At 03:05 PM 2/1/2002 -0500, Gavin Thomas Nicol wrote:
>On Friday 01 February 2002 02:08 pm, Jonathan Robie wrote:
> > I think XPointer is useful, but it should not own the fragment
> > identifier. That should be reserved for something *much* simpler.
>
>Perhaps that is the *real* issue? We keep hearing about complexity,
>but implementation experience says that XPointer is only incrementally
>worse than XPath (which for many things is way overkill).
Right. I am not saying that XPointer is too complex for a mortal to
implement, I am just saying that it is way too complex to be the only
standard way to point into an XML document. For people who write efficient
server-side software, a pointer must be very simple, not a moderately sized
query language.
Full-featured client-side hypertext systems have very different needs, and
these are the needs directly addressed by XPointer today.
The strongest architectural need is for simple pointers, but there are also
people who need complex hypertext systems. As long as we don't conflate the
two, we can have both.
Jonathan
