socket timeout behavior change in v1.20

[manix]

I have this bit of code that used to work with rp2-pico-w-20220926-unstable-v1.19.1-451-gbdbc44474:

sock = socket.socket()
osock = sock

# allow 90 seconds for initial ws connection
sock.settimeout(90)

sock = ussl.wrap_socket(sock)

addr = socket.getaddrinfo(uri.hostname, uri.port)
sock.connect(addr[0][4])

# Omitted a bunch of code handling headers and stuff
# not relevant to the question

# Once a connection is established
osock.settimeout(1)

while True:
    try:
        bytes = sock.read(2)
        # process incoming data..

    except OSError as e:
        # OSError -110 is timed out, so do nothing
        if e.errno != -110:
            raise e

    # do other things here

The way this code used to work was it froze for a second waiting for incoming data, if there was none then it timed out and went ahead to process other things and then repeated.

I have just updated to rp2-pico-w-20230426-v1.20.0 and it goes into the while loop, reads the first incoming websocket messages but after it times out subsequent loops no longer read but rather instantly throw OSError -110.

How can I resume the old behavior?

[massimosala]

I haven't a pico, but usually I do this:

from select import select

# osock.settimeout(1)

while True:
    try:
       rx, _, errors = select( [sock], [], [sock], 1 )

       if errors :           # the peer has close the socket, the network link is down, etc...
             ...

       if rx :               # there is incoming data
             try :
                 bytes = sock.read(2)
                 # process incoming data..