Gabriel39 commented on PR #68667: URL: https://github.com/apache/doris/pull/68667#issuecomment-5925931334
Fixed both issues exposed by the regression and local end-to-end validation: - Legacy Variant is not a legal ARRAY() argument. The suite retains scalar Prepare/GetSchema/execution coverage in V1 and checks nested ARRAY/MAP/STRUCT values and empty results in V2. - V2 exposed a separate schema RPC failure: constructing an empty RecordBatch requires nested extension builders that Arrow does not provide. Schema discovery now writes a schema-only IPC stream, which both FE and remote BE readers already support. Before pushing, I reproduced the failures locally, built the complete current-branch BE with the fix, and ran the full affected Groovy suite against a local 1 FE / 2 BE cluster in both V1 and V2 modes. Both passed. Python ADBC integration passed in both modes, and nested ADBC ExecuteSchema/Prepare/fetch checks passed with native mode false/true/false. All 51 selected BE tests and clang-format 16 also passed. Full cloud topology validation remains with CI. -- 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]
