typcasting a struct

[vladic256]

Hello,

Currently am working on a project which creates a fake ip package and transmits that packet to a second copy of the same program with a different address. My goal is is to create a fake TCP struct and transmit that over the ip packet.

as of now for testing my struct is define as this

struct tcp
{
   string message;
};

an ip struct is define as this.

struct ip_packet
{ byte source_addr[6];
  byte dest_addr[6];
  short int protocol;
  short int length;
  byte content[0]; };

where the byte content[0] is where you store the data. Byte is declare like this "typedef unsigned char byte;"

my idea is to piggy back the tcp struct inside the ip packet by using typecast like this

tcp a;
ip_packet * x;

x->content[0] = (byte)&a;

it compiles just fine, but gives me a warning that states
warning: cast from pointer to integer of different size

when the program runs on and the second copy of the program gets the ip packet I just get a segmentation fault.

on the receiving end of the code I have it like this

   byte b;
    tcp rev;
        b = p->content[0];
     
      rev = (faketcp &)b;

any idea of how to properly typecast this?

[VictorN]

an ip struct is define as this.

struct tcp
{
   string message;
};

tcp a;
ip_packet * x;
x->content[0] = (byte)&a;

Get rid of tcp struct.
just use something like

   string message;
....
p_packet * x;

x->content = (byte*)message.c_str();

[ZuK]

x->content = (byte*)message.c_str();

You can't assign to an array.

Use strcpy() instead.

strcpy( (char *)x->content, message.c_str());

Kurt

[vladic256]

Well the TCP struct was just for testing. My plans is too add more to that struct and be able to transmit the complete struct to the other end of my transmission.

[VictorN]

You can't assign to an array.

Use strcpy() instead.

strcpy( (char *)x->content, message.c_str());

Kurt

Well, before copy you have to allocate a buffer to copy to.

[ZuK]

Well. I understand that arrays of size 0 are not legal. Some/most C compilers let you get away with this kind of structures just to be able to append some data to this structures.
Usually used this way

char * content = "some string";
struct ip_packet * packet = malloc( sizeof(struct ip_packet) + strlen(content) +1 );
strcpy(packet->content, content);
packet->length = strlen(content);

Kurt

[S_M_A]

That way of doing it (ending a struct with a zero length array that is) even has a name that I unfortunately not can remember at the moment...

[Lindley]

Well the TCP struct was just for testing. My plans is too add more to that struct and be able to transmit the complete struct to the other end of my transmission.

The key point which you seem to have missed is that you cannot simply send a "flat" struct if it contains a non-POD type such as a std::string. You'll be transmitting the bytes which contain the string internals you don't care about, not the bytes which contain the string data that you do want.

[Peter_B]

That way of doing it (ending a struct with a zero length array that is) even has a name that I unfortunately cannot remember at the moment...

In C99 they are called 'flexible array members', and are declared like this:

int array_name[];

[Paul McKenzie]

[QUOTE=vladic256;2066868]as of now for testing my struct is define as this

struct tcp
{
   string message;
};

This won't work as Lindley pointed out. You can't take non-POD objects, break them down into bytes, send those bytes over a socket (or save to a file, or whatever you want to do with these bytes) and magically that same object shows up on the other side of the socket. "Teleportation" won't work with C++ non-POD objects. The only types you can transmit are POD types, i.e. 'C'-compatible types -- int, short, double, float, long, char, etc. and arrays/structs containing these types and only these types.

To create C++ non-POD objects, those objects must be properly constructed. This means the object must be created by having the constructor create the object. What you must do is send the data the string object holds, not the object itself. On the other side, you declare a string, and set the string to the data that is coming in.

Regards,

Paul McKenzie