Gabriel39 commented on PR #69171: URL: https://github.com/apache/doris/pull/69171#issuecomment-6248436383
Fixed the remaining UUID/BINARY review finding and the external-regression failure: - Explicit Hive BINARY targets retain the original 16 bytes in the legacy Parquet and ORC readers. File Scanner V2 now converts native UUID values back to big-endian file bytes, including nested values. - Added Hive coverage for Parquet/ORC, both scanner settings, strict/non-strict casts, scalar/array/map/struct values, NULLs, empty collections, byte length and filtering. A direct legacy Parquet reader test also covers plain/dictionary fixtures across row groups and 37-row batches, since timestamp-semantics negotiation can force V2 independently of the session flag. - Regenerated the `test_hdfs_parquet_group2` golden output for ENUM text. The generated output also normalizes three equivalent floating-point representations. Validation: all four focused C++ tests passed after confirming the old-code failures; full ASAN BE build (including glibc compatibility), formatting/diff checks, and four regression suites passed: `test_hive_uuid_binary`, `test_hdfs_parquet_group2`, `test_uuid_parquet_native`, `test_uuid_orc_varbinary_mapping`. The previous FE coverage job 1072210 failed while downloading Maven from the internal OSS URL, before tests started. The external-regression failure in job 1072171 was the ENUM golden mismatch fixed here; its other four failures were already muted (S3 timeouts / unsupported Paimon OBS filesystem). Pushed commit `32fd2fedcdb` and requested fresh `run buildall` and `/review` checks. Those new checks are pending. -- 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]
