Lists Home |
Date Index |
- From: Mark Birbeck <Mark.Birbeck@iedigital.net>
- To: firstname.lastname@example.org
- Date: Thu, 21 Oct 1999 18:48:45 +0100
Steve Muench wrote:
> *Mark's* point about SQL is totally valid
> if what that means is that end-users
> would have to type-in SQL statements as
> their "user experience".
My point is not about users typing in queries!!
What if I write an application on my server that queries your server
every day for new product information which is then stored in my
database for local retrieval. Everyone has agreed that you should give
me that data as XML (well everyone on this list, at least). But if you
accept the query in some XML form as well, then you can do whatever you
want with your data storage and I don't need to change my code.
Your approach is fine for 'local' data - where you are very close to it,
and have a lot of control over it. But the future will surely bring
increasingly complex aggregations of data, where servers are talking to
other servers before returning results sets. For example, how would I
phrase a query to my server that returns results from two other
databases - one relational, one object - especially if I don't control
those other servers? With my approach, my server is like a large XML
document and the two other servers are simply nodes within my document -
and the lot can be queried with the same syntax.
I've tended to use the term 'XML Server' to describe a server that can
do this - XML in and XML out. Whether the XML is then stored as XML or
in a relational or object store is only relevant to the server owner. To
the outside world the are just dealing with a very large XML document. I
wouldn't take that as knocking your RDBMS, Steve. I'm simply saying that
people should put a layer over your product that hides the fact that it
is an RDBMS. To do this they will need to convert the incoming queries
to match the structure they have used.
> *Your* point about not querying XML is totally
> valid since the database is going to do
> that querying and sorting much faster over
> non-trivial amounts of data than
> doing it in an XML document.
I'm also saying you do the query in the 'native' format - that's
obvious. My point is that you arrive at the native query by converting a
more abstract, node query - XPath, XQL or whatever.
xml-dev: A list for W3C XML Developers. To post, mailto:email@example.com
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ and on CD-ROM/ISBN 981-02-3594-1
To unsubscribe, mailto:firstname.lastname@example.org the following message;
To subscribe to the digests, mailto:email@example.com the following message;
List coordinator, Henry Rzepa (mailto:firstname.lastname@example.org)