[Prev][Next][Index][Thread]

Re: function signature type checking

Bruce Hoult <bruce@hoult.org> writes:
> In article <usnayu589.fsf@telus.net>, Ray Blaak <blaak@telus.net> 
> wrote:
> > What I was explicitly trying to do was to split the notion of a
> > callable function from a signature specification into separate types.
> 
> The result of "limited()" isn't a function.  It's a type.  Like "array 
> of integers".

Yes, I know. I was trying to play with the notion of separate functional types
versus signature types, analogous to how classes vs interfaces differ in Java,
or function pointers vs functions differ in Ada. 

I think I will lay low about this for a while until I can understand the
issues better in terms of Dylan.

Given that a Dylan compiler necessarily has the ability to determine signature
matching when doing dispatching, and to raise errors when incorrectly invoking
a method, I would be very interested to know what some of the difficulties
are.

-- 
Cheers,                                        The Rhythm is around me,
                                               The Rhythm has control.
Ray Blaak                                      The Rhythm is inside me,
blaak@telus.net                                The Rhythm has my soul.
Follow-Ups: References: