wenzhenghu commented on issue #65762:
URL: https://github.com/apache/doris/issues/65762#issuecomment-5000838227
Additional static-review finding not yet covered in the issue body:
6. Hive mixed-format splits are accepted by the FileScannerV2 support check
but rejected by HiveReader at runtime.
Evidence:
- FileScannerV2::is_supported accepts Hive ORC/Parquet ranges based on the
current split format:
https://github.com/apache/doris/blob/728d362448a25051293291b42985a79f91c0122c/be/src/exec/scan/file_scanner_v2.cpp#L277-L302
- HiveReader::prepare_split requires every split handled by the same reader
instance to use the same initialized format and returns InternalError otherwise:
https://github.com/apache/doris/blob/728d362448a25051293291b42985a79f91c0122c/be/src/format_v2/table/hive_reader.cpp#L104-L111
So a mixed ORC/Parquet Hive scan can pass the V2 eligibility check first and
then fail later when the scanner advances to a split whose format differs from
the first split.
This is a capability mismatch between the V2 support matrix and the actual
HiveReader contract. The implementation should either reject mixed-format Hive
scans up front or make the Hive V2 path support safe per-split format switching.
--
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]
