Run the v1.0.0 acceptance pass against a real collection #53

New issue

Open

opened 2026-07-28 04:59:59 +00:00 by claude-bot · 0 comments

claude-bot commented

2026-07-28 04:59:59 +00:00

Context

Every milestone has its own exit criteria, and CI verifies the suite on every push. What
none of that establishes is whether the finished product actually works on a real family
collection at real scale â€” which is the only question v1.0.0 genuinely turns on.

Scope

A full acceptance pass over the complete feature set against a real collection.

Implementation notes

Verify the acceptance criteria from all four phases in docs/circa-spec.md section 14,
as a checklist, against the built software.
Run against a real collection of meaningful size, not seeded fixtures. Real scans carry
the messy filenames, missing backs, odd EXIF, and genuine near-duplicates that fixtures
never reproduce â€” those are exactly the cases worth finding before declaring 1.0.
Walk the complete journey: scan, ingest, AI and OCR analysis, review, resolve duplicates,
propagate constraints, dispute and resolve with a second reviewer, back up, and export.
Verify the export output independently of Circa: check the manifest hashes, confirm EXIF
dates read correctly in a normal photo application, and confirm the folder structure
matches the spec.
Exercise restore from backup into a scratch environment, proving the backup is real.
Record anything found as issues rather than fixing in place, so the trail is preserved.
Measure how long reviewing a batch actually takes â€” if the workflow is too slow to be
used at collection scale, that is a v1.0.0 finding regardless of correctness.
Confirm the review history has been preserved intact throughout: no lost decisions, no
broken provenance chains, no missing audit events.

Done when

Every phase acceptance criterion is verified against the built software
The full journey completes on a real collection
Export output is verified independently of Circa
Restore from backup is proven in a scratch environment
Review history and audit trail are confirmed intact
Findings are recorded as issues and triaged before release

References

docs/circa-spec.md section 14 (phase acceptance criteria)
docs/circa-phase1-plan.md section 12

Depends on: every other milestone.

## Context Every milestone has its own exit criteria, and CI verifies the suite on every push. What none of that establishes is whether the finished product actually works on a real family collection at real scale â€” which is the only question v1.0.0 genuinely turns on. ## Scope A full acceptance pass over the complete feature set against a real collection. ## Implementation notes - Verify the acceptance criteria from all four phases in `docs/circa-spec.md` section 14, as a checklist, against the built software. - Run against a real collection of meaningful size, not seeded fixtures. Real scans carry the messy filenames, missing backs, odd EXIF, and genuine near-duplicates that fixtures never reproduce â€” those are exactly the cases worth finding before declaring 1.0. - Walk the complete journey: scan, ingest, AI and OCR analysis, review, resolve duplicates, propagate constraints, dispute and resolve with a second reviewer, back up, and export. - Verify the export output independently of Circa: check the manifest hashes, confirm EXIF dates read correctly in a normal photo application, and confirm the folder structure matches the spec. - Exercise restore from backup into a scratch environment, proving the backup is real. - Record anything found as issues rather than fixing in place, so the trail is preserved. - Measure how long reviewing a batch actually takes â€” if the workflow is too slow to be used at collection scale, that is a v1.0.0 finding regardless of correctness. - Confirm the review history has been preserved intact throughout: no lost decisions, no broken provenance chains, no missing audit events. ## Done when - [ ] Every phase acceptance criterion is verified against the built software - [ ] The full journey completes on a real collection - [ ] Export output is verified independently of Circa - [ ] Restore from backup is proven in a scratch environment - [ ] Review history and audit trail are confirmed intact - [ ] Findings are recorded as issues and triaged before release ## References - `docs/circa-spec.md` section 14 (phase acceptance criteria) - `docs/circa-phase1-plan.md` section 12 Depends on: every other milestone.

claude-bot added this to the v1.0.0 milestone

2026-07-28 04:59:59 +00:00

claude-bot added the

testing

label

2026-07-28 04:59:59 +00:00