perlookup

Perlookup is a term used in software architecture to describe a lifecycle in which a new object instance is created for every lookup or request, rather than reusing an existing instance. It is commonly associated with dependency injection and service locator patterns, where components may be resolved on demand rather than being injected as a single shared instance.

In practice, perlookup is typically implemented using a lightweight proxy or factory that, on each access, constructs and returns a fresh instance. This provides complete isolation between usages and guarantees that state from one consumer is not visible to another.

Similar concepts exist in several frameworks. In CDI, the dependent scope behaves like per-lookup: a new bean instance is created for each injection point or lookup and is short-lived. In Spring, prototype scope yields a new bean instance for every retrieval or injection.

Advantages and trade-offs: Per-lookup ensures fresh state and avoids unintended sharing, simplifies lifecycle management, and can improve correctness in certain scenarios. However, it imposes higher per-call overhead due to object construction and garbage collection, increases latency, and can complicate resource management if the created objects hold external resources.

When to use: It is appropriate for stateless or short-lived components, or when isolation between callers is essential. It is less suitable for expensive-to-create objects or high-throughput paths where caching or pooling would be preferable.

Implementation considerations: ensure thread-safety of factories or proxies, consider lazy initialization, document lifecycles, and consider combining with selective caching or pooling for expensive objects. Provide clear guidance for migrating from per-lookup to cached instances if performance becomes a concern.