Issues with (fix)

Hello all,

Two questions really, I'm trying to test whether a real ends with .0 (eg 200.0)
I've tried using (fix) then subtracting the real and integer to see if it's 0.0. This seems to work generally but now and again (fix) is giving me a different figure (sometimes something along the lines of 3.54451e-006, sometimes 199) so I was wondering if this is normal behaviour. If it is then, is there another way to check for the above?

Thanks, Steve

 

It has to do with the way 'puters deal with ans stores REAL numbers.

I'd use something like;

(equal (rem 200.0 1) 0 1e-14)

Your never going to 100% sure with large values.

-David

 

It's a rounding problem with some REALS that cannot be represented exactly
by binary.

One way around it:

(defun IsWholeNumber (x)
(= 0.0 (atof (rtos (- x (fix x)) 2 8))))


--
R. Robert Bell, MCSE
www.AcadX.com


| Hello all,
| Two questions really, I'm trying to test whether a real ends with .0 (eg
200.0)
| I've tried using (fix) then subtracting the real and integer to see if
it's 0.0. This seems to work generally but now and again (fix) is giving me
a different figure (sometimes something along the lines of 3.54451e-006,
sometimes 199) so I was wondering if this is normal behaviour. If it is
then, is there another way to check for the above?
|
| Thanks, Steve
|

 

Yup, that's better... no string conversion.

--
R. Robert Bell, MCSE
www.AcadX.com


| It has to do with the way 'puters deal with ans stores REAL numbers.
|
| I'd use something like;
|
| (equal (rem 200.0 1) 0 1e-14)
|
| Your never going to 100% sure with large values.
|
| -David
|
| stevej wrote:
| > Hello all,
| >
| > Two questions really, I'm trying to test whether a real ends with .0 (eg
200.0)
| > I've tried using (fix) then subtracting the real and integer to see if
it's 0.0.
| > This seems to work generally but now and again (fix) is giving me a
different
| > figure (sometimes something along the lines of 3.54451e-006, sometimes
199) so I
| > was wondering if this is normal behaviour. If it is then, is there
another way
| > to check for the above?
| >
| > Thanks, Steve
| >
|

 

Robert,

On second thought

(rem 1.99999999999999999999999 1) wouldn't be a good thing. -David