EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·22d ago·5 min read
Deterministic Core, Non-Deterministic Shell
TL;DR
Gary Bernhardt's concept of Functional Core, Imperative Shell divides software architecture into a purely functional core for business logic and an imperative shell for state management and IO. This separation enhances testing and determinism in applications.
✦ Why It Matters
Consider adopting the Functional Core, Imperative Shell architecture to enhance your application's testability and reliability.
Key Takeaways
- Implement a Functional Core to isolate business logic from IO operations.
- Use an Imperative Shell to manage state and external interactions.
- Enhance testability by keeping the core free from side effects.
- Adopt this architecture to improve software reliability and maintainability.
Full Summary
Functional Core, Imperative Shell architecture, introduced by Gary Bernhardt, categorizes code into two distinct parts: the Functional Core and the Imperative Shell. The Functional Core is purely functional, meaning it does not perform input/output (IO) operations or modify state destructively, focusing solely on business logic.
In contrast, the Imperative Shell handles state management, interacts with external systems, and performs IO operations, such as database queries or GUI updates. This separation allows for more predictable and testable code, as the core remains isolated from external influences.
By querying the core and processing its outputs, the shell can effectively manage interactions with the outside world. This architecture not only clarifies the roles of different components but also improves the reliability of software systems.