Yes, comparison with 0 is always exact; that is a consequence of the definition of tolerance.
The right domain of 5&p: is integers. Thus: 5&p: +/1000#0.01 ok because there is an integer n such that n=+/1000#0.01 5&p: 10-10 ok because there is an integer n such that n=10-10 5&p: -~+/1000#0.01 ok because there is an integer n such that n=-~+/1000#0.01 5&p: 10-+/1000#0.01 not ok because there is no integer n such that n=10-+/1000#0.01 5&p: 2^_44 not ok because there is no integer n such that n=2^_44 = above is tolerant equality. ----- Original Message ----- From: "bill lam" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Wednesday, April 26, 2006 9:46 PM Subject: [Jprogramming] 5&p: 0 I remembered doc said comparison with zero is always exact. So that is the following behavior a design feature? 5&p: +/1000#0.01 4 5&p: 10-10 0 5&p: -~+/1000#0.01 0 5&p: 10-+/1000#0.01 |domain error | 5&p:10-+/1000#0.01 5&p: 2^_44 |domain error | 5&p:2^_44 ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
