About null pointers. Yep, we’ve heard that the one of the main people involved in popularizing null pointers regretted it. Why?
20180918/https://en.wikipedia.org/wiki/Null_pointer
Well, the goal here was to have the compiler do all type safety checks at compile-time to guarantee that the complex data types are absolutely safe. No runtime type checking necessary. But what if you have a function that can fail at runtime and can’t really return a valid object? Well, as I understand it, the story must work like this. You use two separate variables, one that is always the correct, static object type, and another to indicate if failure has happened. Then it is up to the program to check for failure at runtime. If it doesn’t, the program will pass around and attempt to process a dummy object of the correct type, which should fail at runtime, maybe even cause a crash if the complex type code is poorly written.
So, the point that I wanted to put forward is this. There is no easy way to do “safe typing” with complex types. By definition, the data and behavior must be more complex, and that requires more logic, which means there will be more points of failure.