What is the difference between following declarations ?

[PrafullaVedante]

1 )

int a;
a = 5;

2 )

int a(5);

3 )

int a = 5;

Prafulla Vedante

[Lost User]

In case 1 you have declared the variable a but it has no value; in reality it will but its contents will be random. The second statement assigns the value 5 to variable a. These two statements do not need to be together, as a can have its value set or reset anywhere it is in scope. Cases 2 and 3 are the same, the variable is declared and initialised at the same time.

One of these days I'm going to think of a really clever signature.

[Orjan Westin]

1. Declaration (after which the variable will be uninitialised, with an unknown value) followed by assignment. 2. Declaration with initialisation. For POD (Plain Old Data, built-in) types, this is effectively no different to 3. but with a class/struct type this would call a different constructor (taking an int parameter, rather than the default constructor). 3. Declaration followed by assignment. Again, for a POD type no different from 2. but with a class/struct type the value 5 could be implicitly converted to the class/struct type in a custom constructor before the assignment, or an assignment operator taking an int could be called. In questions of declaration, initialisation, assignment and so on, what POD types do is well known and specified in the standard, but what class/struct types do depend on their implementation to a much larger extent.

[Stefan_Lang]

Not quite correct. Cases 2 and 3 are equivalent - the assignment operator is, in fact, not used in case 3, not even for a custom class with an override for operator=(). See GotW#1 for further explanations. I have to admit I wouldn't have known the (lack of) difference before checking that site.

[Orjan Westin]

Ah, thank you.