Problem
A manifest renames entry whose from matches no active rule is silently ignored at materialization (gate phase-26 follow-up, at archive/pre-github). A user who mistypes a rename target, or whose target rule left the active set after a method remove/migrate, gets no signal — the overlay entry is simply dead. The Phase-26 note flagged "a validate warning surfacing dead rename entries" as a possible later nicety.
Direction
Core, read-only: validate emits a non-blocking warning for each renames entry that resolves to nothing against the active materialized set — mirroring the existing policy.dangling warning for a policy naming an inactive method (D23/D25). Additive: a new advisory diagnostic, no exit-3, no schema change. Touches P4.
Forks
- Do nothing — dead entries are harmless (ignored, not misapplied); the cost is only a silent typo.
- Refuse at
manifest rename add (usage error) instead of warning at validate — but the pack may legitimately be inactive-but-forthcoming (the forward-looking case spec 04 §4.3 supports), so a hard refuse would break that; a warning is the right register.
- Fold into the rename-kind-extension issue — kept separate because this applies to today's rule renaming, independent of adding new kinds.
Problem
A manifest
renamesentry whosefrommatches no active rule is silently ignored at materialization (gate phase-26 follow-up, atarchive/pre-github). A user who mistypes a rename target, or whose target rule left the active set after amethod remove/migrate, gets no signal — the overlay entry is simply dead. The Phase-26 note flagged "avalidatewarning surfacing dead rename entries" as a possible later nicety.Direction
Core, read-only:
validateemits a non-blockingwarningfor eachrenamesentry that resolves to nothing against the active materialized set — mirroring the existingpolicy.danglingwarning for a policy naming an inactive method (D23/D25). Additive: a new advisory diagnostic, no exit-3, no schema change. Touches P4.Forks
manifest rename add(usage error) instead of warning atvalidate— but the pack may legitimately be inactive-but-forthcoming (the forward-looking case spec 04 §4.3 supports), so a hard refuse would break that; a warning is the right register.