HyukjinKwon commented on PR #58604: URL: https://github.com/apache/spark/pull/58604#issuecomment-5594965368
Thanks @sunchao for the careful post-merge review. I've addressed all three findings in a follow-up: #58642. 1. **Enforce across both analyzer paths.** `HybridAnalyzer.apply` now routes a restricted-mode session to the fixed-point analyzer (where `checkRestrictedMode` runs) in every single-pass mode, including the fully-enabled mode that skips `ResolverGuard`. Test: enforcement holds with `spark.sql.analyzer.singlePassResolver.enabled`/`enabledTentatively`. 2. **Reject before class initialization.** `CallMethodViaReflection.checkInputDataTypes` short-circuits in restricted mode, so a call that is going to be rejected never resolves the class via `Utils.classForName` (no static initializer). Test uses a fixture whose initializer sets a system property and asserts it stays unset. 3. **Traverse each subquery once.** The `innerChildren` recursion now skips plans that are already the node's subqueries (reached via `SubqueryExpression`), keeping validation linear; added coverage for a feature nested in a scalar subquery and for deeply nested subqueries with no restricted feature. All changes are scoped to restricted mode, so default behavior is unchanged. -- 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]
