More precisely, NaN denotes many bit patterns in the IEEE standard.
J displays them all as _. and 128!:5 is 1 for each of them.
On input into J, _. is converted into a single one of such patterns.
Thus:
hex =: ([EMAIL PROTECTED]) }. 2&(3!:3)
unhex=: 3!:2 @ ((}: 2 (3!:3) 0.5) , ]) " 1
H=: '0123456789abcdef'
hex _.
fff8000000000000
] t=: 'fff8',"1 H{~7 12 [EMAIL PROTECTED] 16
fff87acee515487a
fff80d41773eb545
fff8006d81d67daf
fff869b83bf61c7a
fff87ac325f456a2
fff8b2444f99e537
fff81e712ef5847e
unhex t
_. _. _. _. _. _. _.
t -: hex unhex t
1
128!:5 unhex t
1 1 1 1 1 1 1
----- Original Message -----
From: Don Guinn <[EMAIL PROTECTED]>
Date: Friday, February 29, 2008 4:23
Subject: Re: [Jbeta] Use of the name 'NaN' deprecated
To: Beta forum <[email protected]>
> There are many other values in the IEEE standard which represent invalid
> numbers other than NaN. How does 128!:5 treat them? Does it treat them all
> as NaN or as numbers?
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
