Norman Walsh writes:
> Is the following observation correct: there's no dispute about
> the semantics for exceptions or interfaces, the only point of
> contention in the current proposal is whether or not
> superclasses need to be called out in some more explicit way
> than order among siblings.
I think that's roughly right, though your asking and our discussion
to this point makes me think that the set of superclasses / interfaces
should be separate from the set of allowed exceptions. This is mostly
a result of wanting to allow "exceptions" which are distinct from
"classes" (which *is* the case for Modula-3, but there is no concept
of exception inheritance: exceptions cannot be derived from other
exceptions) [note: I'm not actually aware of any language which
distinguishes exceptions from classes *and* supports exception
inheritance].
If DocBook is to support such languages, there needs to be a marked
separation between what something is (superclasses & "implements"),
and what exceptions can be raised by constructors. That might be the
way out: the class itself doesn't raise exceptions, only it's
methods.
-Fred
--
Fred L. Drake, Jr. <fdrake at acm.org>
Corporation for National Research Initiatives
