To my understanding, XML is a new way of representing data. It's different from both databases and programming languages, OO and non-OO alike. It's based on documents, documents being images of a section of reality.
After all, modelling the world in XML often takes less tweaking than either relational database modelling or object-oriented modelling do - at least when you want to implement the models in executable software.
Everybody around me agrees that XML is going to be _the_ means of system integration. Where the viewpoints of folks usually diverge is in the very practical question "so how much work should we put into developing yet another XML language - right now".
I think one can't transform data without understanding its meaning. This understanding is best achieved through designing a neutral XML Schema definition that has semantic names of elements and attributes. Mapping this neutral XML to any programming language-specific or database-specific XML's is a hub-and-spokes architecture, while doing ad-hoc transformations between the different, db- or language-specific XML's will lead to spaghetti code.
I usually push for the neutral, semantics-based XML.
Jan Dvorak,
MathAn Praha, Ltd.
(jan.dvorak@mathan.cz)
|