CareerCup

My answer:
- verify that the frequency is distributed evenly for all the possible numbers
- Verify that the distance between repititions is about the size of the set (1000) for each number
- Verify that the distance between consecutive random numbers is random in itself
- Verify that the numbers 0-999 are only generated
- Verify that in each run you do not get the same sequence of random numbers. Seed it differently
- For visual identification of anomalies, plot the values (y-axis as random number, x-axis as each attempt)

1. Random number => no predictable pattern in numbers generated => the diff between the nos should not be the same, etc

2. Each number in between 0-999 has to have equal probability of occurrence. So, we should maintain frequency counts and verify that no single number stands out w.r.t. frequency.

Chi Square test.

I would be referring to diehard tests. Check Wikipedia.
And, of course that any returned int fits the specified range.
For really depth testing, we could test it is not leaking memory, etc.