Also Chicken Scheme:

  (- 0.3 (+ 0.1 0.2))
    -5.55111512312578e-17

Of course you can always (use numbers), but then you have to use (/ 1 20) and (/ 2 10). "0.3" is still considered "inexact", because of the decimal point. I don't like this - finite decimals are perfectly rational numbers and trivially convertible, why can't the interpreter treat them as such?

Another less-common language. The docs suggests they use "C doubles" for all floats.

    Welcome to SWI-Prolog (Multi-threaded, 64 bits, Version 6.6.6)

    ?- Sum is 0.1 + 0.2.
    Sum = 0.30000000000000004.