achandel-01 opened a new pull request, #605: URL: https://github.com/apache/atlas/pull/605
…impala_process with empty outputs[], breaking lineage ## What changes were proposed in this pull request? earlier for impala insert overwrite (DML QUERY) shows the wrong lineage graph : <img width="556" height="204" alt="image" src="https://github.com/user-attachments/assets/22d9397e-ae37-4ac9-9258-dc0577a95d8d"; /> now after these changes lineage will look like this : <img width="532" height="204" alt="image" src="https://github.com/user-attachments/assets/19c69c40-6e54-4abf-9417-7344f92fcdb7"; /> explanation : getEntities() walks inputNodes and outputNodes and fills the process’s inputs and outputs lists, using a single processedNames set to avoid duplicates. When the same table is both read and written (e.g. self-lineage / INSERT OVERWRITE into the table you read from), its qualified name appears in both node lists. The input loop runs first and registers that name in processedNames. The output loop then treats the name as “already processed” and skips adding that table to outputs, even though it is a real write target. The fix is to dedupe per side (separate sets for inputs vs outputs) so the same qualified name can appear in both inputs and outputs when the lineage says so. ## How was this patch tested? unit testing , mvn build -- 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]
