Question

Here is my “upsert” code:

UPDATE LastTicket SET LastTicketNumber=LastTicketNumber+1
    OUTPUT INSERTED.LastTicketNumber WHERE CategoryId='1';

IF @@ROWCOUNT=0 INSERT INTO LastTicket (CategoryId,LastTicketNumber)
    OUTPUT INSERTED.LastTicketNumber VALUES ('1','2')

So, when the row exists, it successefully updates, the OUTPUT returns the new, incremented LastTicketNumber.

On the other hand, when the row does not exist, the sql server successefully creates it and populates with the data I am passing to SqlCommand (1,2). So, it creates the row, but returns null. Meaning nothing! Why is that? And why when i replace the “INSERTED.LastTicketNumber” with the “INSERTED.CategoryId” is BEGINS to return not-null, the category id. Why is that? And how to return what I need?

The table has only these two columns and nonclustered primary composite key on both of them.

(MSSQL 2008)

Answer 1

If no row exists in the table, the first time the batch runs it will return two result sets – the first being empty (because there is no row to update) and the second containing the inserted Id.

Perhaps you are seeing the first result set and not the second.

Try the following:

DECLARE @t table (LastTicketNumber int)

UPDATE LastTicket SET LastTicketNumber=LastTicketNumber+1
    OUTPUT  INSERTED.LastTicketNumber INTO @t (LastTicketNumber) WHERE CategoryId='1';

IF @@ROWCOUNT=0 INSERT INTO LastTicket (CategoryId,LastTicketNumber)
    OUTPUT INSERTED.LastTicketNumber INTO @t (LastTicketNumber) VALUES ('1','2')

  select LastTicketNumber from @t