Why deferred imports fix circular dependency

Question

Accepted Answer

Top-level import statements execute at module load time - exactly when circular dependencies cause problems. A function body, on the other hand, executes only when the function is called. By then, the interpreter has finished loading all top-level code in every module, so all names exist. The sys.modules cache still applies - Python does not reload the module. It just performs the attribute lookup to bind the name in the function's local scope, and that lookup succeeds because the module is now complete. The minor downside is a tiny overhead on each function call (a cache lookup). In practice that cost is negligible, and this pattern is a well-accepted idiom for breaking cycles in Python codebases.