[Ecls-list] Floating point number precision...

[Goffioul Michael]

Look at the following session:

> (format nil "~G" 1d22)
"10000000000000000000000.    "
> (format nil "~G" 1d23)
"099999999999999990000000.    "
> (format nil "~G" 1d24)
"1000000000000000000000000.    "
> (format nil "~G" 1d25)
"10000000000000000000000000.    "
> (format nil "~G" 1d21)
"1000000000000000000000.    "

The case of 1d23 is somehow annoying and buggy.
(Note: this was a CLISP reported bug, I just tried it out in ECL to
see if it had it as well)

Michael.