POD-ness of a type

[Dave]

Hello all,

It appears that 3.9/10 and 9/4 taken together define POD. The definition
does not seem to exclude base-class subobjects that are non-POD (though it
does exclude data members that are non-POD). So...

class base // non-POD due to str
{
public:
string str;
};

class derived: public base // POD?????
{
public:
int a;
};

According to my interpretation, derived would be POD. This sure doesn't
seem right though. What am I missing?

Thanks,
Dave

 

[Victor Bazarov]

It appears that 3.9/10 and 9/4 taken together define POD. The definition
does not seem to exclude base-class subobjects that are non-POD (though it
does exclude data members that are non-POD). So...

class base // non-POD due to str
{
public:
string str;
};

class derived: public base // POD?????
No.

{
public:
int a;
};

According to my interpretation, derived would be POD. This sure doesn't
seem right though. What am I missing?

You're missing the word "aggregate" in 9/4. Read the definition of
an aggregate in 8.5.1.

Victor

 

[Andrey Tarasevich]

...
It appears that 3.9/10 and 9/4 taken together define POD. The definition
does not seem to exclude base-class subobjects that are non-POD (though it
does exclude data members that are non-POD). So...

class base // non-POD due to str
{
public:
string str;
};

class derived: public base // POD?????
{
public:
int a;
};

According to my interpretation, derived would be POD. This sure doesn't
seem right though. What am I missing?
...

Read the definition again. Only _aggregate_ classes can be POD-classes.
The term "aggregate" is defined in 8.5.1/1