Recv - Dynamic Buffer

[Bloodx]

Hello, is there any way to make the Buffer for Recv Dynamic ?

this is the way I actually use it ..

bool CClient::ClientThread()
{
    std::vector<char> szBuffer(65535);
    int iReceivedBytes;
    do
    {
        iReceivedBytes = recv(this->ClientSocket, szBuffer.data(), szBuffer.size(), 0);
        
        if (iReceivedBytes <= 0)
        {
            closesocket(this->ClientSocket);
            return 0;
        }
        this->PacketControl( szBuffer.data() );

    } while (this->ClientSocket != INVALID_SOCKET);

    return 0;
}

 

[GHOST107]

Most packets have a header, wich in most cases have a data size, I don't know what you receive but reading the fulll data in the TCP stream, not too good Idea, you can break the recv in 2 , 1 to read the header, and 1 to read th data.

Here are 2 scenarios:
- If you receive 2 or more packets while you process one it will be stored in the TCP stream, and when you use recv to read from the TCP stream it will read the buffer size you put in recv.
- If the packet is too big while the data is send, the packet is broken into 2 packets or more packets, while you recv the TCP stream, you would read a incomplete packet.

It is simple to recv the header and data, and continue to processing if you have a complete packet(this depends on how your application works, and if the packet contains a header to specify the data size)

 

[Bloodx]

well Packet[0}] contains a unsigned short ( WORD ) with the Size of Packet also..

unsigned short sSize = *(unsigned short)*&Packet[0];

 

[GHOST107]

It is possible to make the buffer dynamic, but not recommended, since if you're using pointers, pointers are not too safe when allocating memory, plus it uses the heap, which is slow if you have to do lots of calculations. vector is a manage version of an array, but it still allocates memory every the capacity of the vector is reached(the STL vector allocates 1.5x - 2x more memory then it is needed).

For example a vector with capacity of 5, when you insert 6 elements it will reallocate memory for 6 elements, and when you add another element it will reallocate again(assuming the allocation capacity is 1x), reallocation's are slow when dealing with lots of calculations.