Optimise longValueExact(BigInteger n) in BigIntegerUtil

[mpnd]

Given a value n of type BigInteger, longValueExact() computes the exact long value of n, that is n > Long.MaxValue or n < Long.MinValue result in ArithmeticException. Currently, this conversion is calculated by comparing n to BigInteger of Long.MaxValue and Long.MinValue. Is there a solution which doesn’t involve comparison of two BigIntegers?

[Wilko Henecka]

If you assume that long is always a 64 bit integer in two's complement format, than you could test:

if (n>=0 and n.bitlength()<64) or (n<0 and (n-1).bitLength()<64) then 
    return n.longValue()
else 
    throw Exception

But is comparing two BigIntegers such a big deal?

[mpnd]

The method longValueExact(BigInteger n) has been removed in bb46cb6.