Hi Huanbing, Thank you for your interest and insightful question about the TraversalState concept. You've identified exactly the right challenge with the custom executor approach.
However, as Ashutosh suggested, let's keep this thread focused on the current PGQ patch. Your question about DFS backtracking and PlanState management deserves a thorough discussion, which we can have once the current work is committed. I'm looking forward to discussing VLE implementation details with you and the community then. Best regards, Henson
