thank you

thank you very much

Thanks & regrads

Harishkumar.M

I just inherited a DB table that contains an ID column and an order number column that was not originally an Identity column. So, all records are unique on the ID PK column, but *should* also be unique on the order number as well…they are NOT.

Some records have a NULL order number and some order numbers are duplicated. The difficulty is in making the decisions of how to rectify these issues:

Can we delete the data if it is older? That is simple.

If not, how to make this an Identity column and still retain the existing order numbers for historical purposes? SQL Server 2005 doesn’t let you make an existing column IDENTITY: “You cannot modify an existing table column to add the IDENTITY property.”

How to find gaps (un-used numbers) around the same date range of the records with NULL or duplicate order numbers to try to use those to fix the bad records…we could simply use the next highest order number, but the users intuitively know the approximate date of the orders based on the order number and this would throw them off.

If anyone has suggestions for these real-world issues, I’d be interested in hearing them.

When we went to move this table from one database to another, the identity column was reseeded. This of course caused havoc with our data since records were being paired incorrectly.

Is there a way around this?

In the above example, the identity missing was found by manually. Is there any way to find the same using SQL statement ?

ie, How we can find the list of tables whose identity was missed(not is sequential order) within the entire database?

Can you kindly elaborate your solution about triggers

use the updated trigger on the table and get the result in sequence.

like 1,2,3,4,5,6,7
no 1,2,3,5,7,10

IF you need consecutive IDs, create a VIEW that simply ORDERS the table with consecutive IDs.

Trust me, you don’t need consecutive IDs in the TABLE.

Without actually knowing what they were doing, kind of hard to say it was a design flaw, I think the whole point of the article was how to reindex identity column so that they are in sequence.

Now if had been something I worked, then I would not care what is in the id column, I would have a second column that would be seen by the user.

i.e. patient number, product sku, the id column would be something that could use to build a relationship lookup or nothing at all. Again depending on the design.

Nice article, but I am sure that the author pointed out how to get around it and then probably asked the developer why they were doing that and point him/her in the right direction. :)

IMHO, any solution that uses integers as identity values should also be able to use GUIDs without any loss of basic functionality. If you can’t do that, you need to rethink what you’re doing.

if the order matters, generate it at run time!

CREATE TABLE TableID (ID INT IDENTITY(1,1), Col VARCHAR(10))
GO
/* Insert 10 records with first value */
INSERT INTO TableID (Col)
VALUES (‘First’)
GO 10
/* Check the records in table */
SELECT *
FROM TableID
GO
/* Delete records from middle and bottom both*/
DELETE
FROM TableID
WHERE ID IN (4,7,11,12)
Go
/* Check the records in table */
SELECT *
FROM TableID
GO
/*
Now check the missing identity values
To Do so we will create a memory table with a sequential number,
from 1 to the ident_current value and join that to the live table.
*/
Declare @rows Int
Declare @numberOfRowsToInsert Int

declare @seq table (
seq int not null primary key
)
/* Get current Max Value and reseed table */
SELECT @rows = MAX(ID) FROM TableID
DBCC CHECKIDENT(‘TableID’, RESEED, @rows)
set @rows = (select Ident_Current(‘TableID’)) + 5
/* Insert values into Memory table */
declare @i int
set @i = 1
while @i <= @rows
begin
insert @seq values( @i )
set @i = @i + 1
end
/* Now set the Identity Insert On, so that we can insert any number */
Set Identity_Insert TableID On

INSERT INTO TableID(Id,col) Select seq,’Third’
From @seq left outer join TableID T on seq = T.ID
Where T.ID Is Null

Set Identity_Insert TableID OFF
/* Check the records in table */
SELECT *
FROM TableID
GO
/* Clean Database */
DROP TABLE TableID
GO

-Vinit