Why do some people not use XSLT to process XML-formatt

Hi Folks,

I am working on a project that has created a large XML Schema. The creators of
the XSD deliberately avoided using xs:choice and several other XSD constructs
(instead of xs:choice, they used xs:sequence containing optional elements).
They did this because: (1) the XML documents -- which conform to the XSD --
will be processed by Java, and (2) the skeletal structure of the Java is to be
auto-generated from the XSD using a data binding tool and the data binding
tool doesn't support xs:choice and the other XSD constructs.

Thus, a limitation in the data binding tool dictated how the XSD was
designed.

If they were using XSLT to process the XML, then the creators of the XSD
wouldn't have been under the limitation, i.e., schema-aware XSLT has no
problem with xs:choice or any other XSD construct.

I don't understand why anyone would not use XSLT to process XML documents. The
XSLT language was explicitly designed for processing XML documents, i.e., XSLT
is a domain specific language (DSL). When a DSL exists, shouldn't it always be
chosen over a general-purpose language such as Java?

The project is in its early stages. There might be an opportunity to convince
people to use XSLT rather than Java. What argument should I make to persuade
them to use XSLT?

SAXON processes XSLT using Java. Why? Why doesn't SAXON use XSLT to process
XSLT? Should I infer that the XSLT language is not powerful enough with
certain types of programming?

/Roger