Blizzara commented on issue #14632: URL: https://github.com/apache/datafusion/issues/14632#issuecomment-2654253000
Actually, this is related to an interplay between the substrait stuff and the analyzer. Seems like what happens is: - substrait is kind of handled "correctly", in that the produced & consumed plan reports the correct schema. Thus the consumer's [rename_expressions](https://github.com/apache/datafusion/blob/980931c5a70b8b08c918195803145f64f1ec901f/datafusion/substrait/src/logical_plan/consumer.rs#L926) does not wrap the expressions in renaming casts. - analyzer (called by the [roundtrip code](https://github.com/apache/datafusion/blob/980931c5a70b8b08c918195803145f64f1ec901f/datafusion/substrait/tests/cases/roundtrip_logical_plan.rs#L1373) but also my use-case in production) however sees that the branches of the case-when return different types (Struct({f1:Utf8}) vs Struct({c0:Utf8}) or something like that), and solves that by casting the then-arm to match the else-arm. I'm not sure what the correct solution here would be, but maybe the use of `named_struct` in substrait is a mistake on my part given the rest of substrait is not "named".. -- 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: github-unsubscr...@datafusion.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org For additional commands, e-mail: github-h...@datafusion.apache.org
