Sam Steingold <ยทยทยท@goems.com> writes:
> Of course, I can print most-positive-short-float to a string,
> measure it's length, multiply by (log 10 2) and get "a good
> approximation":
>
> (round (* (log 10 2)
> (1- (position-if #'alpha-char-p
> (format nil "~a"
> most-positive-single-float)))))
>
> (1 is subtracted to account for the decimal point).
>
> Any suggestions?
There are better ways to do this. (Among other things, you're
assuming that the representation of floats is binary. :-) I think you
want to look DECODE-FLOAT, FLOAT-RADIX, FLOAT-PRECISION, etc. in CLHS.