Re: More finalize woes

[Per Bothner]

tephen Crawley wrote:
> It is NOT a logical extension of what I'm saying. I'm saying that
> non-trivial finalizers should always include a call to super.finalize().

In "contract-oriented" programming more generally, every method
that overrides another should super-call the latter.  (The
langauges Simula67 (the first object-orineted language) and BETA
enforce this by turning this around:  Instead of a overriding
method calling super, the base methiod calls INNER to call the
overriding method.)

> It does NOT follow that every class should have a finalizer.

Ok.

> Adding a trivial finalizer does NOT help in the event that the
> programmer changes the classes position in the class hierarchy.
> But having a (previously redundant) super.finalize() call in a
> non-trivial finalize method may turn out to be a GOOD THING.

True.  I wasn't thinking clearly.

I still don't think it is a good idea reason to call super,finalize
if super is Object - and I don't think we should warn in that case.
--
        --Per Bothner
address@hidden   http://www.bothner.com/per/