atomr-view 0.3.3

Backend-agnostic UI/visualisation system for atomr — actor-driven scenes across native, web, and mobile.

atomr-view — Heavy Client UI for Federated atomr Systems

Status: Implementation Phase Complete / Testing Suite Active

atomr-view is an actor-based view management system designed for heavy client applications (Desktop, Mobile, Wasm) that integrate with federated atomr clusters. It treats the UI rendering surface as a supervised peer with network-shaped failure modes, using the actor model to bridge the gap between remote state authority and local pixels.

Mental Model

In atomr-view, the UI is not a separate layer bridged by callbacks; it is a subgraph of the local ActorSystem.

• WindowActor: Owns an OS window, its declarative scene, and child regions.
• RegionActor: Manages a stateful subtree of the UI, typically subscribed to a remote sharded entity or a local streaming feed.
• UiBridge: An asynchronous, backpressured conduit between the actor system and the rendering thread (e.g., winit + wgpu).
• SceneDescription: A declarative, keyed tree of semantic UI intent (Button, Text, etc.) that the backend reconciles against native widgets.

Key Features

• Location Transparency: UI actors interact with cluster-sharded actors using the same tell/ask/watch primitives as local actors.
• Backpressure Discipline: The UiBridge uses conflation for state updates, ensuring the UI thread only ever processes the freshest state available.
• Declarative Reconciliation: Keyed tree-diffing computes minimal patches, preserving local UI state (like cursor position) across updates.
• Multi-Backend Strategy: Supports winit + wgpu + egui for desktop, with architectural hooks for Wasm (DOM) and Mobile (SwiftUI/Compose).
• Python-First API: Full integration via PyO3, allowing UI actors and complex scene manipulation to be authored in Python.

Platform Backends

atomr-view supports multiple rendering surfaces via the UiBackend strategy:

• Desktop (winit + wgpu + egui): The default reference implementation for high-performance desktop apps.
• Bevy (BevyBackend): Integrates the actor system with Bevy's ECS. Use the ActorBacked component to mirror ECS entities with long-lived actors for persistence and netcode.
• Wasm (WebDomBackend): Targets browser environments with a cooperative single-threaded scheduler that yields to requestAnimationFrame.
• Mobile (NativeShellBackend): Bridges to SwiftUI (iOS) and Jetpack Compose (Android) via uniffi, maintaining native UI performance while keeping business logic in Rust actors.

Quick Start

Rust

use atomr_view_core::prelude::*;
use atomr_core::prelude::*;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let sys = ActorSystem::create("App", Config::reference()).await?;
    
    // 1. Initialize the bridge and backend
    let (cmd_tx, cmd_rx) = mpsc::channel(100);
    let (evt_tx, evt_rx) = mpsc::channel(100);
    let bridge = UiBridge { cmd_rx, evt_tx };
    
    // 2. Start the bridge actor
    let bridge_actor = sys.actor_of(Props::create(move || {
        UiBridgeActor::new(cmd_tx.clone(), evt_rx)
    }), "ui-bridge")?;

    // 3. Define a window actor
    let main_window = sys.actor_of(Props::create(move || {
        WindowActor::new("main", bridge_actor.clone())
    }), "main-window")?;

    // 4. Run the backend (blocks main thread)
    WinitWgpuBackend.run(bridge);
    
    Ok(())
}

Python

import atomr_view as view

ui = view.PyUiSystem()
ui.create_window("main", "My Federated App")

# Update the scene declaratively
root = view.SceneNode(
    key=view.SceneKey(),
    kind=view.NodeKind.Container(),
    properties=view.PropertyMap({"layout": view.PropertyValue.String("vertical")}),
    children=[
        view.SceneNode(
            key=view.SceneKey(),
            kind=view.NodeKind.Text(),
            properties=view.PropertyMap({"text": view.PropertyValue.String("Hello from Python!")}),
            children=[]
        )
    ]
)
ui.set_scene("main", view.SceneDescription(root))

Implementation Status

• Core Foundation: UiBridge, SceneDescription, keyed diffing engine.
• Desktop Backend: winit + wgpu + egui reference implementation.
• Python Bindings: PyO3-based bridge with full scene manipulation support.
• Testing: Property-based tests for diffing invariants and async integration tests.
• Bevy Integration: BevyBackend bridging actors with ECS entities.
• Wasm Support: WebDomBackend with cooperative cooperative scheduler hooks.
• Mobile Flow: NativeShellBackend with uniffi bindings for native shells.

Project Structure

• crates/atomr-view-core: Core actor logic and scene engine.
• crates/atomr-view-backends: Swappable rendering strategies.
• crates/atomr-view-python: PyO3 integration.
• ai-skills/: Skills for AI assistants working with atomr-view.

License

Apache-2.0