turborepo

Monorepo build caching speeds up development by using content-aware hashing to skip unchanged tasks, restoring previous outputs from local or remote cache instead of recomputing them, so incremental builds finish quickly.

Task graph orchestration runs dependent tasks in the correct order across many packages and apps. It maps dependencies declared in your pipeline configuration and executes builds, tests, and lints in parallel where safe, ensuring upstream packages finish before downstream consumers start.

Configuring pipeline dependencies requires declaring task relationships and input/output globs in turbo.json. You specify which tasks depend on others, what files trigger cache invalidation, and what artifacts get cached, enabling correct selective execution across pnpm workspace packages.

Remote caching works across CI and local environments by storing build artifacts in a shared remote store. When any machine runs a task whose inputs match a prior run, it fetches the cached output instead of rebuilding, preventing duplicate computation across the team.

Build caching is needed when a microservices monorepo grows large enough that rebuilding and retesting all services on every change becomes slow. Dependency-aware caching and selective rebuilds eliminate redundant task execution, reducing CI time and improving local development feedback loops.

Content-aware hashing requires accurate declaration of task inputs and outputs in turbo.json. If input globs omit files that affect a task or outputs are underspecified, the cache may serve stale results. Tasks with hidden dependencies or non-deterministic side effects can also produce incorrect cache hits.