Question

We have an Oracle type that’s declared as such:

CREATE OR REPLACE TYPE its_accountarray;

There is no base type. This type is then used in Procedures and packages (and in Java Stored procedures as an array descriptor). In the procedure/package it uses extend function to add to array.

<declare section>
    v_account latax.ITS_ACCOUNTARRAY := new ITS_accountArray();
...
BEGIN
    ...
    ...
    v_account.extend(1);

and in Java, it’s used as:
oracle.sql.ARRAY ls_accountARRAY, ls_periodARRAY;

oracle.sql.ArrayDescriptor ad =   ArrayDescriptor.createDescriptor(**"ITS_ACCOUNTARRAY"**, oconn);

ls_accountARRAY = new ARRAY(ad, oconn, arg_accounts);
ocs.setARRAY(2, ls_accountARRAY);

I am curious as to how this works. Even though the name has Array in it, it is not defined as an array or table type, like I typically see. It works, but is this a legal usage or should I declare the type to be of some array type explicitly?

Thanks

Sam

Answer 1

I tried to reproduce what you say is happening, and got an error:

CREATE OR REPLACE TYPE its_accountarray;

DECLARE
  v its_accountarray := its_accountarray();
BEGIN
  null;
END;
/

PLS-00311: the declaration of "ITS_ACCOUNTARRAY" is incomplete or malformed

So, you’re probably looking at the wrong type definition (e.g. you’re looking at it in the wrong schema), or not looking at the entire type definition (e.g. there is a TYPE BODY that provides the implementation). Since this seems to be acting like a nested table type, the former sounds more likely.