Re: ratio implementation

[Bill Schottstaedt]

Kevin Ryde wrote:

> > I notice that r5rs seems to
> > imply that (inexact->exact .3) should return 3/10 (see the rationalize
> > example which is assuming this)
> >    
>
> Isn't that the effect of the rationalize function, rather than
> inexact->exact?
>  
The example I was referring to was:

(rationalize
 (inexact->exact .3) 1/10)            ==> 1/3    ; exact


which only makes sense if (inexact->exact .3) does not
return 0, but 3/10.  Then rationalize of that with an
error of 1/10 finds the simpler ratio 1/3. In Guile
currently (I'm typing from work, so I can't actually
check this...), the (inexact->exact .3) returns 0,
so rationalize would also return 0.