Recover and retain existing observation history safely #74

New Issue

2026-09-13T21:03:13Z

xavierk commented

2026-09-13 21:03:13 +00:00

Parent

Implement Fenris TUI polish and hourly history

What to build

Users upgrading with surviving raw samples or older summaries get all recoverable history without invented precision or lost evidence. Repeated repair and retention remain safe and their outcomes are visible to readers. Covers historical repair/retention in TPH-6 and legacy eligibility prerequisites for TPH-7.

Implement the approved companion contract and applicable TPH acceptance amendments linked from the parent. Preserve the frozen redesign, lifecycle/security boundaries, existing evidence thresholds and mathematics except explicit approved accounting amendments. Reuse existing seams; any necessary prefactoring precedes behavior changes within this slice. No release, deployment, or parent-issue changes.

Acceptance criteria

Normal collection, legacy import and recovery use the same evidence rules. Automatically derive only surviving supported evidence, transactionally and idempotently; preserve import markers, quarantine diagnostics, rename-after-commit and no-destructive-recreation guarantees.
Rerunning or interrupting repair produces no duplicated intervals or totals. Empty/incomplete reconstruction cannot replace valid older summaries; a failure preserves retryable evidence and is surfaced explicitly.
Prune ordinary raw samples after 14 days only after durable derivation; retain boundary anchors required by successor evidence. Keep intervals/hour/day history indefinitely; viewport ranges never affect retention.
Legacy day-only summaries retain actual represented precision and known eligibility. Missing metadata, hour detail and boundary shares remain unavailable, not reconstructed by interpolation or waiting; summaries and underlying intervals never double-count.
Read-only TUI/status distinguish insufficient evidence, repair failure and store fault, with explicit missing-detail facts and no reader-side repair writer.
Demonstrate migration then collection then reader consumption on stores containing surviving raw evidence, trusted old day-only data, mixed controller/period summaries and unfinished boundary anchors. Exercise retry, interruption, pruning, concurrent reader consistency and newer-schema refusal using existing migration/legacy/pruning seams.

Blocked by

Publish trustworthy first usage history

## Parent [Implement Fenris TUI polish and hourly history](https://git.bongbetic.com/xavierk/Fenris/issues/72) ## What to build Users upgrading with surviving raw samples or older summaries get all recoverable history without invented precision or lost evidence. Repeated repair and retention remain safe and their outcomes are visible to readers. Covers historical repair/retention in TPH-6 and legacy eligibility prerequisites for TPH-7. Implement the approved companion contract and applicable TPH acceptance amendments linked from the parent. Preserve the frozen redesign, lifecycle/security boundaries, existing evidence thresholds and mathematics except explicit approved accounting amendments. Reuse existing seams; any necessary prefactoring precedes behavior changes within this slice. No release, deployment, or parent-issue changes. ## Acceptance criteria - [ ] Normal collection, legacy import and recovery use the same evidence rules. Automatically derive only surviving supported evidence, transactionally and idempotently; preserve import markers, quarantine diagnostics, rename-after-commit and no-destructive-recreation guarantees. - [ ] Rerunning or interrupting repair produces no duplicated intervals or totals. Empty/incomplete reconstruction cannot replace valid older summaries; a failure preserves retryable evidence and is surfaced explicitly. - [ ] Prune ordinary raw samples after 14 days only after durable derivation; retain boundary anchors required by successor evidence. Keep intervals/hour/day history indefinitely; viewport ranges never affect retention. - [ ] Legacy day-only summaries retain actual represented precision and known eligibility. Missing metadata, hour detail and boundary shares remain unavailable, not reconstructed by interpolation or waiting; summaries and underlying intervals never double-count. - [ ] Read-only TUI/status distinguish insufficient evidence, repair failure and store fault, with explicit missing-detail facts and no reader-side repair writer. - [ ] Demonstrate migration then collection then reader consumption on stores containing surviving raw evidence, trusted old day-only data, mixed controller/period summaries and unfinished boundary anchors. Exercise retry, interruption, pruning, concurrent reader consistency and newer-schema refusal using existing migration/legacy/pruning seams. ## Blocked by - [Publish trustworthy first usage history](https://git.bongbetic.com/xavierk/Fenris/issues/73)

xavierk added the ready-for-agent label 2026-09-13 21:03:13 +00:00

xavierk referenced this issue

2026-09-13 21:03:21 +00:00

Browse local daily writes and hourly detail #75

xavierk referenced this issue

2026-09-13 21:03:25 +00:00

Show evidence-anchored projection rates #76

xavierk added a new dependency 2026-09-13 21:04:12 +00:00

#73 Publish trustworthy first usage history

xavierk added a new dependency 2026-09-13 21:04:14 +00:00

#75 Browse local daily writes and hourly detail

xavierk added a new dependency 2026-09-13 21:04:16 +00:00