robert burrell donkin wrote:
how's JiBX's schema generation?
Decent, within the limits of both JiBX bindings and schema. But schema
is such a horrible, convoluted, mess that it's really difficult to try
to establish complete equivalences between it and general Java code. The
frameworks that come closest to full schema support, such as XMLBeans,
do so by building Java code that's specifically designed to match schema
constructs and even then fall short in some areas (substitution groups
in the XMLBeans case, for instance).
On the other hand, most forms of JiBX binding definitions have schema
equivalents, and most forms of schema constructs have JiBX equivalents.
I've already got a sort of active model of the JiBX binding definition
that provides the binding validation support (validates the actual
binding XML structure, makes sure that referenced classes and methods
are available, checks for type conflicts, etc.). I've been working on a
similar representation for schema so that I can improve the code that
goes code+binding -> schema, and also provide better
schema->code+binding support. My expectation is that if I can tie these
together I'll be able to support fast consistency checking that'll allow
highlighting of problems in IDEs and even suggested fixes.
- Dennis
