Re: [Digir-protocol] XPath/XQuery

[Guillaume R. <ro...@cc...>]

Le Mardi 29 Octobre 2002 21:12, Hobern, Donald a =E9crit :
> In the meetings in Brazil Guillaume suggested that we might standardise a=
ll
> our data storage as XML documents and then rely on XPath pattern matching.
> (If this is a misrepresentation, please forgive me.)  I argued that this
> might be undesirable in that it may be difficult to turn arbitrary XPath
> patterns into SQL statements for arbitrary database schemas and that this
> might therefore place an unnecessary technology constraint on provider
> implementers.
Actually, i only suggested to standardise external view of our data storage=
 as=20
virtual XML documents, whetever technology is used behind the scene, so as =
to=20
express our queries following a consensus model, the federative schema.

The question of how to translate an incoming xpath query into the most=20
efficient native-format query is just a technical issue. Anyone able to=20
produce the global DOM tree (or global SAX events set) from its dataset is=
=20
able to run the xpath statement on it anyway.

> However from what I have read of XQuery, it seems to be aiming to become a
> plausible standard for addressing our problem and I need to understand it
> further.  We certainly do not want to end up with a totally bespoke XML
> language for defining queries if a widely accepted standard for doing the
> same thing is likely to appear in the near future.
A rapid web research shows that XQuery is official W3C effort toward XML qu=
ery=20
language, and it is going to be largely incorporated in XPath 2.=20

But there are also other query languages developped elsewhere:
=2D XQL (http://www.ibiblio.org/xql)
=2D QUILT (http://www.almaden.ibm.com/cs/people/chamberlin/quilt.html)
=2D LOREL (http://www-db.stanford.edu/lore)
=2D XML-QL (http://www.w3.org/TR/1998/NOTE-xml-ql-19980819)

I guess however that XQuery is the safer bet.
=2D-=20
Guillaume Rousse <ro...@cc...>
GPG key http://lis.snv.jussieu.fr/~rousse/gpgkey.html