Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!elroy.jpl.nasa.gov!usc!apple!uokmax!munnari.oz.au!goanna!ok
From: ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe)
Newsgroups: comp.lang.misc
Subject: Re: composition of functions
Keywords: composition, function, first-class object
Message-ID: <4801@goanna.cs.rmit.oz.au>
Date: 20 Feb 91 07:04:05 GMT
References: <1991Feb8.191014.6430@spool.cs.wisc.edu> <1991Feb15.230112.2902@csrd.uiuc.edu>
Organization: Comp Sci, RMIT, Melbourne, Australia
Lines: 27

In article <1991Feb15.230112.2902@csrd.uiuc.edu>, bliss@sp64.csrd.uiuc.edu (Brian Bliss) writes:
> >the question was:  how do I write a function
> 	typedef int i2ifn(int);
> 	i2ifn compose(i2ifn f, i2ifn g) { /* fill this in */ }
> >so that compose(f,g)(x) == f(g(x)).

> you say

> int (*f1)(), (*f2)();

> int retval (x) { (*f1)((*f2)(x));}

> int (*compose (f, g))()
>    int (*f)(), (*g)();
>    { f1 = f; f2 = g; return retval; }

Spare me days!  This completely (almost definitively) misses the
point.  If I do
	fg = compose(f, g);
	hk = compose(h, k);
the later call to compose() should not invalidate the earlier result!
This atrociously buggy pseudo-implementation *guarantees* that
	compose(compose(f,g),h)(x)
will not work, which it would in a language with first-class functions.
 
-- 
Professional programming is paranoid programming