Re: Chaining and mixing of supertype-subtype and type-instance relationships

On Sun, Oct 02, 2011 at 02:38:47PM +0200, Johannes Schmidt wrote:

> E sub D isa C sub B sub A
>
> 1) D is an instance of B and A (due to transitivity in "C sub B sub A")
> 2) E is an instance of C, B, and A (due to transitivity in "E sub D" and  
> "C sub B sub A")
>
> I found this in the latest TMRM draft in 5b) [1]:
>
> "[...] Additionally, whenever a proxy a is an instance of another c,  
> then a is an instance of any superclass of c: if x isa,m c and c sub,m  
> c' , then x isa,m c' is true."
>
> This supports 1), however I did not find any notes in the standards  
> supporting 2).
>
> What do you think?

That TMRM tries to follow the "intuitive" meaning (whatever that is)
of subclassing and instantiation? That E should be an instance of C
is in no way intuitive to me. Do you have a real-life example?

\rho