[
Lists Home |
Date Index |
Thread Index
]
David Carlisle wrote:
>I have yet to see this done for any seriously complicated transform.
>If the transformation is sufficiently simple then you could do this
>without the type system, but are you really expecting to be able to
>typecheck a function mapping (say) docbook to XHTML and guarantee that
>given a valid docbook instance a valid XHTML instance will come out?
>
>This may be just about possible with the XQuery style, but is highly
>dubious with the recursive template style in XSLT. You really don't have
>a good enough handle on when the templates will fire in order to
>typecheck the whole transform (as opposed to individual component
>functions).
>
Are you talking about validating an ad-hoc transform or letting the user
select one from a generated set of possible valid transforms?
My experience generating XSLT type taggers from XSDL suggests that using
XSLT modes to represent FSM states can make this area rather more
tractable than I, for one, had expected.
Francis.
|
