Re: [stack] stackless concatenativity + visual dataflow

On May 20, 2008, at 7:52 AM, Chris Double wrote:

> Based on the things I've written in Factor a 'fixed arity' restriction
> would have little to no effect on what I've written.

Very good to hear! I was half-expecting it to be shot down for being  
overly restrictive.

> It did in the past when 'curry' was discouraged and quotations passed
> to 'map' and the like would often access items deeper in the stack...
>
>  5 { 1 2 3 } [ over + ] map
>  5 { 1 2 3 } [ + ] map-with

The version using map-with is certainly clearer. Perhaps there really  
is a readability benefit in gently forcing the programmer to use  
combinators with fixed stack effects.

> Would your restriction still allow curry, compose, etc?

Yes, and you wouldn't need different versions of 'compose' for each  
possible arity combination of the functions being composed. In other  
words, '[0] [+] compose' and '[swap] [over] compose' would both be  
fine, even though the arities of the input quotations and the arities  
of the resulting composed quotations do not match at all.

- John