robius-matrix-integration_skill

Overview

This skill provides a practical pattern for integrating the Matrix SDK into Makepad apps using Robrix and Moly-inspired approaches. It focuses on non-blocking async architecture: sliding sync for room lists, per-room timeline subscribers, and a request/response worker pattern that keeps UI and networking separate. The goal is reliable real-time room/timeline handling, media fetches, pagination, and UI notifications for Makepad-based clients.

How this skill works

The UI thread submits MatrixRequest enum variants into a single async worker via an unbounded sender. The worker task matches requests and spawns dedicated Tokio tasks for long-running operations (pagination, joins, media fetch). Per-room state stores a timeline and a crossbeam channel sender; timeline.subscribe streams diffs which are converted into TimelineUpdate messages sent to the UI. A SegQueue holds high-frequency room list updates and SignalToUI::set_ui_signal() notifies the UI to drain pending updates.

When to use it

• Building a Matrix client UI in Makepad that must stay responsive while performing async Matrix operations
• Implementing sliding sync and efficient room list updates for large accounts
• Managing per-room timeline subscriptions and real-time message flows
• Handling media downloads, avatars, typing/read indicators, reactions, and pagination without blocking the main loop
• Coordinating results from async workers back into the app action loop

Best practices

• Never block the central request receiver—spawn a new task for each long operation
• Use crossbeam channels per room for timeline updates to avoid global contention
• Always call SignalToUI::set_ui_signal() after enqueueing updates so the UI can poll safely
• Skip or defer requests for rooms not yet present in the joined-rooms map
• Abort and clean up background tasks when room state is dropped to avoid leaks
• Use a lock-free queue (SegQueue) for very high-frequency room-list updates

Example use cases

• Paginating a room timeline: submit PaginateRoomTimeline and spawn a task that sends PaginationRunning / PaginationIdle updates
• Joining a room: submit JoinRoom, let the worker spawn a task to join, then post a JoinRoomResultAction back to the UI
• Real-time timeline: spawn_timeline_subscriber subscribes to timeline.stream and emits NewItems and diffs to the view
• Media fetch: request FetchMedia with a callback and update per-message timeline entry on completion
• Unread counts and room list changes: enqueue RoomsListUpdate entries and signal UI to process them