actioncable

Presence is usually overkill, but for collaboration features it’s valuable: show who’s online in a room. I identify connections with current_member in ApplicationCable::Connection, then in a channel I broadcast updates when members subscribe/unsubscri

If a feature is fundamentally “live” (comments, activity), I reach for model broadcasts. Rails can broadcast Turbo Stream fragments on after_create_commit, and the UI subscribes with turbo_stream_from. The beauty is that it’s still server-rendered HTM

Never subscribe clients to guessable user-specific streams. Use signed_stream_name so a user can only subscribe to their own broadcasts. This is essential when streaming private notifications.

Typing indicators can be done without a complex protocol. I broadcast a small turbo stream replace to a typing_indicator target when a user starts typing, and another replace to clear it after a timeout. On the client, a Stimulus controller sends “typ

Even with model broadcasts, it’s useful to know how streams map to ActionCable channels. Turbo::StreamsChannel is essentially a channel that streams from a signed stream name. When I need custom behavior, I still follow the same scoping rules: identif

When I’m on Rails 7+, I like broadcasts_to because it makes realtime behavior obvious in the model. Instead of writing explicit after_create_commit hooks, I declare that a model broadcasts to its parent or to a scope. Then Turbo uses conventional part

For long operations (imports, conversions), run work in a job and stream progress updates. The job emits broadcast_update_to so the UI updates live without polling.

For a notifications dropdown, I keep the list server-rendered and let Turbo Streams keep it fresh. The dropdown content sits in a turbo_frame_tag (so you can also navigate to a full notifications page), and the unread badge is a separate small target

Presence is often more about “who is here now” than perfect accuracy. Use Redis sets with expiries updated by pings. This keeps the system simple and operationally understandable.

When updates can happen from multiple places, model-level broadcasts keep the UI consistent across tabs without sprinkling stream logic in controllers. Use after-commit hooks so broadcasts only occur once the write is durable.