Incorrect C header declaration in LIFX Protocol V2 Message headers docs

[xoblite]

Hi,

please be aware that at least with some compilers (e.g. MSVC Community 2019 and its bundled toolchain), the C header declaration included as part of the LIFX Protocol V2 Message headers documentation is incorrect in that it doesn't lead to the intended/required packing of the header, specifically the following part of the frame typedef...

uint16_t   protocol : 12;
uint8_t   addressable : 1;
uint8_t   tagged : 1;
uint8_t   origin : 2;

...which instead should be all of the same type uint16_t in order not to confuse the compiler...

uint16_t   protocol : 12;
uint16_t   addressable : 1;
uint16_t   tagged : 1;
uint16_t   origin : 2;

Thanks for making the LAN protocol option available; keep up the good work!

BR//KHH (@xoblite)

[delfick]

I'm unfamiliar with MSVC, what does it complain about? Surely it doesn't matter if it's a uint8 or unit16 for these given they specify only a small portion of bits anyway?

[pkruger]

Haven't used MSVC in a while. I think the issue comes in with how the compiler maps the bits, i.e. it is expecting to map all the bits into a uint16_t, i.e. the same uint16_t, and having uint8_t in there as well seems to imply they additional bits should go into another variable which doesn't make sense since there is none. We can definitely look info this.

[delfick]

@xoblite if you have the following, does MSVC complain?

typedef struct {
  /* frame */
  uint16_t size;
  uint16_t protocol : 12;
  bool  addressable : 1;
  bool  tagged : 1;
  uint16_t  origin : 2;
  uint32_t source;
  /* frame address */
  char  target[8];
  char  reserved_1[6];
  bool  res_required : 1;
  bool  ack_required : 1;
  uint8_t  reserved_2 : 6;
  uint8_t  sequence;
  /* protocol header */
  uint64_t reserved_3;
  uint16_t type;
  uint16_t reserved_4;
  /* variable length payload follows */
} Header;