[ogg-dev] Zero granule pos

[Conrad Parker]

2008/7/25 Ralph Giles <giles at xiph.org>:
> On 24-Jul-08, at 6:24 AM, ogg.k.ogg.k at googlemail.com wrote:
>
>> I've seen several implementations of Ogg demuxing that use a zero
>> granulepos to detect headers. However, I do not recall seeing this
>> in the Ogg docs - is this an abuse that happens to work because
>> Vorbis is timed by end granule, or is it a proper way to check ?
>
> We've tried to make it "best practice" that ogg mappings be designed
> so that this is possible, but you're correct that it's not required
> by the spec.

like, for CMML we alsos allow a zero granulepos to indicate a clip
that starts at the beginning of the stream; perhaps a similar thing
might make sense with Kate.

Looking for zero granulepos is not a reliable way to determine which
packets are headers, which is why there's a "Number of header packets"
field in  skeleton fisbone.

K.

K.