examples.md

Mediator — where it lives outside this repo

Cited, real implementations to study (or point an agent at) when designing or reviewing coordinator-shaped code.

Python standard library

• queue.Queue / asyncio.Queue — the degenerate mediator: producers and consumers know the queue and never each other; the pairwise coupling that would exist lives in one thread-safe object. docs.python.org/3/library/queue.html
• Tk variable tracing — tkinter widgets coordinate through shared Variable objects with trace callbacks rather than direct references. docs.python.org/3/library/tkinter.html (unverified)

Major ecosystems

• Django Form.clean() — cross-field validation in one method: fields that depend on each other never reference each other; the form owns the rule ("if shipping is express, phone is required"). docs.djangoproject.com/en/stable/ref/forms/validation/#cleaning-and-validating-fields-that-depend-on-each-other
• python-telegram-bot's Application — handlers register with one dispatcher; updates route through it; handlers never call each other. docs.python-telegram-bot.org (unverified)
• Message brokers (RabbitMQ, Kafka) — the mediator at architecture scale: every producer and consumer couples to the broker's topology, none to each other. The god-object risk scales up too — topic sprawl is recheck sprawl. (concept citation)

What to notice across all of them

Each one is defined by the references it removes — Django fields don't import each other, queue consumers can't name their producers. And each bounds the mediator's scope: clean() owns validation only, a queue owns transport only. When reviewing mediator code, ask what pairwise references died, and what stops the hub from absorbing rules that belong to the domain.