Managing background thread termination during script cancellation

Recommended approach

Streamlit’s st.stop() only halts the script’s execution flow; it does not automatically signal background threads to stop. To ensure those threads terminate cleanly, coordinate them with an explicit stop flag that the script sets before (or as part of) calling st.stop().

Implementation steps

1. Create a shared threading.Event (or a boolean flag in st.session_state) that is initially cleared.
2. Pass this event to each background thread when you start it.
3. Inside the thread’s work loop, periodically check event.is_set() (or the session‑state flag). If the flag is set, break out of the loop and return.
4. When the user triggers cancellation (e.g., clicks a button that calls st.stop()), first set the event/flag, then call st.stop() to halt the script.
5. Optionally, join the threads with a timeout after setting the flag to wait for graceful shutdown.

Why daemon threads alone are insufficient

Setting daemon=True makes threads terminate when the Python process exits, but it does not give them a chance to clean up resources. Relying on daemon status can leave files open, locks held, or temporary data incomplete.

Verification

• Add a log or print statement immediately after the flag check inside the thread to confirm the exit path is taken.
• After triggering st.stop(), use threading.enumerate() to verify that the background threads are no longer listed (or have joined).
• Observe whether any resource cleanup code runs; if not, adjust the flag‑checking frequency or ensure the flag is set before st.stop().