So I. will never be made tolerant? If so then forget my remark. However, if
there is a good chance that at some time in the future it might, then how
about accepting I.!.0 but any value other than zero for the tolerance be a
domain error for now? Wouldn't hurt anything and it would let people make
sure I. is not tolerant if they need to.

On Wed, Mar 17, 2010 at 9:24 AM, Roger Hui <[email protected]> wrote:

> Tolerance and ordering do not mix well.  Watch this:
>
>   y=: 1 + (2^_44) * 400 ?...@$ 200
>   # ~. y
> 67
>   # ~. y {~ ?~#y
> 70
>   # ~. y {~ ?~#y
> 71
>   # ~. y {~ ?~#y
> 73
>   # ~. y {~ ?~#y
> 70
>   # ~. y {~ ?~#y
> 69
>   # ~. y {~ ?~#y
> 70
>   # ~. y {~ ?~#y
> 70
>   # ~. y {~ ?~#y
> 68
>
>
>
> ----- Original Message -----
> From: Dan Bron <[email protected]>
> Date: Wednesday, March 17, 2010 7:07
> Subject: Re: [Jprogramming] How to use dyad I.
> To: 'Programming forum' <[email protected]>
>
> > RM=Raul Miller, DB=me:
> > RM>  I. becoming tolerant would be a breaking change
> > RM>  in the language ... So I do not consider potential
> > RM>  language change to be a valid design issue.
> >
> > I have less confidence in this inference than you do, especially
> > in the context of tolerance (or other implementation, as opposed to
> > notation, changes) [1].
> >
> > DB>  Nested trains can definitely be leveraged to re-use
> > calculations.
> >
> > RM>  I usually prefer to avoid such approaches
> >
> > To each his own, of course.  And my threshold for "going
> > explicit" is higher than most.
> >
> > -Dan
> >
> > [1]  Though the fact that I.'s intolerance is called out in
> > its very definition is a good sign (OTOH, it could also just be a
> > warning; I suspect if Roger could've easily made I. tolerant, he
> > would have, and I.'s intolerance stems from fundamental
> > difficulties in providing tolerance, or reliance on other
> > intolerance functions).
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to