Morris Kwan wrote:
The runtime method resolving solution also has a benefit that a parameter can
bind to different types at runtime. If the users value ease of use and
extensibility as more important, this might be a better solution.
Yes, I agree. In the case of Java types, however, different runtime
types may still be allowed via subtyping/subsumption. I.e., a user may
declare the type of a parameter to be a base type of the possible
runtime types (the current type system may need to be tweaked for this,
but it is doable I think). The only caveat is that a Java method that
takes such an XSLT parameter would have to be declared using that base
type for the static binding algorithm to succeed.
-- Santiago
