Question

I have a set of structs which come from some C code I’m converting to C++, they should be POD types according to what I understood. Here are some examples:

struct Data {
  u16 type : 12;
  u8 variant : 3;
  bool isTop : 1;
};

struct DData {
  u16 type : 12;
  u8 variant : 3;
  u8 layer : 1;
};

struct TData {
  struct Data data1, data2;
  struct DData ddata[MAX];
  u16 x;
  u8 y;
  u8 s : 4;
  u8 l : 4;
  u8 wl : 3;
  u8 wr : 3;
  u8 lt : 2;
};

I have many methods which up to now had pointer to these structs as parameters and I was planning to move them directly inside the struct declarations to save a lot of typing, eg.

struct Data {
  u16 type : 12;
  u8 variant : 3;
  bool isTop : 1;

  inline bool hasFlag(u64 flag) { return Types::specs[type].flags & flag; }
 };

I was wondering if every method I can add in this way will be safe for this struct, these structures are serialized over the network and into binary files so I need to be sure that this can’t potentially break anything. If I’m right they should remain POD types so no particular problems or overhead should appear.

Answer 1

That should remain a POD.

You can be sure by using the std::is_pod trait:

std::is_pod<Data>::value;

will be true if Data is a POD, false otherwise.

Anyway, I wouldn’t bother moving the functions into the struct body. You’re not really gaining anything besides a different calling syntax.