# Read a floating point on 32 bits and return it as a `Float`
	#
	# Using this format may result in a loss of precision as it uses less bits
	# than Nit `Float`.
	#
	# Returns `0.0` when an error is pending (`last_error != null`).
	fun read_float: Float
	do
		if last_error != null then return 0.0

		var b0 = read_byte
		var b1 = read_byte
		var b2 = read_byte
		var b3 = read_byte

		# Check for error, `last_error` is set by `read_byte`
		if b0 < 0 or b1 < 0 or b2 < 0 or b3 < 0 then return 0.0

		return native_read_float(b0, b1, b2, b3, big_endian)
	end