jizezhang commented on issue #17964: URL: https://github.com/apache/datafusion/issues/17964#issuecomment-3495438041
@Jefffrey - Follow-up on `array -> make_array`,as implementation of `coalesce_type` method is somewhat different btw the two (`make_array` uses `type_union_resolution` https://github.com/apache/datafusion/blob/631f9abf6953f8f80dc6e552fed70b96ee5dc8d3/datafusion/functions-nested/src/make_array.rs#L132 while `spark_array` uses `comparison_coersion` https://github.com/apache/datafusion/blob/f7a9f2449f7e45bf4dbff49979089e5c619b1faf/datafusion/spark/src/function/array/spark_array.rs#L134 from `type_coercion::binary module`), do we want to consolidate the implementations or leave as is? - For `exp1m -> exp`, from what I understand, datafusion `exp` is created with the macro `make_math_unary_udf` that accepts input type `Float32` or `Float64` https://github.com/apache/datafusion/blob/f2437d13c08283cb1dd1241b716c4e3a4e027648/datafusion/functions/src/macros.rs#L211 while spark `exp1m` accepts a scalar or array of `Float64` https://github.com/apache/datafusion/blob/f2437d13c08283cb1dd1241b716c4e3a4e027648/datafusion/spark/src/function/math/expm1.rs#L74-L83 If that is the case, we would not consolidate the two? Thanks! -- 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]
