[Haskell-cafe] Re: Knot tying vs monads

[apfelmus]

ChrisK wrote:
>  The data dependency is circular.

Yes and no. The input and outputs pairs are dependent on each other, but 
the integer doesn't depend on the string. Thus, I'm pretty sure that

   (Int, String) -> (Int, String)

can be refactored into

   Int -> (Int, String -> String)

This is related to attribute grammars, I finally found the reference

   Designing and Implementing Combinator Languages.
   S. Doaitse Swierstra, Pablo R. Azero Alcocer, João Saraiva
   http://people.cs.uu.nl/doaitse/Papers/1999/AFP3.pdf

I'd even add  after  to the result of the functions in order to avoid 
the O(n^2) degenerate case.


In any case, I prefer Wadler's combinators. With  line  being more rigid 
than  Brk ,  nest  and  group  basically factor the monolithic  Blk 
which makes more laws and available and hence gives a more elegant 
implementation.


Regards,
apfelmus