Hi, i got your point basically you wanted to make this logic to be useful for normal join also. But for the same thing i had raised a discussion before, you can check here. <http://apache-carbondata-dev-mailing-list-archive.1130556.n5.nabble.com/DISCUSSION-Join-optimization-with-Carbondata-s-metadata-tt103186.html#a103187>
Since Spark already handling in new version, everyone's opinion was not to make before spark. SO this will be specific for CDC as here its little different as we are joining intermediate dataframe with source to get the files to scan. SO this should be fine. Only problem with the cartesian product as mentioned in design doc, you can check and give your inputs on that, i also have one more solution to search in a distributed way with a interval tree data structure. Thanks, Akash -- Sent from: http://apache-carbondata-dev-mailing-list-archive.1130556.n5.nabble.com/