C + + code style guidelines: POD

An idiot with a rib
Links: https://www.zhihu.com/question/36379130/answer/69853366
Source: Know
Copyright belongs to the author. Commercial reprint please contact the author for authorization, non-commercial reprint please specify the source.

POD, full name plain old Data,plain means it is a normal type, old means it can be compatible with C, and can be manipulated using the most primitive functions of C such as memcpy (). The pod is divided into two basic concepts in the c++11.That Ordinary (trival) and standard layout.

1. The first is the trivial (trival) definition, usually an ordinary class or struct that needs to meet the following definitions:

• Have trivial default constructors and destructors. The default meaning is automatically generated by the compiler for us, not allowed to be defined by ourselves, but because C++11 provides default, it can also be defined by their own =default, such as

struct trival{
Trival () {}=default;
}
is to satisfy this requirement, and
struct notrival{
Notrival () {};
}
This requirement is not fulfilled (even if there is nothing in the body of the constructor we define). This requirement is not tied to a constructor with parameters. You can customize the constructor with parameters.

• Have trivial copy constructors and move constructors. The default meaning is ditto, and you can use =default.
• Has a trivial copy assignment operator and a move assignment operator.
• Cannot contain virtual functions and virtual base classes.

2. Next is the definition of the standard layout:

• All non-static members have the same access level ( the access level is public,private,protected),

struct t1{
Private:
int A;
Public
int b;
The standard layout is not satisfied because a, B access level is different.

• In class and struct inheritance, you need to meet one of the following two scenarios:
• There are non-static classes in a derived class, the derived class can have only one base class that contains only static members.
• The base class has non-static members, and non-static members are not allowed in the derived class.

（ These two sentences look quite around the mouth, in fact, is to illustrate a fact, about the fact that non-static data, derived classes have non-static data then its base class can only be static, and the base class can only have one. If the base class has non-static, then the derived class cannot have non-static. There is a seesaw feeling, non-static opposite sitting is static, father and son class is sitting on the seesaw at both ends of this correspondence relationship. )

• The first non-static type in a class is not of the same type as the base class. Like what

struct a:b{
b b;
int C;
} is not eligible for this condition. Because the first member of a is of the base class B type.

• No virtual classes and virtual base classes (duplicates in trival)
• All non-static data members conform to the requirements of the standard layout, which is actually a recursive definition.

so in c++11, Pod is the two aspects of satisfying the ordinary (trival) and standard layout. You can use the Is_pod<t>::value in <type_traits> to determine if T is a pod type.

Having said so much, then why do we need the pod to satisfy the condition of the data?

1. You can use byte assignments, such as memset,memcpy operations
2. Compatible with C memory layouts.
3. This ensures that static initialization is safe and effective.

C + + code style guidelines: POD