10.0 * 0.1 =?= 1

[utzoo!henry]

A friend of mine once made a very true comment:  "floating point is an
analog box anyway".  To my mind, the only thing one is justified in
assuming about floating-point arithmetic is that the result is somewhere
near what it should be.  Assuming that 0.5 * 2 == 1 is just as foolish
as assuming 10.0 * 0.1 == 1;  there will always be some machine where
it's not, for one reason or another.  Pray for machines which can support
efficient interval arithmetic.

This is not to dispute that {x *= y} != {x = x * y} is wrong code.

						Henry Spencer
						decvax!utzoo!henry