kadirozde commented on PR #2428: URL: https://github.com/apache/phoenix/pull/2428#issuecomment-5688268524
> @kadirozde Would it be possible for you to go through the Copilot findings and click 'Resolve Conversation' button for everything you think does not apply? @apurtell Thanks — I’ve gone through your conditional-approval items and the Copilot findings. Conditional approval items V2 default off for first ship — done. DEFAULT_WHERE_OPTIMIZER_V2_ENABLED is now false (opt-in via phoenix.where.optimizer.v2.enabled=true). Docs updated accordingly. I’ll also update the PR description, which still incorrectly says “default on.” collapseToSingleBoundingRange — not removed. Copilot correctly flagged that compounds.subList(0, bound) narrows the union and can drop matching OR branches (false negatives a residual cannot recover). That helper is now used as the over-budget path: collapse to a covering envelope and keep predicates in the residual via the new approximated plumbing. So it is no longer dead code. DESC separator in buildRvcEqualitySpace — done. Variable-width packed RVC-IN literals now stop on both SEPARATOR_BYTE and DESC_SEPARATOR_BYTE. KeyRange.hashCode memoization — done. Comment corrected (fields are not final; readFields mutates them), and init() clears cachedHashCode. Copilot correctness findings Cartesian bound (allowed = d + 1) — fixed to stop before the overflowing slot (allowed = d), with approximated=true so dropped-slot predicates stay in the residual. Residual after algebra/extractor approximation — KeySpaceList.isApproximated() / KeyRangeExtractor.Result.approximated / V2ScanBuilder.Result.approximated plumbed through; WhereOptimizerV2 does not strip visitor-consumed nodes when approximated. Point-lookup classifier accepting leading EVERYTHING — fixed; any unconstrained user dimension rejects the native point-lookup path. Docs claiming wrong default — fixed with the default-off change. I’ll click Resolve Conversation on the Copilot threads that are addressed by the above. -- 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]
