The gcvt function places the converted string in a buffer and
    returns the address of the buffer. If possible, gcvt produces
    ndigit significant digits in F-format, or if not possible, in
    E-format. Trailing zeros are suppressed.

    The ecvt, fcvt, and gcvt functions represent the following
    special values specified in the IEEE Standard for floating-point
    arithmetic:

    Value         Representation

    Quiet NaN     NaNQ
    Signalling    NaNS
    NaN
    +Infinity     Infinity
    -Infinity     -Infinity

    The sign associated with each of these values is stored into the
    sign argument. In IEEE floating-point representation, a value
    of 0 (zero) can be positive or negative, as set by the sign
    argument.

    See also fcvt and ecvt.