Filed by the domain:ui execution seat (session session_012u2pRjcqAYtoEjgr3wwhnK), surfaced while reviewing #5671 (#4972).
Measured
packages/types/src/index.ts:912 publishes:
ComponentInputasPluginComponentInput,
Repo-wide, excluding node_modules/ and dist/, that is the only occurrence of the name — verified against origin/main. Nothing imports it. It is exported and never pulled.
Why now
Before #4972 the alias at least pointed at a different declaration: plugin-scope.ts declared its own 9-key ComponentInput, so PluginComponentInput named a genuinely distinct type. #5671 converges that declaration onto base.ts, so once it lands the alias is a published name duplicating another published name in the same entry point, with no consumer on either.
That is not a regression introduced by #5671 — it is a pre-existing zero-consumer export whose last remaining justification #5671 removes.
Ask
Retirement candidate under implementation-first / startup scope discipline.
Removing a published export is a surface change and needs the normal treatment: a bump judgment under the repo's policy, and the ADR-0087 registries if the removal is spec-facing.
Filing rather than doing, for one reason worth stating: "no importer in this repo" and "no importer" are different claims, and only the second one licenses deleting an export from a published package. This card holds the first; someone has to decide whether that is enough here.
Filed unassigned.
Filed by the
domain:uiexecution seat (sessionsession_012u2pRjcqAYtoEjgr3wwhnK), surfaced while reviewing #5671 (#4972).Measured
packages/types/src/index.ts:912publishes:Repo-wide, excluding
node_modules/anddist/, that is the only occurrence of the name — verified againstorigin/main. Nothing imports it. It is exported and never pulled.Why now
Before #4972 the alias at least pointed at a different declaration:
plugin-scope.tsdeclared its own 9-keyComponentInput, soPluginComponentInputnamed a genuinely distinct type. #5671 converges that declaration ontobase.ts, so once it lands the alias is a published name duplicating another published name in the same entry point, with no consumer on either.That is not a regression introduced by #5671 — it is a pre-existing zero-consumer export whose last remaining justification #5671 removes.
Ask
Retirement candidate under implementation-first / startup scope discipline.
Removing a published export is a surface change and needs the normal treatment: a bump judgment under the repo's policy, and the ADR-0087 registries if the removal is spec-facing.
Filing rather than doing, for one reason worth stating: "no importer in this repo" and "no importer" are different claims, and only the second one licenses deleting an export from a published package. This card holds the first; someone has to decide whether that is enough here.
Filed unassigned.