Hi. I'm getting some really strange behavior on LWM: (* 0.02 5) ==> 0.099999994 I can't imagine I'm the first person to notice this. Am I missing something fundamental about how to carry out floating point arithmetic in LispWorks? Thanks, -Chris
> I'm getting some really strange behavior on LWM: > > (* 0.02 5) > ==> 0.099999994 which LispWorks version do you use? I can't reproduce it in LispWorks 4.3.7, Windows version. But your problem is a general problem, because not every decimal number can be expressed as a floating point number. If you are calculating with money, try using cents. Regards, Frank
On Oct 30, 2004, at 2:41 PM, Frank Buss wrote: > which LispWorks version do you use? I can't reproduce it in LispWorks > 4.3.7, > Windows version. LispWorks 4.3.7 running on OS X 10.3.5. Incidentally, MCL 5.0 gives the correct answer on the same system. > But your problem is a general problem, because not every decimal > number can > be expressed as a floating point number. If you are calculating with > money, > try using cents. In this case, I am able to represent the quantities as integers. However, while I certainly understand that there are limits to floating point representation, it seems that LispWorks should be able to adequately represent integer multiples of 0.02. Comparing MCL to LispWorks, I find LispWorks: least-positive-single-float ==>1.4012985E-45 (float-precision 0.02) ==> 24 MCL 5.0: least-positive-single-float ==> 5.0E-324 (float-precision 0.02) ==> 53 Are the LW settings different for PC and Mac? If so, is there a way to adjust these values? Thanks, -Chris S
> Comparing MCL to LispWorks, I find > LispWorks: > least-positive-single-float ==>1.4012985E-45 > (float-precision 0.02) ==> 24 on Windows it has other limits: CL-USER> least-positive-single-float 4.9406564584124657E-324 CL-USER> (float-precision 0.02) 53 CL-USER> least-positive-double-float 4.9406564584124657E-324 Regards, Frank
On Oct 30, 2004, at 2:17 PM, Chris R. Sims wrote: > > I'm getting some really strange behavior on LWM: > > (* 0.02 5) > ==> 0.099999994 For what it's worth, I get exactly the same answer (i.e., 0.099999994) from all of the following on Mac OS X 10.3.5 on a dual G5: sbcl OpenMCL LWM 4.3.7 acl trial edition clisp cmucl In fact, MCL 5.0 is the *only* common lisp implementation of the 7 running on this machine that returns 1.0. Thought you'd want to know. regards, Ralph Raffael Cavallaro, Ph.D. raffaelcavallaro@mac.com