Release Convergence
The state in which a release candidate's implementation is consistent with the governed operational intent established for it — what was built matches what was governed.
Definition
Release convergence is the state in which a release candidate’s implementation is consistent with the governed operational intent that was established for it. A release has converged when what was built matches what was governed.
Convergence is established through convergence validation — the process of checking that the release candidate meets the governed constraints documented in the operational lineage.
Convergence vs. Correctness
Release convergence is not the same as software correctness. A test suite validates correctness — that the software behaves as specified by its tests. Convergence validation confirms that the software was built under the governed constraints that were established for it.
A release can be correct and not converged: all tests pass, but the implementation used an unapproved dependency version, or was built in a session that bypassed a governance checkpoint.
A release can be converged and still have bugs: the implementation followed all governed constraints, but there is a logic error that the tests did not catch.
Both properties matter. Correctness and convergence are complementary, not substitutes.
What Convergence Covers
Convergence validation in Yanzi checks that:
- The implementation is consistent with the intent artifacts in the operational lineage
- No active Rule constraints were violated during development
- All required governance checkpoints were passed
- The Role authority boundaries were observed throughout
Convergence does not check: test coverage, performance metrics, security scan results, or deployment readiness. These remain in the domain of traditional CI/CD pipelines.
The Convergence Point
Release convergence is determined at a specific point in time — when certification is run. A release that converges at certification time may diverge from new governance constraints established afterward. This is expected and documented in the canonical lineage: the convergence was valid relative to the corpus state at certification time.
This is why the corpus hash is recorded at certification — it anchors the convergence claim to a specific, reproducible corpus state.