Why does Go forbid circular dependencies, and how do you resolve them?

WHAT THIS TESTS: This question probes whether you understand the architectural rationale behind Go's package system, not just the mechanical fact that import cycles fail. The interviewer wants to see that you value directed acyclic graphs for build performance and separation of concerns, and that you can refactor code to restore a clean dependency structure.

A GOOD ANSWER COVERS: First, explain why the compiler forbids cycles. Go requires package dependencies to form a DAG because that enables fast incremental compilation; the compiler can skip unmodified packages and walk the graph in a single pass. It also enforces cleaner architecture by making coupling explicit. Second, describe at least two refactoring patterns. The first pattern is relocating logic to the lower-level package. If package A and B import each other, ask which concern is more fundamental and move the dependent code there. The second pattern is merging packages. When two packages are so tightly coupled that separating them creates constant back-and-forth, combining them into one package removes the cycle outright. The third pattern, dependency injection via interfaces, is especially useful for layer violations. If an abstract domain package depends on a concrete storage package, define a storage interface in the domain layer and inject the concrete implementation at startup. This inverts the dependency so the concrete layer depends on the abstract interface.

COMMON WRONG ANSWERS: A major red flag is suggesting compiler flags, build tags, or single-package workarounds to trick the toolchain. Another weak response is claiming cycles are forbidden only to make the compiler easier to write; the correct answer ties the restriction to incremental builds and architectural clarity. Candidates who can only describe one fix, such as merging everything into one package, signal shallow experience with larger codebases.

LIKELY FOLLOW-UPS: The interviewer may ask how you would detect a cycle in a large repository, how dependency injection affects testability, or whether Go modules changed anything about cycle detection. They might also ask you to sketch a specific refactor for a hypothetical service with domain, storage, and API layers.

ONE CONCRETE EXAMPLE: Imagine an e-commerce app where a product package imports inventory to check stock, inventory imports order to subtract pending orders, and order imports product to sum prices. To break the cycle, move the InStock method from product into the inventory package, since stock status is inventory's concern, not the product's. If inventory and product remain too entangled, merge them into a single goods package. Finally, if inventory directly calls a concrete inventory_storage package, introduce a storage interface inside inventory and inject the storage implementation during initialization so the domain layer no longer imports the concrete layer.

Source: appliedgo.net