Event‑Driven Programming and I/O Models in Python

The article begins by contrasting the traditional linear programming flow, where execution order is predetermined by code blocks, with the event‑driven model that waits for external events such as mouse clicks or timers before invoking callbacks.

It then outlines three common server handling models: creating a new process per request, creating a new thread per request, and using an event‑driven approach (the most widely adopted for network servers).

Examples of event‑driven programming in a web UI are shown, highlighting how the onclick attribute can trigger a JavaScript function without blocking the main thread.

The article introduces fundamental concepts needed for I/O handling, including user space vs. kernel space, process switching, blocking, file descriptors, and buffered I/O.

Five I/O models are described in detail:

Blocking I/O – the process is blocked during both waiting for data and copying data.

Non‑blocking I/O – the system call returns immediately with an error if data is not ready, requiring the application to poll.

I/O multiplexing (select/poll/epoll) – a single process monitors multiple sockets, blocking only on the multiplexing call.

Signal‑driven I/O – the kernel sends a signal (e.g., SIGIO) when a descriptor becomes ready.

Asynchronous I/O – the kernel performs the entire operation and notifies the process via a signal when finished, eliminating all blocking.

Each model is illustrated with concise Python code snippets, such as a simple blocking server, a non‑blocking server using setblocking(False) , and a select‑based multiplexed server.

Further, the article explains the difference between level‑triggered and edge‑triggered event notifications, noting that epoll supports both.

Finally, the Python selectors module is presented as a high‑level abstraction that automatically chooses the optimal underlying I/O multiplexing mechanism, with a complete example showing registration of sockets, accept and read callbacks, and the event loop.