http://dev.clojure.org/jira This file is an XML representation of an issue en-us 4.4 649 25-07-2011 http://dev.clojure.org/jira/browse/CLJ-1006 Clojure <p>Hi,</p> <p>As discussed on the mailing list in the message "When arithmetic on a computer bite back" (01/jun)</p> <p>There may be bug in the way quotient is implemented for bigdec.</p> <p>user&gt; (quot 1.4411518807585587E17 2) ;; correct with doubles<br/> 7.2057594037927936E16<br/> user&gt; (quot 1.4411518807585587E+17M 2) ;; wrong with BigDecs<br/> 72057594037927935M </p> <p>&#8211;<br/> Laurent</p> Linux 3.2.0-24-generic #39-Ubuntu SMP i686 GNU/Linux CLJ-1006

Quotient on bigdec may produce wrong result

Defect Major Closed Declined Unassigned laurent joigny bug Fri, 1 Jun 2012 16:12:34 -0500 Fri, 1 Mar 2013 12:47:08 -0600 Fri, 9 Nov 2012 08:49:48 -0600 Release 1.4 0 1 <p>I can reproduce the bug when using BigDecimal constructor on String.<br/> See attached file for a test class.</p> <p>More infos :<br/> java version "1.6.0_24"<br/> OpenJDK Runtime Environment (IcedTea6 1.11.1) (6b24-1.11.1-4ubuntu3)<br/> OpenJDK Client VM (build 20.0-b12, mixed mode, sharing)</p> <p>A simple test file, that you can drop in Clojure sources and execute to reproduce the bug on BigDecimal constructor using String as argument.</p> <p>Seems to be a general precision problem. Note that in</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>user&gt; (quot 1.4411518807585587E17 2) ;; correct with doubles 7.2057594037927936E16 user&gt; (quot 1.4411518807585587E+17M 2) ;; wrong with BigDecs 72057594037927935M </pre> </div></div> <p>the double result is actually wrong and the bigdec one is correct. The problem which lead to the wrong conclusion is that in your calculation the input number is already wrong.</p> <p>So the moral is: don't use any floating points (neither doubles nor bigdecs) for computations involving divisibility tests.</p> <p>For bigdecs, you can set the math context for making computations throw exceptions if they lose precision, though:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>user&gt; (binding [*math-context* (java.math.MathContext. 1 java.math.RoundingMode/UNNECESSARY)] (quot (bigdec (Math/pow 2 58)) 2)) ;Division impossible ; [Thrown class java.lang.ArithmeticException] </pre> </div></div> <p>Not a bug. Just floating-point arithmetic.</p> Global Rank