Question

I have a set of tables with different data type for the columns and I need to consolidate a way for retrieving data. I thought using a function would be a good idea, but I don’t know how to define one function having different return types.

For example, how to define this function to be able to use different definitions for tabletype.

CREATE OR REPLACE FUNCTION retrieve_info(field_id in integer)
RETURN pintegertypetable -- <-- how to change this to return a more generic record built dynamically in the code below?
AS
  r pintegertypetable := pintegertypetable ();
BEGIN
  r.extend;
  r(i) := pintegertypetable (someinteger);
  return r;
END;

Is that possible?. Is there a better way to handle this problem: different columns stored originally in a lot of legacy tables, and given that every column has different data types, in which way we can retrieve the most recent information conserving the original data types without hardcoding views neither storing everything in varchar2 and casting again in client code?

Answer 1

You can implement this by using a weakly-typed Ref Cursor as the return type. This is especially easy to implement from a client interface using JDBC, as the returned cursor type can be stepped through just like any query result and the metadata can be interrogated from ResultSet.getMetaData(). Here’s an example:

CREATE OR REPLACE PROCEDURE retrieve_info(field_id in integer, p_cursor in out sys_refcursor)
AS
BEGIN
  open p_cursor for 'select * from emp';
END;

The query in quotes could be anything returning any type, for any number of columns.