Convergence Milestone Releases (C0-C3)
We introduced explicit convergence milestones so developers have clear, recommended jump-off points for new state work.
Convergence states are not just another node in the graph. They are curated checkpoints where multiple concerns have been intentionally composed and validated.
Why We Added Convergence States
- Keep transitional states focused and easy to reason about.
- Provide stable branch points for new enhancements.
- Avoid repeatedly re-solving baseline architecture choices in each new state.
Current Convergence Levels
C0:003-containerized-compose-runtimeC1:006-observability-lgtm-composeC2:008-order-management-matcherC3:011-platform-convergence-c3
Dotted-Line Inheritance
Convergence states can declare dotted-line lineage for documentation and learning context, while keeping a single publish parent for branch ancestry.
That lets us carry forward improvements into the best “next base” without creating ambiguous multi-parent publish history.
Convergence View
Contribution Guidance
- For most new features, start from the nearest convergence state that already contains needed prerequisites.
- If a convergence state is changed, document explicit rationale.
- Use dotted-line lineage only when it clarifies learning inheritance, not branch ancestry.
Decision + Navigation Links
- ADR: ADR-008 Convergence State Model
- Visual learning graphs: /docs/learning-paths
- State catalog: /docs/spec-kit/state-docs