Sebastian,

I hope you get a good reply.  I am in a similar situation.  The XML files I 
need
to process are very complex also.  I don't care about formating to text, I 
need to
move the data to Access tables for more processing, and then returned to XLM
format.  There would be many tables and some large tables.  

I have not seen any XLM examples that come close to the size and complexity
I have.


-- 
hgeron


"chairman@s..." wrote:

> I have a question concerning the organization of highly complex (xml-)
> data.
> The data basically shall be used to control a workflow engine but at
> the same moment shall hold data required for other, general purposes.
> The question now is how this data shall be organized. Requirements are:
> 
> * Multible user access (both read and write)
> * Largely normalization (not to reduce amount of data but to reduce
> administration efforts)
> * Easy to Access (SQL, XPath, ..)
> * Output in XML Format
> 
> >From this list of requirements some design decisions are relatively
> obvious:
> A database is required that enables multible user access for read and
> write transactions (including locking mechanisms and all that fine
> stuff).
> The question of the database type though is not that ovbious:
> (I realize that asking the following question will attract a whole
> bunch of people that keep saying "Damn, why are people to stupid to
> realize that XML is no database but only a language to structure data."
> over and over again thus not helping anybody at all. To you guys I can
> only say: look around what XML is capable of. It has grown far beyond
> simple document structuring.)
> What kind of database shall be used?
> 
> * a XML enabled database (a relational Database that can store XML
> data)
> * a XML database (a database that natively stores XML)
> * a relational database with an XML Mapper
> * a object oriented database with an XML Mapper
> 
> I'd like to describe the data structure further to give you a better
> insight in what I exactly would like to do:
> 
> The "core" data can be mapped relatively simple to a relational data
> model, yet there are certain elements that are rather complex (in terms
> of storing in a relation and in terms of converting to XML) for example
> a binary tree like:
> 
> <node>
> 	<node>
> 		<leaf/>
> 		<node>
> 			<leaf/>
> 			<leaf/>
> 		</node>
> 	</node>
> </node>
> Of course it is possible to store this as a relation but you would have
> to go through a quite complex conversion to get this back into an XML
> format.
> Also another point is to be considered (when deciding to store the data
> in XML format):
> There may be about 2000 of such binary trees requried in the database
> but there would only be about 20 that significantly differ from the
> others, most of them are equal except from one leaf. So I would like to
> store the 20 different trees and somehow link the differing leafs where
> required. I don't think this is possible in XML via XPoint or XLink due
> to XMLs document oriented principles...
> 
> As you might realize I'm just poking arround in this problem and have
> not yet come arround to test or implement many existing solutions I
> just would like a pointer in the right direction or any comments about
> technologies that you had good/bad experiences with.
> 
> Any comments on this would be greatly appreciated.
> Thanks in Advance
> 
> 
> Sebastian Schaetz
> 
>