Hugo Pipes fingerprinting and CI/CD cache invalidation

Asset Pipeline Consistency

Hugo Pipes utilizes a content-based hashing mechanism via the fingerprint function to enable long-term caching of CSS and JS assets. In local development, the build process relies on filesystem modification timestamps to determine when an asset requires re-processing.

When deploying via CI/CD pipelines, the environment often starts with a clean slate or utilizes external caching layers. This introduces uncertainty regarding how Hugo determines if a cached asset is stale when the source file content remains identical but the build environment's metadata differs.

Does the fingerprint function rely solely on the file's checksum, or does the underlying Hugo Pipes cache check filesystem timestamps before calculating the hash? How does this behavior change when the site is built in a non-persistent CI environment compared to a local development server?