Lists Home |
Date Index |
Not unrelated, what does anyone think of using XSLT to specify a
transformation between two XML structures?
Thanks, Bill for that question. I was starting to get a bit tired of the
current Byzantine fight :-) (In my imagination XML-Dev resemble a lot to a
kind of parliament, a lot of talk, very little action :-)
If I compare it to DSSSL it is less explicative about the transformations
applied but a lot easier to use. However, with DSSSL I know the underlying
model (a grove) and with XSLT I don't. This is probably why DSSSL is more
explicative since it provides a set of operators on a grove (i.e. a data
model). In day to day cases, I tend to find XSLT easier to deal with since a
lot of things are hidden. Anyway this is the case also with other languages
like for instance C++ or Java. When I do a "new" operation, I am not
necessarily aware that objects are part of a collection (i.e. a heap), an
instantiation model; objects are created in some kind of ether.
Recently dealing with some research subjects about prototype based and class
based programming I encoded data model in XML and used XSLT to express
strategy patterns. Thus, I used an XML language to encode object models and
XSLT to specify a strategy (i.e. an action performed on these objects), I
also used XSLT to encode object model transformation into an other object
model. In most cases, people (my experimenters) got an adequate overall view
of the two models and the required transformation to map one into another
when I used a graphical notation like an extended UML, but got more
difficulties to get that from XSLT templates.
Conclusion: XSLT is quite poor to convey information about model
transformation but is quite efficient to perform transformation on XML
structures. Said, differently, XSLT convey very little information but is
tremendously useful in action.
Didier PH Martin