XianReallyHot-ZZH commented on PR #13762: URL: https://github.com/apache/ignite/pull/13762#issuecomment-6874305277
@chesnokoff implemented your idea in a stacked commit (the first solution's history is preserved — reverting it recovers the statusCode version, as you asked): - Dropped the feature-gated `statusCode` field, the `CoreFeatureRegistry` registration and the wire-compatibility baseline update — the PR no longer touches the wire format at all: `GridQueryFailResponse` is field-identical to master. - Added `SQL_ERROR = 3` to the existing `failCode`; the constructor sets it when the error carries an `IgniteSQLException` in its cause chain (checked after the existing CANCELLED/RETRY branches, so those keep their current priority). - The reducer reconstructs `CacheException(mapperFailedMsg, new IgniteSQLException(errMsg, UNKNOWN))`; the outer message is byte-identical to master. Old nodes are unaffected: master's `fail()` dispatches unknown fail codes to the final `else` (causeless `CacheException`), so an old reducer receiving `SQL_ERROR` behaves exactly as today. Two implicit choices worth stating: a `QueryRetryException`/`QueryCancelledException` wrapping an `IgniteSQLException` is still classified RETRY/CANCELLED (pre-existing priority kept), and the reconstructed cause's message is the map node's top-level `errMsg` per your suggestion, not the nested exception's own message. No information is lost vs the previous approach — the ticket's scenario carried `UNKNOWN` on the wire anyway; the existing test (unchanged) still passes 2/2. Also rebased on latest master; the `CoreFeatureRegistry` conflict is gone along with the registration itself. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
