Question

declaring a struct Table:

struct Tables {
       int i;
       int vi[10];
       Table t1;
       Table vt[10];
};

Tables tt;

assuming that a user-deault contructor is defined for Table.

here tt.t1 will be initialized using the default contructor for Table, as well as each element in tt.vt.

On the other hand tt.i and tt.vi are not initialized because those objects are not of a class type.

so we remain with a semi-initialized object tt.

if I understood well – if tt.i or tt.vi won’t be explicitly initialized i the code, after creating tt, an error will be thrown if we try to read a value from them?

2) can someone explain it to me, why cpp designers didn’t want to simply initialize the built-in types int and int[] to zero?

Answer 1

No, no error will be thrown. You will have a mild case of undefined behaviour. However, as integers don’t have trap values, there is no way the behaviour can be detected.

Note that the C++ language itself throws exceptions only very, very rarely – about the only times I can think of where it does it is when performing an invalid cast to a reference via dynamic_cast, and when new fails to allocate. Of course the Standard Library may throw in a number of error conditions.

As to why C++ works that way (and C too), well initialisation takes time, and if it is not needed that is time wasted. For example, if you were going to read user input immediately into those variables, there is little point in initialising them before you do so.