Re: Including 'fragment identifier semantics' in MIME media type registration?

> I have some questions about fragment identifiers.

> When the media type is determined by negotiation, how can users 
> specify fragment identifiers in advance?  I'm happy to see 
> that the TAG is already considering this issue.

When the media type is determined by negotiation, any fragment
identifier specified in advance should be one that is usable
with all of the possible media types that might result.

> If fragment identifiers are interpreted by clients, why should 
> they be standardized as part of URIs? To me, they are not part 
> of URIs or protocols, but they are instructions to clients.  In 
> other words, why different applications for the same media type 
> have to agree on fragment identifiers?

Suppose there were a new media type for 'whizzymovie' with two
players, but player XYZ thought '#part=1-3' meant the first three
seconds, while player ABC thought '#part=1-3' meant the first
three minutes.  I couldn't send you something with a link to 
   http://example.org/event.whizzymovie#part=1-3 
and know what it was you would see -- would it be
the first 3 minutes or the first 3 hours?

Suppose I also had a version in the 'oldmovie' media type and offered
'whizzymovie' and 'oldmovie' by content negotiation.
I couldn't send you a link
    http://example.org/event#part=1-3

unless the definition of the fragment identifiers for
'video/whizzymovie' and 'video/oldmovie' agreed.

So it seems like a good idea to try design fragment
identifiers that can be used consistently across
different media types. It's also hard to do.

Larry
-- 
http://larry.masinter.net

Received on Sunday, 8 September 2002 20:34:50 UTC