Performing boundary value testing on float range

Question

I want to perform boundary value analysis on a program which takes 3 float variables as input and returns the largest out of them. There's no range specified hence I impose the range to be the one implied by the no. of bytes used to store float variables, in my case 4 bytes. Hence, my float range becomes -3.4+38 to +3.4+38

Problem is: in boundary value u check these points: say your range is : 1<= x <= 10, the points which you need to check are : {1, 2, 9, 10}

In my float example I can't come to understand which value comes just above -3.4+38 and which value falls just below +3.4e+38. someone help here.I don't get this float representation.

Answer 1

Let us assume that you are programming in C. Any other high-level language should map the concepts to similar notations.

The float value just above -3.4e+38f is nextafterf(-3.4e+38f, 0). The float value just below 3.4e+38f is nextafterf(3.4e+38f, 0), and you can also compute it as the opposite of the previous one if you find that clearer.

You can also obtain these constants by typing -3.399999749e+38 and 3.399999749e+38 respectively if you do not have a function nextafterf.

---

Note that 3.4e+38f is not the largest float, only a rough approximation of it (and 3.4+38, or approximately 41.4, even less so).