Re: [ubl-dev] UBL and a functional SQL schema

["David Webber \(XML\)" <david@drrw.info>]

Alexi,

Agreed - that's why I mentioned the XML storage type - 
such as Oracle has - which pre-indexes the XML content
for you on INSERT (you pay the piper upfront with 
slower storage times) - but then you can access the
content "normally" using SQL extensions and XPath
expressions to reference the content directly.

Internally its using those mini-indexes at the record
level it created to index into the content directly.
Neat, simple, easy, fast!

Thanks, DW

----- Original Message ----- 
From: "Aleksei Valikov" <valikov@gmx.net>
To: <ubl-dev@lists.oasis-open.org>
Sent: Thursday, February 24, 2005 8:38 AM
Subject: Re: [ubl-dev] UBL and a functional SQL schema


> Hi.
> 
> > This is what we have been espousing as the true benefit of XML for
> > 7+ years now.
> > 
> > If you are using an RDBMS that supports XML storage types then
> > its even more effective.
> > 
> > Why disassemble and reassemble the XML at the atomic level
> > when you do not have to?!?
> 
> The essential question is querying. If you don't need any structural 
> querying, CLOBs are the best choice. If you do need querying, the 
> question is how complex are your queries? Would XML indexing 
> capabilities of your database or search engine like Lucene be enough? 
> Finally, if you assume your queries will be "quite" complex or you can't 
> predict their complexity, you may finally run into trouble with 
> CLOB-like storage.
> 
> Bye.
> /lexi
> 
>