Rationale

1. Why does a binary operation on two safe<int> values not necessarily return another safe type ?
2. Why is there no policy driven design for handling overflows
3. Why is Boost.Convert not used.

1.	Why does a binary operation on two `safe<int>` values not necessarily return another `safe` type ?
	There are a number of it was hard to implement. it doesn't really seem necessary. We always do SOMETHING with result of the operation. This will result in an assignment or a conversion to some other type where the result will be validated.
2.	Why is there no policy driven design for handling overflows The question was - to which type does one apply it to? Consider the following example: safe<int, overflow_policy_1> t1 = 2; safe<int, overflow_policy_2> t2 = 4; unsigned int x = t1 - t2; // which policy should be invoked?
3.	Why is Boost.Convert not used. I couldn't figure out how to use it from the documentation.