Re: storing xml files into database

[Bill Lindsey <bill@blnz.com>]

Hi Chris,

Yes, B-Bop's Xfinity server can (and does) evaluate queries
across large document collections as a single operation.
The query performance cost of increasing the collection size
is typically on the order of O(log n).    

We do node level indexing. Attribute values and 
PCDATA text is indexed, with the limitation Chris Maden 
mentioned. Our currently shipping version, does not
include a full-text, inverted word index.
  
For the kind of application with a single, very stable, data 
oriented schema, the "table-per-element-type" approach might 
get you faster query results.

If you're not running queries across documents or assembling
fragments from many documents, a "BLOB" approach will give
you better space utilization.

I'm comfortable with our technology's ability to store, index,
query and faithfully reconstruct any well formed documents 
(or document fragments).

Bill Lindsey
B-Bop

Chris Parkerson wrote:
> 
> Chris:
> 
> I'm curious, is B-Bop able to do cross-document XPath queries?  I'm
> curious if they've got that working.  This is what most of our customers
> want to do and they want to do it over thousands of documents.  Do they
> do node level indexing, especially word indexing (for word searches) on
> individual text nodes?
> 
> If someone (or some company) actually gets the XML data model to work
> really well for any arbitrary XML document on top of the relational data
> model, I will be very impressed (and will buy stock in whoever does it
> ;->).  What I've found is that such a model works for a subset of XML
> documents, but tends to fail on other sets.  Our goal is to work well
> with ANY XML document, as long as it is well-formed.  And so far, we've
> been able to do that pretty darn well ;->