What
Right now, only one UDF can be registered under a single name and used on columnar tables. We want multiple overloaded UDFs to be able to be used for different use cases in a columnar table query.
Why
Function overloading is a built-in feature of Postgres. We should support it!
How
Infer the signature from the logical plan, pass the signature along with the function name as an argument to the UDF, use non-lossy type conversion (#826) to get the corresponding function OID and call the function.
What Right now, only one UDF can be registered under a single name and used on columnar tables. We want multiple overloaded UDFs to be able to be used for different use cases in a columnar table query.
Why Function overloading is a built-in feature of Postgres. We should support it!
How Infer the signature from the logical plan, pass the signature along with the function name as an argument to the UDF, use non-lossy type conversion (#826) to get the corresponding function OID and call the function.