wenzhenghu commented on PR #65126: URL: https://github.com/apache/doris/pull/65126#issuecomment-5091795706
Review follow-up summary, audited against the current head `2266c7ef964fb4b63c3c786c9f562468c89c558f`: - I reviewed all 25 inline review threads across the 39 submitted reviews, together with the follow-up findings raised in PR comments. All 25 inline threads are currently resolved. - The correctness and regression findings were addressed in code, including cache-only replay fallback, canonical invalidation, cold-event generation fencing, delegated-session point lookup, per-key object/ID publication ordering, named replay with disabled object caches, and lazy `StripeState` allocation. - Findings that describe intentional eventual-consistency boundaries were handled by clarifying the PR description and code comments. In particular, old-generation work must not publish, while a refresh that starts after the generation bump or overlaps invalidation may still publish under the accepted Caffeine semantics. The PR does not claim full global invalidation linearizability. - Larger or baseline follow-ups remain tracked independently: #65750 for the remaining HMS event consistency gaps, #65779 for names-snapshot and selective-invalidation performance, #65792 for unified cache statistics, #66001 for global invalidation ordering, and #66024 for the pre-existing `MetaInfoAction` privilege-filtering issue. I also compared the latest PR head with the last metadata-cache implementation commit. The final merge from `master` only changed `fe/pom.xml` and `fe/fe-core/pom.xml`; it did not touch the metadata-cache production code or related tests. Static inspection found no regression or weakening of the fixes accepted in the resolved review threads. This was a static regression audit only; no UT or checkstyle run is claimed for the latest head in this follow-up. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
