For loop linguistic abstraction (usage)

[Filip Konvička]

>  > The compiler is fine....BTW, I think that the A.B:=C notation could 
>  > perhaps be extended to the case that I described above, and everything 
>  > would work just fine.
>  > 
>
> you may have a context where the compiler doesn't know if X is a
> dictionary or a record of cells (or a banana ...!) so you would
> sometimes have to defer to  a runtime test.
>   
I thought that the Value.dotAssign procedure does just that (detect A's 
type at runtime).
> In general I am very wary about adding such magic.  
>   
No problem, that was just a thought..... :-) However, I think that 
there's so much magic about A.B:=C already, that my proposal would maybe 
make it complete. And, while we're talking about magic - why not make Oz 
a bit more magical? :-D

F.