+/**
+ * Maximum int64_t value that can be stored in a double without chancing losing precision.
+ * IEEE doubles have 53 bits of mantissa, 10 bits exponent, 1 bit sign.
+ * IBM Mainframes have 56 bits of mantissa, 7 bits of base 16 exponent, 1 bit sign.
+ * Define this constant to the smallest value from those for supported platforms.
+ * @internal
+ */
+static const int64_t MAX_INT64_IN_DOUBLE = 0x001FFFFFFFFFFFFFLL;