Python-在以太网数据包中发送/接收一个numpy数组

Question

I need to set up a network to transfer a numpy array from a host computer to a client computer. This can be as simple as possible except I have to use 16 bit integers. Am I doing it right? Are there better functions/commands I should be using? (Win7, 64bit, Python 2.7) So far, it seems to work but I am a total noob to networking and I haven't found a decent tutorial that explains how all this fits together. (Please suggest one) Thanks!

server.py

import socket, time, numpy as np
myDict = {1:1, 2:2, 3:3}
s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_BROADCAST, 1)
while 1:
    data = np.array(myDict.values(), dtype = np.int16).flatten()
    s.sendto(data, ('<broadcast>',5252))
    time.sleep(1)

client.py

import socket, numpy as np
s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
s.bind(('', 5252))
data = np.empty((1,6), dtype = np.int16)
s.recv_into(data)
print data

Answer 1

You are using datagram sockets (UDP) to transfer your data. UDP is best effort service and you are not guaranteed that every datagram you send will be delivered at the remote end.

An alternative is to use TCP, but the you will need to serialize your numpy data before sending it over the socket and then deserialize it when you receive it.

If you are fine with using unreliable UDP transfers, then what you are doing now is OK.

Edit: When using broadcast - you are basically sending same piece of data to all devices in your directly attached network. Any device listening to UDP port 5252 will be able to receive data that you are sending.