zhangshenghang commented on PR #10331: URL: https://github.com/apache/seatunnel/pull/10331#issuecomment-3754478252
<!-- code-pr-reviewer --> This PR has several blocking issues that must be addressed: **[CRITICAL] Partition column types lost** `HiveSourceConfig.java:420` - `appendPartitionColumnsAsString()` hardcodes all partition columns as `BasicType.STRING_TYPE`, ignoring original types from Hive Metastore. This breaks type correctness and may cause Sink type conversion failures at runtime. **[CRITICAL] Missing test coverage** - `HiveIT.java` - PR description mentions ORC tables but E2E tests only cover TEXT and PARQUET formats - `HiveIT.java:108-131`, `HiveSourceConfigEmptyFilesTest.java` - No tests for partitioned tables with empty partitions, despite PR claiming to fix this scenario **[MAJOR] Unsafe error handling** `HiveSourceConfig.java:220-234` - `parseFilePaths()` catches all IOExceptions as "directory not found", silently treating permission errors and other I/O failures as empty directories. **[MAJOR] Missing null checks** `HiveSourceConfig.java:365` - `buildRowTypeFromHiveMeta()` calls `table.getSd().getCols()` without null checks. If StorageDescriptor is null, NPE will cause connector initialization failure. `HiveSourceConfig.java:368-372` - Loop over `cols` doesn't check if FieldSchema objects or their methods return null, risking NPE with incomplete metadata. These issues should be fixed before merging to ensure correctness and test coverage for the new empty table fallback behavior. -- 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]
