member variable :1 syntax

[Angus]

In a class I see this syntax for member variables:

int m_myvar:1;

But if I try to access this variable it seems set to zero? So what is
this all about?

 

[Victor Bazarov]

In a class I see this syntax for member variables:

int m_myvar:1;

But if I try to access this variable it seems set to zero? So what is
this all about?

Bit fields. Read about them. The problem is that 1 bit for an integer
is probably not enough to keep any value. Try making it 'unsigned'.
And then assign 1 to it and try reading it back, see what you get.

V

 

[Angus]

Bit fields.  Read about them.  The problem is that 1 bit for an integer
is probably not enough to keep any value.  Try making it 'unsigned'.
And then assign 1 to it and try reading it back, see what you get.

V

Ah yes I see now (I think).

This sort of idea is used a lot in the code I am looking at:
bool m_setting1 :1;
bool m_setting2 :1;
bool m_setting3 :1;
bool m_setting4 :1;

etc etc.

I presume it is for efficiency. A bool variable would normally take
one whole byte. But if you had 8 bool values as above then feasibly
they could all be stored in a bit each - and possibly only use one
byte of storage. Probably not worth doing in many cases but if for
example you were storing vlaues in a database with many rows... then I
can see point.

 

[Alf P. Steinbach]

* Angus:

This sort of idea is used a lot in the code I am looking at:
bool m_setting1 :1;
bool m_setting2 :1;
bool m_setting3 :1;
bool m_setting4 :1;

etc etc.

I presume it is for efficiency.

It's unlikely to improve efficiency, but it can be a good way to conform to some
externally imposed memory layout.


Cheers & hth.,

- Alf

 

[Angus]

* Angus:






It's unlikely to improve efficiency, but it can be a good way to conform to some
externally imposed memory layout.

Cheers & hth.,

- Alf

This example demonstrates a memory advantage:

class myclass
{
public:
bool getboolval() { return m_setting1; }


private:
bool m_setting1:1;
bool m_setting2:1;
bool m_setting3:1;
bool m_setting4:1;
bool m_setting5:1;
bool m_setting6:1;
bool m_setting7:1;
bool m_setting8:1;
};


int main(){
int size = sizeof(myclass);
std::cout << size; //reports 1


If however, the :1 syntax above is removed, then size is reported as
8.

My test was using MSVC++v6 on Windows XP (32 bit).

 

[Alf P. Steinbach]

* Angus:

This example demonstrates a memory advantage:

class myclass
{
public:
bool getboolval() { return m_setting1; }


private:
bool m_setting1:1;
bool m_setting2:1;
bool m_setting3:1;
bool m_setting4:1;
bool m_setting5:1;
bool m_setting6:1;
bool m_setting7:1;
bool m_setting8:1;
};


int main(){
int size = sizeof(myclass);
std::cout << size; //reports 1


If however, the :1 syntax above is removed, then size is reported as
8.

My test was using MSVC++v6 on Windows XP (32 bit).

<digression>
You really should upgrade compiler. You'd get the same result, but MSVC 6 is a
pre-standard compiler and has trouble with modern template code, forces you to
write non-standard code, plus it has a host of bugs in floating point support.
AFAIK the latest MSVC compiler is bundled with the SDK, so it's free even
without installing the lobotimized free version of Visual Studio.
</digression>

Yes, it's a memory advantage, but have you tested efficiency?


Cheers, & hth.,

- Alf