Comments on: Horrible Java http://davec.org/2009/05/horrible-java/ Mon, 19 Mar 2012 16:57:24 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: dave http://davec.org/2009/05/horrible-java/comment-page-1/#comment-1970 dave Fri, 15 May 2009 10:57:26 +0000 http://davec.org/?p=631#comment-1970 Doubles can represent integers. In fact, half of all the possible values of a double are integer values. The problem is that, beyond a certain magnitude, a double will not have enough precision to distinguish even from odd. The <em>real</em> problem is that no data type, including the "arbitrary precision" ones, can possibly represent all integers up to infinity. Consider <a href="http://mathworld.wolfram.com/GrahamsNumber.html" rel="nofollow">Graham's Number</a> for a start. The starting point for understanding how this code works is in recognising that it is patently impossible. :P Doubles can represent integers. In fact, half of all the possible values of a double are integer values. The problem is that, beyond a certain magnitude, a double will not have enough precision to distinguish even from odd.

The real problem is that no data type, including the “arbitrary precision” ones, can possibly represent all integers up to infinity. Consider Graham’s Number for a start.

The starting point for understanding how this code works is in recognising that it is patently impossible. :P

]]> By: Neil http://davec.org/2009/05/horrible-java/comment-page-1/#comment-1967 Neil Fri, 15 May 2009 01:16:13 +0000 http://davec.org/?p=631#comment-1967 While I have no idea what Java is doing to deal with a concept of infinity, I have a problem with the code before it gets as far as using whatever infinity it has... The parity of a number is only defined for integers, so declaring i to be a double is not technically correct, and we all know technically correct is the best type of correct :-). While I have no idea what Java is doing to deal with a concept of infinity, I have a problem with the code before it gets as far as using whatever infinity it has… The parity of a number is only defined for integers, so declaring i to be a double is not technically correct, and we all know technically correct is the best type of correct :-) .

]]>