Conserve local-day evidence #98

Closed
opened 2026-09-27 20:01:06 +00:00 by xavierk · 0 comments
Owner

Parent

Deepen observation-history publication, local-day evidence, and activity selection.

What to build

Give drive owners honest local-day read/write evidence through collection and the visible activity views. Derive known volume from compatible measured intervals inside recorded local boundaries, retain a local-midnight-spanning measurement once as shared local-day evidence, and show known amounts explicitly as incomplete when allocation is ambiguous or evidence is missing. Keep UTC projection evidence distinct.

Acceptance criteria

  • A measured read/write interval crossing a UTC-hour boundary but wholly inside one local day contributes its full measured amount to that day's known activity; include the reproduced 5,120,000-byte write case.
  • An actual measured interval crossing Asia/Kolkata local midnight produces one shared evidence item, not full known volume in both adjacent days; a synthetic 100-byte case demonstrates conservation independently of display rounding.
  • TUI daily totals and selected activity use the recorded local date and show known writes/reads and shared volume separately, with incomplete/so-far/zero/unavailable states that cannot be mistaken for a complete day. CLI/status facts agree where they display activity.
  • Verify 23-/25-hour local days, non-whole-hour offsets, timezone changes, repeated local clock labels, controller segments and monitoring periods; no fabricated interpolation or regression in established UTC projection rules.
  • Tests cross controlled acquisition and clock, temporary observation store and ordinary reader paths, with focused evidence-invariant checks only where necessary.

Blocked by

## Parent [Deepen observation-history publication, local-day evidence, and activity selection](https://git.bongbetic.com/xavierk/Fenris/issues/95). ## What to build Give drive owners honest local-day read/write evidence through collection and the visible activity views. Derive known volume from compatible measured intervals inside recorded local boundaries, retain a local-midnight-spanning measurement once as shared local-day evidence, and show known amounts explicitly as incomplete when allocation is ambiguous or evidence is missing. Keep UTC projection evidence distinct. ## Acceptance criteria - [ ] A measured read/write interval crossing a UTC-hour boundary but wholly inside one local day contributes its full measured amount to that day's known activity; include the reproduced 5,120,000-byte write case. - [ ] An actual measured interval crossing Asia/Kolkata local midnight produces one shared evidence item, not full known volume in both adjacent days; a synthetic 100-byte case demonstrates conservation independently of display rounding. - [ ] TUI daily totals and selected activity use the recorded local date and show known writes/reads and shared volume separately, with incomplete/so-far/zero/unavailable states that cannot be mistaken for a complete day. CLI/status facts agree where they display activity. - [ ] Verify 23-/25-hour local days, non-whole-hour offsets, timezone changes, repeated local clock labels, controller segments and monitoring periods; no fabricated interpolation or regression in established UTC projection rules. - [ ] Tests cross controlled acquisition and clock, temporary observation store and ordinary reader paths, with focused evidence-invariant checks only where necessary. ## Blocked by - [#97 Retain and recover pending publication](https://git.bongbetic.com/xavierk/Fenris/issues/97).
xavierk added the ready-for-agent label 2026-09-27 20:01:06 +00:00
xavierk added a new dependency 2026-09-27 20:01:07 +00:00
xavierk added a new dependency 2026-09-27 20:01:10 +00:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Reference: xavierk/Fenris#98