Re: fragment identifiers

At 10:07 AM 2002-07-23, Tim Bray wrote:

>Roy T. Fielding wrote:
>
>[Good stuff, Roy]
>
>>  In HTML, a fragment
>>identifies a portion of the complete virtual document, not
>>just the bits within the HTML framework.
>
>I'm not sure about that - in *valid* HTML, in principle, a fragment could refer to a "portion" of the data; in real life a fragment seems effectively an arrow pointing at a location within the document... I've never seen a user agent that would process foo.html#bar by selecting out the element that had name="bar" or id="bar", only by maneuvering to where that element starts.

It is true that headers and #fragments in HTML are safer to interpret as start-points than as sub-range references.  

But a start-point is a Dedekind cut of the stream.  It suffices to identify a 
tail, a terminal sub-stream, which fits the general model category of a 
sub-object.

Al

For an object class which has both stream and tree properties well developed,
consider the ANSI/NISO Z39.86 digital talking book.

 http://www.loc.gov/nls/niso/

Sorry for the arcana, but this matters for the contemplated Timed Text media 
class where instances may either be segments or be emergent streams.

 http://www.w3.org/AudioVideo/timetext.html


>Not that this changes anything in the underlying argument. -Tim

Received on Tuesday, 23 July 2002 11:57:06 UTC