Approval requested for explicit onboarding closure re-evaluation lineage
Sales proposes ADR-0050 to close the post-abort policy re-evaluation gap in ADR-0048. The proposal is distinct from the published response-loss and lease-recovery proof: those cases resume the same immutable key, while this decision concerns explicit authorization and lineage for a fresh child operation after a settled predecessor.
Current strict v1 carries no general predecessor identity. Revenue permits a narrow replacement while an exact predecessor fence remains, but explicitly rejects unfenced post-abort replacement. Delivery requires a proven aborted predecessor for a new lesson operation. Sales will continue to leave unsupported cases pending until the owners accept and publish a compatible representation. No proposed endpoint, new selector, policy override or rollout is being implemented by this request.
Requested decisions
Revenue, please approve or amend the separately versioned additive command/read approach and explicit predecessor operation/fingerprint binding. Specify the durable settlement evidence that proves old admitted/provider/canonical work cannot commit, compare-and-set lineage-head rules, conflict codes, post-abort handling and migration of existing journals/fences. Confirm current financial eligibility and refund rules remain unchanged. A replacement must not be inferred from chronological recency or an absent fence.
Delivery, please approve or amend the child intent and exact release-predecessor linkage, lesson/lock/version serialization, old-generation event handling and partial-child abort behavior. With Revenue, specify authoritative proof when a prior intent never admitted a release; a missing read alone is not sufficient. Confirm completed outcomes and immutable receipts remain protected.
Platform, please review only authentication and event-envelope impact: whether existing code-selected closure capabilities and current scopes suffice for the proposed additive surfaces, with unchanged subjects and tenant/org reach. No new event or envelope field is proposed. If authority must change, identify the separate explicit policy decision before implementation.
Each domain should reply on this thread with its position. An approval of the ADR must include acknowledges: [ADR-0050]; objections and conditional implementation requirements should be explicit in the body and any structured commitment/gate fields. Revenue and Delivery own the final exact wire schemas and producer eligibility/settlement policy. No response is presumed approval.
Proposed consumer behavior and compatibility
Sales would preserve ordinary same-key Retry cancellation and add an explicit operator Re-evaluate cancellation action after current exact evidence qualifies the predecessor. Its journal would retain append-only generations and immutable old requests/receipts, with transactional protection against duplicate clicks and stale previews. Outreach remains paused, new obligations remain blocked, and Lost requires final current evidence for every linked obligation.
Prefer new strict protocol-version 1.1 re-evaluation endpoints while v1.0 commands, fingerprints, receipts and reads remain unchanged. The ADR names concrete candidate paths for review, not a settled contract. Acceptance must resolve how new lineage-aware receipts are read without sending incompatible shapes to old consumers. Additive storage, compatibility tests, producer settlement/race/crash proofs and a fresh complete Sales rehearsal precede activation. Existing production gates and broader ADR-0048 commitments remain open.
Revenue and Delivery must explicitly resolve mixed-version admission: the existing v1.0 narrow fenced-replacement path must advance the same authoritative lineage head and obey the same unique-child constraint as v1.1. Require real races in both winning orders, including migration and restart, with one committed child and unchanged historical receipts/fingerprints. If any legacy writer cannot participate, keep the new path disabled. Do not treat unchanged old wire syntax as proof that old and new writers cannot branch.