The compiler gives error if any occurs...then how comes the run time error if compiler had already compiled ?

Answer 1

Consider:

int array[5];

std::cout << array[7];

// The array index is clearly out of bound, the array is uninitialised and contains garbage values which may cause errors to your remaining code, but the code is correct syntax-wise.

// tl;dr - There are some things that the compiler does not check. Successful compilation does not guarantee proper program flow during execution. This is why these errors are called "run-time" errors.

A better analogy would perhaps be for a car to contain all required components, but incorrectly assembled during manufacturing process. Imagine, for example a car has the steering, the gears, the airbags, the seats, the engine; The 'compiler' confirms that all exist and is installed, but the compiler may not know that your steering is installed upsidedown. Driving the car into a wall would be more like input error.

Answer 2

To continue Conway's analogy, there might be no mechanical problem with the car - so it starts just fine. However, you can give it the wrong input and crash it into a wall...

Similarly, the code could be just fine:
print a/b

However, b is supplied by the user - and if the user enters 0, you get a ZeroDivisionError...

Answer 3

Think of this as a car. The compiler only check if you car has any problem. It does not know how you will drive car.