Copilot commented on code in PR #11993: URL: https://github.com/apache/gluten/pull/11993#discussion_r3425661887
########## backends-velox/src/main/scala/org/apache/gluten/execution/VeloxBroadcastBuildSideCache.scala: ########## @@ -37,6 +38,10 @@ case class BroadcastHashTable(pointer: Long, relation: BuildSideRelation) * * The complicated part is due to reuse exchange, where multiple BHJ IDs correspond to a * `BuildSideRelation`. + * + * This implementation supports two modes: + * 1. Driver-side build (new): Hash table is built and serialized on driver, then broadcast 2. + * Executor-side build (legacy): Each executor builds its own hash table from broadcast data Review Comment: Comment typo: "then broadcast 2." reads like an accidental artifact and is confusing in a high-level description of the cache modes. ########## ep/build-velox/src/get-velox.sh: ########## @@ -17,9 +17,9 @@ set -exu CURRENT_DIR=$(cd "$(dirname "$BASH_SOURCE")"; pwd) -VELOX_REPO=https://github.com/IBM/velox.git -VELOX_BRANCH=dft-2026_06_06 -VELOX_ENHANCED_BRANCH=ibm-2026_06_06 +VELOX_REPO=https://github.com/JkSelf/velox.git +VELOX_BRANCH=dft-2026_06_06-serialize-hashtable +VELOX_ENHANCED_BRANCH=ibm-2026_06_06-serialize-hashtable Review Comment: The default `VELOX_REPO` now points to a personal fork (`github.com/JkSelf/velox.git`). This makes builds non-reproducible and can break CI or downstream users if the fork/branches are rewritten or removed. Prefer an official upstream (e.g., IBM/velox) and pin to a commit hash (or keep the fork only behind an explicit `--velox_repo` override). -- 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]
