Dart Concurrency and Isolates

Priority: P1 (HIGH)

Core Concepts

Dart uses a single-threaded event loop. All Flutter code runs on the Main Isolate by default. Blocking it causes jank.

• async/await: For non-blocking I/O (network, file). The event loop continues while waiting.
• Isolates: Dart's lightweight threads with isolated memory. Communicate via message passing only.

Decision Matrix

Condition	Approach
I/O bound (HTTP, database)	async/await on Main Isolate
CPU-bound, < 16ms	async/await on Main Isolate
CPU-bound, one-off heavy task	Isolate.run()
Continuous background processing	Isolate.spawn() with ports

Workflow: Offloading Heavy Computation

• 1. Identify the CPU-bound operation blocking the UI.
• 2. Extract computation into a standalone top-level or static function.
• 3. Ensure the function accepts exactly one argument (Isolate constraint).
• 4. Call Isolate.run(() => myFunction(data)).
• 5. await the result on the Main Isolate.

Workflow: Long-Lived Worker Isolate

• 1. Create a ReceivePort on the Main Isolate.
• 2. Spawn worker with Isolate.spawn(entryPoint, mainPort.sendPort).
• 3. In worker, create its own ReceivePort and send its SendPort back.
• 4. Store worker's SendPort for bidirectional communication.
• 5. Close ports and kill isolate on dispose.

See examples for complete code.

Anti-Patterns

• No JSON parsing on Main Isolate: Large JSON decoding (>1MB) blocks frames. Use Isolate.run.
• No shared mutable state: Isolates cannot share memory. Pass data via messages.
• No FutureBuilder in build without caching: FutureBuilder re-fires on every rebuild if the future is created inline.

Verification

• No frame drops during heavy computation (check with DevTools).
• Worker isolates are disposed when no longer needed.
• flutter test passes.

References

• Isolate Examples

Canonical response anchors

When this skill applies, preserve the following domain terminology or equivalent concrete examples in the answer when relevant:

• Dart Concurrency and Isolates