dmitry-chirkov-dremio commented on code in PR #48597: URL: https://github.com/apache/arrow/pull/48597#discussion_r3581629693
########## cpp/src/gandiva/annotator.cc: ########## @@ -93,6 +93,25 @@ void Annotator::PrepareBuffersForField(const FieldDescriptor& desc, } } +const Status Annotator::CheckEvalBatchFieldType( Review Comment: I think this validation is incomplete for the new caching model. It checks type mismatches for columns that are present in the record batch, but it never verifies that all referenced input fields were found. `PrepareEvalBatch()` still skips unmatched columns, so if an expression references a field that is absent from the batch, the corresponding eval-batch slots are left unset rather than failing here. Since the earlier schema-equality / field-existence checks were removed, this now looks like a correctness/stability regression. -- 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]
