Join Strings through Rapids

[mtsol]

Hi, if I have two spark columns, how can I join those using rapids, and which function is available in cudf for that if I am trying to do that in evalueteColumnar function?

[jlowe]

Can you elaborate a bit on what you mean by "join"? Given the reference to evaluateColumnar I assume this is for a simple UDF expression which must return one output row for every input row. That implies "join" isn't like a SQL join. Does "join" mean concatenating the strings together for each row like the Spark string concatenation function, i.e.: concat? If that's the case, this is already implemented by the RAPIDS Accelerator. See https://github.com/NVIDIA/spark-rapids/blob/branch-23.04/sql-plugin/src/main/scala/org/apache/spark/sql/rapids/collectionOperations.scala#L60-L66 for the source code.

If "join" means something else, it would be great to have a toy example showing the semantics of the operation required.