Dbochman/polling-loop-finalization-escape

Polling Loop Finalization Escape

Problem

A polling loop detects a trigger condition (e.g., "return home detected"), executes finalization steps (dock Roombas, send notification, update state), then should exit. But if any exception occurs during finalization that isn't caught by the inner try/except blocks, the outer loop's generic except Exception handler catches it, logs the error, and continues the loop — causing the trigger to be re-detected and finalization to re-execute on every poll cycle.

Context / Trigger Conditions

• A polling loop with an if trigger_condition: ... return finalization block
• Individual finalization steps wrapped in their own try/except for resilience
• An outer except Exception around the entire poll iteration that logs but doesn't return
• Symptoms: the same action (dock, notify, cleanup) executes repeatedly at poll intervals
• Log pattern: alternating "trigger detected" and "Error: ..." messages in a loop

Solution

Ensure the outer exception handler checks whether the trigger was already detected, and if so, exits instead of looping:

# BEFORE (broken): exception during finalization causes loop to continue
while polling:
    try:
        if detect_trigger():
            try:
                finalize_step_1()
            except Exception:
                log("step 1 failed (non-fatal)")
            try:
                finalize_step_2()  # <-- unexpected error here
            except Exception:
                log("step 2 failed (non-fatal)")
            return  # never reached if step_2 throws something uncaught
    except Exception as e:
        log(f"Error: {e}")  # catches it, but doesn't exit!
    await sleep(interval)   # loops back and re-triggers

# AFTER (fixed): outer handler respects the trigger state
while polling:
    trigger = None
    try:
        trigger = detect_trigger()
        if trigger:
            # ... finalization steps ...
            return
    except Exception as e:
        log(f"Error: {e}")
        if trigger:
            log("Exiting despite error (trigger was already detected)")
            return  # don't loop back and re-trigger
    await sleep(interval)

Verification

After fix, logs should show exactly one "trigger detected" + "docking" sequence, then the monitor ends. No repeated dock commands at 30s intervals.

Notes

• This is especially dangerous when finalization has side effects (sending commands to devices, sending notifications) — each re-trigger executes all side effects again
• In the Roomba case, repeated stop+dock cycles interrupted the robots mid-dock-return, leaving them stopped in the middle of the floor
• The root cause exception may itself be subtle (e.g., Python 3.14 UnboundLocalError on a variable that appears unreachable) — fixing the exception alone isn't sufficient because other future exceptions could cause the same loop behavior
• Always set a flag/variable BEFORE entering the finalization block so the outer handler can check it