Richard E Maine wrote:
> of course it is trivially possible for default
> initialized components to get undefined.
Of course, components can. But most of the cases (events) where a whole
object of a derived type becomes undefined do not apply when default
initialization is specified. For example, INTENT(OUT) dummies. Maybe not
all. I recall it was you that was making a point along these lines when
we had a J3 interp request for assumed-size dummies and default
initialization. I suggested we simply forsake default initialization
(for laziness) but you were making the point that such a thing must not
happen since those objects would start in a bad (undefined) state, which
is against the purpose of default initialization.. But I probably did
not quite understand what you were saying, or said it wrong :-)
> Thus, the fact that a type does or does not have default initialization
> is not inherently a bar to interoperability.
Good, no interps!
Thanks,
Aleks
|