ZMQ `recv_json_with_retry` returns `None` on exhausted retries, but callers treat return value as a valid list — potential `TypeError` crash

[GaneshPatil7517]

Problem Description

The recv_json_with_retry() method returns None after 5 failed receive attempts (concore.py line 84). The caller in read() then attempts to check if isinstance(message, list) and len(message) > 0 — this works for None (the isinstance check fails), but the fallback path return message (line 308) returns None to the user's control loop, which typically expects a list.

Technical Analysis

ZMQ receive (concore.py, recv_json_with_retry, line 84):

logging.error("Failed to receive after retries.")
return None

Caller in read() (concore.py, lines 299–308):

message = zmq_p.recv_json_with_retry()
if isinstance(message, list) and len(message) > 0:
    first_element = message[0]
    if isinstance(first_element, (int, float)):
        simtime = max(simtime, first_element)
        return message[1:]
return message   # <-- returns None if all retries failed

A downstream controller calling ym = concore.read(...) and then u = controller(ym) will pass None into a numerical computation, causing a TypeError crash. The ZMQ error except blocks (lines 309–313) only catch zmq.error.ZMQError and Exception — they don't catch the None-return path because no exception is raised.

The send_json_with_retry has a similar issue: it returns None on failure (implicit return), but the caller doesn't check whether the send actually succeeded.

Proposed Improvement

def recv_json_with_retry(self):
    for attempt in range(5):
        try:
            return self.socket.recv_json()
        except zmq.Again:
            logging.warning(f"Receive timeout (attempt {attempt + 1}/5)")
            time.sleep(0.5)
    raise TimeoutError(f"ZMQ recv failed after 5 retries on {self.address}")

Change the method to raise an exception on exhaustion instead of returning None. The existing except blocks in read() will then catch it and return default_return_val, which is the correct behavior.

Impact

• Reliability: None propagation causes crash in user code unrecoverable at runtime
• Julia impl: Must decide on error signaling strategy (exceptions vs Nothing vs Result type)
• Cross-language: Only affects Python ZMQ path; C++ has no ZMQ support yet

[GaneshPatil7517]

hey @pradeeban Sir can i start working on this....

[pradeeban]

sure, go ahead.