Re: false and types

[alms@cambridge.apple.com (Andrew LM Shalit)]

At  2:09 PM 10/22/92 -0700, thant@void.esd.sgi.com wrote:
> (define-class <foo> (<object>)
>   
>   (a type: <integer>
>      init-value: #f))               ;;  Why can you do this...
> ;Value: <foo>

This is an error.  Thomas should not let you do this.

> (define foo (make <foo>))
> ;Value: foo
> 
> (a foo)
> ;Value: ()
> 
> ((setter a) foo 3)
> ;Value: 3
> 
> ((setter a) foo #f)                 ;; ...if you can't do this?
> 
> 
> I figured out that I could do (singleton #f) to create a type for 
> a method that dispatched specifically on false, but that isn't
> a solution for the above.  In C and C++ NILL is kind of an exception
> to the type system, and I thougth #f might play the same role.
> 
> What is the *right* thing to do?

In this case, the right thing to do in Dylan is to not supply an
init-value.  If the slot is accessed before a value is stored in
it, you will get an error.

If you want to supply an init value, it must be of the type specified
for the slot.

I thought the Dylan manual was clear about this, but a quick glance
shows that it doesn't spell out the restriction on init-value specifically.