The collector still gathers the legacy review: {open, note} field — a count of open translation-review issues plus a config note. The render-lanes work (#3, built in #4) keyed the Human Review column on the verdict store instead and moved the legacy count into a tooltip, explicitly leaving this field's fate undecided. With #3 closed, that decision has no open tracker except this issue.
The field measures something the lanes don't: the pre-verdict, issue-based review backlog. Concretely, lecture-python-programming zh-cn has 24 review issues open since 2026-04-11, currently visible only in a tooltip.
Two ways to settle it:
- Retire the field with a follow-up schema bump — one review signal, one substrate (the verdict store), less collector surface; accepts that the issue-based backlog goes untracked here.
- Keep it and render it as a distinct "issue backlog" signal alongside the lanes (the audit's original §1 ask) — the campaign-era review work predates verdicts and still exists.
Decision only — no build until settled; any schema change lands with whichever package implements the outcome.
The collector still gathers the legacy
review: {open, note}field — a count of opentranslation-reviewissues plus a config note. The render-lanes work (#3, built in #4) keyed the Human Review column on the verdict store instead and moved the legacy count into a tooltip, explicitly leaving this field's fate undecided. With #3 closed, that decision has no open tracker except this issue.The field measures something the lanes don't: the pre-verdict, issue-based review backlog. Concretely, lecture-python-programming zh-cn has 24 review issues open since 2026-04-11, currently visible only in a tooltip.
Two ways to settle it:
Decision only — no build until settled; any schema change lands with whichever package implements the outcome.