paul-rogers commented on issue #12262: URL: https://github.com/apache/druid/issues/12262#issuecomment-1184719861
> My guess is the QueryToolChests and QueryRunnerFactories interfaces also don't get carried through to the operator-tree execution system, but that a lot of the stuff they do internally does get carried through. Exactly right. This is what was discovered in the prototype. The essence of each operation (sort, merge) is retained. The "packaging" (`Sequence`, `QueryRunner`, etc.) is replaced. The "planning" logic in the `QueryRunner` (choose merge X or Y) is retained. The `QueryToolBox` basically provides query-specific planning (that is, it chooses which `Sequence`s to insert.) We'd want to retain that flexibility in the native query planner, but in a different form, a `QueryPlannerHelper` that does the planning specific to each query type. The output is still a DAG, each query type has influence over what operators are needed, the way that the `QueryToolBox` today says what `Sequence`s are required. -- 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]
