Hub-and-Spoke SDLC State Machine

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

Design, implementation and review roles share one coordinator-owned phase. Teaching states are designed→implemented→reviewed, with evidence bound to revision-a. A completed label without artifacts or from an old revision cannot advance it.

Mechanism

Define transition inputs/checks/owners and require artifacts, revision, actual results and open items. The coordinator reads/verifies current-candidate evidence before advancing. Separate completed, unavailable, disabled, skipped and failed. Task termination is not review approval. Use owned/atomic shared-state updates.

Bad example

Let all roles change state and advance implemented→reviewed on any completed label, ignoring old versions, missing artifacts or review dimensions.

Good example

The teaching coordinator owns phase state. Roles provide artifact_path, revision, observed results and unresolved_items. Verify revision-a and required checks before advancement. Old/incomplete/failed review cannot satisfy reviewed. Keep unavailable/disabled/skipped distinct. Update under ownership/atomic comparison and preserve unfinished conditions in handoff.

Why the change matters

Central state provides shared position while evidence gates distinguish ended work from accepted phases. Revision/dimension visibility prevents stale or absent reviews satisfying transitions.

Observable expectation

Missing review artifacts leave implemented; matching complete review permits reviewed. Revision-b changes invalidate automatic use of a results.

Inspect readback before transitions and no skipped/unavailable converted to success. No real coordinator runs here.

Limits

States/phase order are teaching choices, not universal source tooling. Multiple writers need enforced consistency. Bookkeeping grants no authority and does not establish all-product coverage.