At 06:04 PM 1/26/00 -0500, you wrote:
>David Forslund wrote:
>
> >
> > I would recommend using tools where they are designed. That means using
> > UML and XML where appropriate. XML is a representation of data not a
> > design tool. UML can be expressed in XML (using XMI) if you like.
> >
>
>
>It seems to me that the very meta-data aspect of XML that is used to
>tell the XML parser what the XML document is all about is, in fact, a
>modeling function. Models, if there are more than one of them, have to
>be able to describe themselves (I think), and there are always going to
>be a myriad of models in medicine. The entire literature on practice
>variations tends to support this. So how do you handle multiple
>models? XML looks good for simple modeling that can describe itself and
>can easily be translated into an object database. Or am I completely
>off the mark?
It could be used to express models and that is what XMI does by expressing
the UML elements in XML. But writing the XML directly is rather painful
compared to using a UML modeling tool to express the model and then using
XML to send it around. UML can also talk to a database through the Meta
Object Facility. One can move continuously between UML, XML and the MOF as
needed depending on the way the model needs to be used. So your idea is
fine and is being implemented by software vendors in a standard way.
Dave
>John
