These functions return the exponent part of their argument
as a signed integer. When no error occurs, these functions
are equivalent to the corresponding
-.B logb()
+.BR logb ()
functions, cast to (int).
An error will occur for zero and infinity and NaN,
and possibly for overflow.