The classical assumptions behind the two-sample $t$-test

Suppose $t$ test is applied to test the equality of means of two random variables X and Y, where X and Y are independent. Then: