I was using your own definition. You said ~0 was the highest integer that can be represented precisely without using Math::BigInt. That's wrong, and I explained why.

Your original post if still wrong by your new definition. If you want to the highest integer that can be used in a range, both your answer of ~0 and your suggestion to use Math::BigInt for larger integers than ~0 are wrong!!

I have already mentioned in this thread that ranges require signed integers, so ~0 >> 1 (2147483647) would be the answer.

Update: I originalyl had ~-(~0+1) instead of ~0 >> 1, but that won't work if the system's floats have a smaller range than the integers.