Static variable declaration and definition

Question

Consider the following code:

#include "stdafx.h"
#include <iostream>

class Eelement {
public:
    static int m_iVal;
};

int Eelement::m_iVal = 33;

int main()
{
    //...
    return 0;
}

Question:
Why, during the initialization of the static variable m_iVal, I must to precede its name by its type int a second time (first time in class definition) ?

int Eelement::m_iVal = 33;

Why the compiler imposes this syntax which for me looks much more like a double declaration than other things because compiler already knows its type (from the definition of the Element class).

It's the same as with member functions, there can be only one definition. — Lukas-T, Feb 12 '20 at 09:12
@NutCracker Thank you for the link. The article discuss about this point. But something always worries me: if I write this: **char Eelement :: m_iVal = 33;** The compiler will tell me that the declaration is not compatible with **int Element :: m_iVal** therefore, it knows its type well ! — Landstalker, Feb 12 '20 at 09:44
@Landstalker look at the answers in the duplicate question. The last answer shows the improvement introduced in C++17 : `inline static` — Robert Andrzejuk, Feb 12 '20 at 15:12

TonySalimi · Answer 1 · 2020-02-12T11:34:31.303

1

Why the compiler imposes this syntax which for me looks much more like a double declaration ...

It is not double declaration. Inside the class, it is declaration, but when you put:

int Eelement::m_iVal = 33;

outside of the class, it is definition. The declarations may happen in different translation units (TUs). By defining the static variable, you tell the compiler which translation unit to use for putting the static variable there.

edited Feb 12 '20 at 11:34

answered Feb 12 '20 at 09:18

TonySalimi

8,257
4
33
62

Technically a definition is also a declaration. – fefe Feb 12 '20 at 09:45

463035818_is_not_an_ai · Answer 2 · 2020-02-12T11:04:22.060

1

Eelement::m_iVal = 33;

is an assignment to the static member. The int is not superflous, because only this is a definition:

int Eelement::m_iVal = 33;

edited Feb 12 '20 at 11:04

answered Feb 12 '20 at 09:19

463035818_is_not_an_ai

109,796
11
89
185

Thanks. But The compiler reports an error if I try to write: **char Eelement::m_iVal = 33;** and warns me of an incompatibility with the declaration **int Element :: m_iVal** If I suppose that the compatibility between **"the declaration"** and **"the definition"** must be always respected, type mention can be removed in the "second" definition. like that, we are sure that the compatibility of the type will be always respected. – Landstalker Feb 12 '20 at 10:00
But? I didnt imply that you can have a definition that does not match the declaration. Though still you cannot remove `int` in the example, because then it is not a definition – 463035818_is_not_an_ai Feb 12 '20 at 10:57
Yes I understood. it was in the continuity of my reflexion and not in direct relation to your explanation ;) – Landstalker Feb 12 '20 at 11:39

Static variable declaration and definition

2 Answers2