Passing parameters to PL/SQL table types

Question

Hi Everybody,
I have one question about passing PL/SQL tables types and tabs as IN parameter in procedure.I am working in 11.2.0.2.0 environment. I am stuck on how to pass those values to procedure.Please find below more details:
Table 1:
CREATE TABLE ITEMS
ITEM_ID VARCHAR2(40 BYTE) NOT NULL,
ITEM_NAME VARCHAR2(40 BYTE),
SERIAL NUMBER(2),
ADDED_ON DATE);
Table 2:
CREATE TABLE ITEM_ACTIVITY_INFO
ITEM_ID VARCHAR2(40 BYTE) NOT NULL,
ACCOUNT_TYPE VARCHAR2(1 BYTE),
ID_NUMBER NUMBER(3),
ACTIVATION_DATE DATE);
Table 3:
CREATE TABLE ITEM_GROUP
GROUP_ID NUMBER(2) NOT NULL,
ITEM_ID VARCHAR2(40 BYTE),
GROUP_TYPE VARCHAR2(20 BYTE),
GROUP_DATE DATE);
Table 4:
CREATE TABLE ITEM_ADDRESS
GROUP_ID NUMBER(2) NOT NULL,
NAME VARCHAR2(60 BYTE),
ADDRESS VARCHAR2(100));
Following types are created:
CREATE OR REPLACE TYPE ITEMS_TYPE AS OBJECT
ITEM_ID VARCHAR2(40 BYTE),
ITEM_NAME VARCHAR2(40 BYTE),
SERIAL NUMBER(2),
ADDED_ON DATE);
CREATE OR REPLACE TYPE ITEM_ACTIVITY_TYPE AS OBJECT
ITEM_ID VARCHAR2(40 BYTE),
ACCOUNT_TYPE VARCHAR2(1 BYTE),
ID_NUMBER NUMBER(3),
ACTIVATION_DATE DATE);
CREATE OR REPLACE TYPE ITEM_GROUP_COMP_TYPE AS OBJECT
GROUP_ID NUMBER(2) NOT NULL,
ITEM_ID VARCHAR2(40 BYTE),
GROUP_TYPE VARCHAR2(20 BYTE),
GROUP_DATE DATE
ITEM_ADDRESS_IN ITEM_ADDRESS_TYPE);
CREATE OR REPLACE TYPE ITEM_ADDRESS_TYPE AS OBJECT
GROUP_ID NUMBER(2),
NAME VARCHAR2(60 BYTE),
ADDRESS VARCHAR2(100));
CREATE OR REPLACE TYPE ITEM_GROUP_COMP_TAB AS TABLE OF ITEM_GROUP_COMP_TYPE;
Create or replace procedure ITEM_ADD_CHANGE(
ITEM_IN IN ITEMS_TYPE,
ITEM_ACTIVITY_IN IN ITEM_ACTIVITY_TYPE,
ITEM_GROUP_IN IN ITEM_GROUP_COMP_TAB,
ITEM_OUT IN OUT ITEMS.ITEM_ID%TYPE);
Above are the paramteres we are passing to procedure.
I need help in how to pass parameters to above procedure. All comments and responses will be highly appreciated. Thanks everyone for going through the post. Please let me know if more more information is required on this problem.
Regards
Dev

Answer 1

Billy  Verreynne  wrote:
Types used in this fashion, only make sense if the table is based on the type. It makes very little sense to have a table structure and then to duplicate the structure using a type.
The 2 structures may be defined the same. But they are NOT interchangeable and requires one to be converted to the other to use. This is not sensible in my view. It is far easier in that case to simply use the PL/SQL macro +%RowType+ to create a duplicate structure definition - one that can natively be used for touching that table, without conversions required.
If you do want to use types, define the type, then define the table of that type, adding the required constraints (pk, fk, not null, check) to the table's definition.Billy:
Just curious, why do you say it makes very little sense to have a type modeled on a table? I do that a lot. In my case, I am getting the values from an external program, not building them manually, but it makes a lot of sense to me.
One application where I do this a lot has a java front-end that parses HL7 messages. Each message contains at least minimal information about a variable number of entities (and often several rows for an entity) in the database, and must be processed as a single atomic trasnaction. So, rather than have potentially hundreds of parameters to the "main" driver procedures for different message types I created a set of types more or less identical to the tables representing the entities. The java program parses the mesasge and populates the type, then calls the appropriate stored procedure for the message type passing in the populated types. My stored procedure then does inserts/updates or deletes as appropriate over potentially dozens of tables.
John