RE: <boolean> (was Re: object-class)

[Stuart Watt <S.N.K.Watt@open.ac.uk>]

> Thats not the same thing as having a small, concise definition of control
> semantics.  In Smalltalk & Self control semantics are spread throughout
> several relevant classes.

I'd agree. Unfortunately, method dispatch is not complete enough to describe a
good set of control forms. Other forms, such as bind-exit, deal with scoping
and pure O-O approaches get weak here. Continuation-passing is better at this,
and is a concise definition of control. I still think that message-passing best
fits into Dylan as an "invisible cond" in the dispatch system. 

What about a type macro system? <true> could be a synonym for (singleton #t),
but I always feel guilty adding stuff to small languages. Perhaps this can be
integrated with the forthcoming definition of macros so as not to be too big.

Stuart Watt
Human Cognition Research Laboratory, 
The Open University, Milton Keynes, MK7 6AA. UK.
Email: S.N.K.Watt@open.ac.uk; Tel: +44 908 655013; Fax: +44 908 653169