and base
> times properly.
> theora mailing list
> theora at xiph.org
More information about the theora