Share State Across Dart Isolates Without Losing Your Mind: Enter shared_map

Dart’s concurrent programming model relies on Isolates, which do not share memory. Each isolate maintains its own private heap and single-threaded event loop. While this design prevents data races, deadlocks, and threading issues, it also prevents isolates from sharing data. When you need to share data, you typically face two suboptimal solutions: serializing and copying data structures or building a custom message-passing server using ReceivePort and SendPort.

Shared_map offers a superior alternative. Developed by experienced Dart engineer Graciliano M. Passos, shared_map provides a synchronized Map data structure that can be shared across Dart isolates. Key benefits of shared_map include zero dependencies, flawless quality, universal platform support, and familiar async Map methods for easy integration.

The library employs the reference pattern, where a lightweight, serializable token is generated and transferred between isolates. This token allows the worker isolate to reconstruct a proxy instance of the shared Map, enabling seamless reads, writes, and mutations while ensuring synchronization across all isolates. In a practical example, two background isolates read and write to a shared cache, demonstrating how shared_map simplifies cross-isolate communication compared to traditional messaging approaches.