Luca Morandini wrote:
Nevertheless, it is easier to build a tool around a declarative
language expressed as XML, than a procedural language expressed as... a
procedural programming language.
I'm sorry, Luca, but I think that's BS.
A complexity of a tool is given by the degrees of freedom, not by the
flavor/syntax/style that you use to encode the information that your
tool generates.
Sure, XML removes the syntax and validation stages that you get for
free. Then RDF and OWL might even give you some implicit reasoning
abilities. And then an RDF rule language might introduce some
prolog/datalog-like constructs.
And if you think writing a tool for this stuff is easier just because
it's more "declarative" (which more and more reads to me as "the
procedures are implicit"), think again, you are thinking about parser
reuse and *that* is probably 1% of your tool cost anyway.
For example, do you think that if the java classes were expressed as XML
statements that *declarative* describe their methods and variables and
inner classes it would be easier to write a tool like Eclipse?
--
Stefano.
