Is XPATH the way to search a data model?

[Gerard Lemson]

>
>
>
> Brian Thomas wrote:
>
> > 	Hi David,
> >
> > On Monday 17 May 2004 04:35 pm, David Berry wrote:
> >
>
> snip
> >
> >
> >>My question
> >>is, should we be optimising our data models specifically so
> that they can
> >>be searched using XPATH?
> >
> >
> > 	"Yes" (if you mean XQuery + XPath).
> >
> >
> I feel this very strange; optimization could perhaps influence
> implementation,
> but not data models (according to one previous mail of G.Lemson
> if I understood
> it well enough). This is a too short feedback loop and will
> "corrupt" data model
> with ad-hoc restrictive solutions devoted to only one implementation.
> Just my feeling,
> Pierre
>
>
Indeed I think Pierre understood my email correctly.
I think it is not the datamodel that should be optimized for XQuery+XPath
(+XSLT ?)
but (a) possible binding(s) of the data model to an XML schema.
Similar we might want to optimize a relational binding for SQL, and a
Java binding for a particular Java implementation etc.
As I argue in my reply to Francois request (in
http://www.ivoa.net/forum/dm/0454.htm)
I believe in fact that there are probably too many possible specialized
applications
for a data model and consequently different useful optimizations one may
want to
make than can be handled by a single model (binding). I believe therefore
that it is not even
in a presumed standard reference binding that one should think about
optimizations.
Instead I believe we need a way to create derived data models (which usualy
go by the name of "views") based on the reference binding, and it is these
views that are optimized for the
particular goal for which they are created. Luckily XML has in XSLT and
possibly XQuery a way to
do so explicitly, just as relational databases offer the possibility to
define explicit, derived
VIEW-s.
If people are interested I may discuss this framework during the Boston
meeting.

Bye

Gerard