Create table script by passing value from table

Question

I Have a procedure to check if the table exists, if yes drop and re-create else create table.
CREATE OR REPLACE
PROCEDURE test_proc authid current_user
AS
v_cnt NUMBER:=0;
BEGIN
SELECT COUNT(*)
INTO v_cnt
FROM all_tables
WHERE table_name='T1'
AND owner = 'WORKSPACE';
IF v_cnt = 1 THEN
EXECUTE immediate 'DROP TABLE WORKSPACE.T1';
END IF;
EXECUTE immediate 'CREATE TABLE WORKSPACE.T1
ITM_NBR NUMBER ,
LCT_NBR NUMBER
SELECT COUNT(*)
INTO v_cnt
FROM all_tables
WHERE table_name='T2'
AND owner = 'WORKSPACE';
IF v_cnt = 1 THEN
EXECUTE immediate 'DROP TABLE WORKSPACE.T2';
END IF;
EXECUTE immediate 'CREATE TABLE WORKSPACE.T2
LCT_NBR NUMBER(5, 0) NOT NULL ,
PRY_CD NUMBER(5, 0) NOT NULL )';
END;
Requirement: Instead of passing the Owner,tablename & the create table scripts directlyin the query, it should be passed from a database table which will be in given format.
TableName Query Owner
T1 CREATE TABLE WORKSPACE.T1(ITM_NBR NUMBER, LCT_NBR NUMBER) WORKSPACE
T2 CREATE TABLE WORKSPACE.T2(LCT_NBR NUMBER(5, 0) NOT NULL ,PRY_CD NUMBER(5, 0) NOT NULL) WORKSPACE

Answer 1

user12276240 wrote:
Excuse all. The database columns are yet to be finalized and every time changes come in we are currently dropping and re-creating the tables which leads in changing the script, hence we are going in for this approach.
Moreover if i am going to have this script everytime i have to drop and re-create all the tables which can be elimated by this way.So why not have a SQL script as follows?
drop table emp purge;
create table emp(
create index .. on emp( .. );
..It drops the existing table (if exists). It then creates the table with constraints and indexes. This script is now the master definition of the table. You can slap a change log header to it. Any changes that needs the table to be redefined and recreated are done via this script. Which means it is also checked into the source code repository as it is part of the code of that system/application.
I would however remove the drop table from it in case some developer accidentally run it at the wrong time and trash the table.
The bottom line is that DDL code is part and parcel of the installation, configuration and setup code - and that this is as important as the source code. This code also needs to be "formalised" in proper scripts, checked into SVN/CVS/etc, and used to bootstrap the system (or parts of it), as and when needed.