standarisation: UML or XML?

[a.gomez.corona@ibermatica.com (Alberto Gomez Corona)]

Hi.

First, I´m "concerned" about the initial focus of this Forum, about defining standard messages. I want to higlight some remarks, learn from myself from the tremendous success of the Web and the not so well evolvement of apparently better structured information exchange architectures, such is CORBA, COM etc.

 

And here I want introduce the notion of document computing. For me, a document can be seen as the exposition of a object data to anther piece of software. Instead of offering a given interface, a document is an object which shares itself.

 

What advantages such object has in terms of standarisation and evolution (two apparently opposite features)?

 

first, XML, the base format, can be standarised for any information exchange need. and, there are standard procedures to extend , re-define , formally fix and publish to the internationa community since the beginning.

 

If we implement the document visiblity to the external world trough an  object oriented API, we may give away these well proven features, and we may be invloved in re-solve the problem: wether UML must be used or not etc.

 

If we use XML only as another way to serialize binary data, we will drop out the tremendous capablilities of XML as a way to standarize and to make evolvable the information about the status and the bussinees processes of a enterprise.

 

A document describing the product catalog of a company can be used on infinite ways (discovered or undiscovered yet). A Interface to access the product catalog of a company only can be used the way it was designed to be used. I may be very open, but never more open that the document itself.

 

To sumarize, what I propose is the following:

 

In any client-server scenario, to expose the server side of the interaction as a document (XML) . In the client side, to use an standard API (such is XML-DOM or XSLT) and to code the process specific code over the former ifrastructure. That domain specific code will be subject of standarisation after the the supporting data has been standarised. That is, the first priority is the schemas, DTDs and so on of the XML documents to use. The standarisation of the procedure for each individual process can be done by the market.

 

 

Maybe this message can have some utility.