Context
A revised document may be opened as a new Markover review after an agent has completed the preceding feedback round. Markover currently treats those reviews as independent snapshots.
This is a follow-up to #123 and carries forward only the deferred revision-lineage questions from PR #2. Revised and Done lifecycle states do not depend on this work.
Outcome
Determine whether connecting successive reviews provides enough practical value to justify the identity, persistence, interaction, and presentation complexity. Implementation should follow only if the evaluation identifies a clear user need and a small trustworthy model.
Questions to resolve
- Which real workflows benefit from lineage beyond lifecycle state, document path, Git metadata, and pull-request association?
- Is a simple agent-declared predecessor relation sufficient, or would branching and merging be required?
- How should reviews of renamed, moved, copied, or independently modified documents behave?
- Can repository and project-relative path metadata identify candidates without implying ancestry?
- What atomic storage model prevents one-sided links, forks, cycles, or duplicate ordinals?
- Are labels such as v1 and v2 useful, or do they misleadingly imply a canonical document version?
- How should lineage appear in Inbox, Projects, history, and agent-facing JSON without making current work harder to scan?
Constraints
Deliberate deferral
Revisit this only after Revised and Done states have been used enough to provide evidence that lineage solves a meaningful remaining problem. The default acceptable outcome is to retain independent reviews.
Context
A revised document may be opened as a new Markover review after an agent has completed the preceding feedback round. Markover currently treats those reviews as independent snapshots.
This is a follow-up to #123 and carries forward only the deferred revision-lineage questions from PR #2. Revised and Done lifecycle states do not depend on this work.
Outcome
Determine whether connecting successive reviews provides enough practical value to justify the identity, persistence, interaction, and presentation complexity. Implementation should follow only if the evaluation identifies a clear user need and a small trustworthy model.
Questions to resolve
Constraints
Deliberate deferral
Revisit this only after Revised and Done states have been used enough to provide evidence that lineage solves a meaningful remaining problem. The default acceptable outcome is to retain independent reviews.