You will find the best explanation in JfC
http://www.jsoftware.com/help/jforc/odds_and_ends.htm#_Toc191734562
especially the reason: integer calculation should not be slow downed by
comparison tolerance.
Regards, Markus
----- Original Message ---------------
Subject: [Jgeneral] Tolerance's upper limit
From: "Sherlock, Ric" <[EMAIL PROTECTED]>
Date: Fri, 1 Aug 2008 12:38:37 +1200
To: General forum <[email protected]>
>After reading the dictionary definition of tolerant equal
>(http://www.jsoftware.com/help/dictionary/d000.htm) and the Comparatives page
>(http://www.jsoftware.com/help/dictionary/dictd.htm) which suggested "An
>arbitrary tolerance t can be specified by using the fit conjunction (!.)", I
>was confused while trying to experiment with it, when the following gave an
>error:
> 1 =!.0.1 ] 0.899 0.9 1 1.1 1.12
>|domain error
>| 1 =!.0.1]0.899 0.9 1 1.1 1.12
>
> eq=: =!.0.1
>|domain error
>| eq=: =!.0.1
>
>It seemed to me that I was missing something simple (or that there was a bug
>-which I doubted!). I was about to post to the forum for clarification when I
>found an earlier forum poster had encountered the same problem, that is, that
>the upper limit for t is 2^_34 .
>
>After following a link in a forum message to the Tolerant Comparison essay on
>the wiki and a link from there back to the dictionary (Appendix C), I see that
>the dictionary does indeed document this limit. However I respectfully suggest
>it would be worthwhile mentioning the limit (or linking to Appendix C) on
>either or both of the dictionary/d000.htm and dictionary/dictd.htm pages as
>well.
>----------------------------------------------------------------------
>For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
