RE: Indexing XML documents

[Falko Braeutigam <falko@softwarebuero.de>]

On Thu, 29 Jun 2000, Ashvil wrote:
> > -----Original Message-----
> > From: Lars Martin [mailto:lars@softwarebuero.de]
> > some weeks ago we 'discussed' this topic here in the SMB
> > headquarter but didn't
> > found a solution except of starting a new ozone sub project
> > to provide such
> > wanted functionality.
> 
> I think this may be possible with a middleware that sits between XT
> and Ozone-DB, if I understood your suggestion correctly.
> 
> > There are no _free_ third party products out there which
> > provide indexing of
> > XML data. Or did I miss something?
> 
> No. I could not find anything out there.
> 
> >
> > The problem I see is to combine the third party XPath
> > product (like XT) with the
> > generated index stored in ozone. So you have to extend or
> > at least to patch the
> > query engine to react on queries where an index exists. The
> > best way would be
> > a proposal (what about an W3C Draft written by ozone users
> > ;-) for an indexing
> > interface which can be plugged into any XPath engine.
> 
> Hey, I like the idea of a W3C draft type document for an indexing
> interface. I would like to contribute to that document from a
> requirements point of view. Any other volunteers. I am sure that other
> people in the XML community would also be glad to contribute.
> 
> >From a requirements point of view, If more data is to be stored in XML
> databases, then I want to make sure that we can use indexing and
> caching techniques developed by relation databases are used.
> 
> Any other folks out there share this opinion and want to volunteer to
> work on a draft proposal for indexing (and maybe caching) interface
> for any XPath engine. Please raise your hands ;-)
> 
> 
> The ASCII art diagram for this would be (If I understood Lars
> correctly)
> 
> XPath engine (XT) <---> Indexing Interface <---> Ozone-DB.

Yes, this would be great. Then we could provide a ozone optimized
implementation of the indexing *and* the XPath engine could stay independent
of the underlying DOM/Indexing implementations.

A standard to index DOM is the only way to keep the XPath engine independent
of indexing. Otherwise indexing must be incorporated into the XPath engine
directly which is not a good idea.

On the other hand I'm not sure how much indexing would speed up XPath queries.
A XPath contains path navigation and simple searches. For path navigation the
DTD can be used for optimization. For simple searches indexes are need.


Falko
-- 
______________________________________________________________________
Falko Braeutigam                         mailto:falko@softwarebuero.de
softwarebuero m&b (SMB)                    http://www.softwarebuero.de