Lists Home |
Date Index |
10/4/2002 4:25:40 PM, Liam Quin wrote:
>I think XML Query will bring a new twist to hypertext in XML, since
>there's a fine line between hard-coded links, runtime queries, and
>even text retrieval and automatically-supplied links.
I think that's an awfully good point that doesn't get much discussion.
(Sean McGrath alluded to it vis a vis "views" the other day).
In the database world, the CODASYL network data model
data model became an ANSI/ISO standard in the 1970's. It was almost
totally supplanted (in mindshare anyway) by the Relational data model
in the 1980's. The key differentiator was that CODASYL ultimately
relied on "pointers" (which I'm pretty sure are functionally much like
hyperlinks, at least by analogy) to relate bits of information that did
not fit into a hierarchy, whereas the Relational model had a) no native
concept of hierarchical relationships and b) relied on joining the
*values* of tuples in different relations to define relationships.
C.J. Date has some quite illuminating rants against "pointers", both
in the old CODASYL sense and in various SQL-isms that go beyond the
pure relational model.
[forgive me if I've botched this ... I don't claim to be an RDBMS geek]
XQuery supports a join operation much like SQL does ... the result set
can be defined on a match between the values of elements/attributes in
one XML tree or collection with those in another. This could allow some
very powerful applications, and ways of thinking, once the spec (ahem)
gets finalized and the "you can do joins with XML!" meme gets propagated.
For example, I find myself (and noticing others) referencing things by
a Google search that has them on the first page rather than by a long
and fragile URL. That pattern applied with XQuery could lead us towards
a situation not unlike Ted Nelson raves about, i.e., where "hyperlinks"
don't break when things move around any more than relational databases
break when they are repartitioned.
I have no idea if these are just half baked wishfulness, but Liam's
point resonated with me! It also illustrates why I'm so immensely
frustrated with the XQuery WG -- this is in the most basic layer of
XQuery (indeed it was in the very first version of
Quilt that Jonathan showed me nearly three years ago). I want
the world to start driving the Model T version of XQuery
rather than having to wait until the Ferrari concept car
they are experimenting with is ready for production. People who
just need to get to work and back can drive the Model T; people who
need to go 200mph can wait for the Ferrari.