Re: Scalars & atomic values & variables

Joe "Nuke Me Xemu" Foster <joe_at_bftsi0.UUCP> schrieb:
> "Bob Badour" <bbadour_at_golden.net> wrote in message <news:eICdnT6fXqHX2Gqi4p2dnA@golden.net>...
>> "Adrian Kubala" <adrian_at_sixfingeredman.net> wrote:
>> > The definition I proposed elsewhere (for types the values of which are
>> > atomic), is that subtypes are non-atomic. What constitutes a subtype
>> > depends on the type system, I guess. Using the "set of values and
>> > associated functions" definition, my original guess is that A is a
>> > subtype of B if its values and functions are supersets of those of B.
>>
>> Nope. A is a subtype of B if its values are a subset and its operations are
>> a superset. A is a proper subtype of B if its values are a proper subset and
>> its operations are a proper superset.
>
> And then there's Liskov's notion of subtypes:
>
> URL:http://en2.wikipedia.org/wiki/Liskov_substitution_principle
>
> Perhaps the notions of "subtype" and "derived type" should be distinct.

Yes, indeed this is exactly what I was thinking of, though I was trying to frame it in terms of sets of values and functions. I propose that it generally matches our intuition to say that subtypes (according to this definition) are non-atomic. Received on Fri Jan 09 2004 - 19:52:16 CET