Re: Combining media fragments with other time-clipping methods

On Sat, Nov 29, 2008 at 12:03 AM, Yves Lafon <ylafon@w3.org> wrote:
> On Fri, 28 Nov 2008, Silvia Pfeiffer wrote:
>
>>> When we did the temporal URI spec, we found that the best way to look
>>> at temporal URIs is that they always specify a interval, and never
>>> just a offset point. The only sensible use case for a single offset is
>>> when one is trying to extract a keyframe at such an offset rather than
>>> a media fragment - this could be done with content negotiation, but
>>> may not be something we should consider. So, our assumption was that
>>> the time always specified semi-open intervals: [20s,inf[ for #t=20s,
>>> or [20s,40s[ for #t=20s-40s. I think this makes sense for us, too.
>>>
>>> Yeah! Semi-open intervals rule!! :-)
>>> BTW: I was thought to write those sem-open intervals either as "[20s,
>>> 40s>"
>>> (at school) or "[20s, 40s)" (at university).
>>> Is the "[20s,40s[" a notation I'm not aware of, or a typo?
>>
>> Just the way I learnt them. But I also know [20s,40s). I've never seen
>> [20s,40s> though.
>
> I always used [], [[, ]] and ][, never saw the other notations.
>
> But back to the point, pointing to a position in a complete media file is
> indeed interesting, but not directly related to fragments. But as the syntax
> would be quite similar (as using # makes perfect sense there), we need to
> accomodate that in our syntax.

That begs the question: is an image extracted from a certain offset
point not a fragment? It is indeed a different mime type, but I'd
still call it a fragment.

Cheers,
Silvia.

Received on Saturday, 29 November 2008 01:30:00 UTC