What my boss wants me to do is to create a view, in SQL Server 2000,
that will provide not only a row number field of some sort, but that
will produce sequential ordering for arbitrary selects and orderings.
So, if my data is a table with values from A thru D and my user does
SELECT data FROM vwTable, the result would be:
Row Data
--- ----
1 A
2 B
3 C
4 D
But is they did SELECT data FROM vwTable ORDER BY data DSC, they would
get
Row Data
--- ----
1 D
2 C
3 B
4 A
And if the did SELECT data FROM vwTable WHERE Data IN ('B', 'C'), they
would get
Row Data
--- ----
1 B
2 C
In SQL 2005, of course, this would be fairly trivial since I could use
the ROW_NUMBER function. In 2000, though, it seems to be utterly
impossible. My boss, however, is convinced that there must be some way
to create a calculated field to do it.
I'll be cursed if I can figure out a way to do so.
Any suggestions would be appreciated.
Paste the following in Google search box:
"dynamically number rows site:support.microsoft.com"
--
Anith
Madhivanan
Thanks, however, while that is a good way to derive row numbers in a
select statement, unfortunately it isn't quite what my boss is asking
me to do. She wants a view that will produce row counts in a
calculated field regardless of the order that the user uses to select
the data.
I would prefer to require the user to generate the row numbers in their
selects, wjhich wouldd allow for the solution you offered.
Unfortunately, that isn't what I've been tasked to do.
SQL Reporting Services.
Hi Andrew,
The way you describe it here, it's impossible. That holds true for both
SQL Server 2005 and SQL Server 2000. Even ROW_NUMBER() won't help you.
If you need the row numbers to match the order specifiede on the select
and if you want to skip numbers for rows not included in the select,
you'll have to add row numbering logic on the SELECT statement. If you
add row numbers in the view, the numbers won't change if you exclude
some rows or choose a different order when selecting from the view.
Just to prevent misunderstanding - it is NOT impossible to get the
result sets you require. But it's only possible by extending the SELECT
with some row numbering logic. Either using ROW_NUMBER() if you're using
SQL Server 2005, or by using either a correlated subquery or a self-join
and a GROUP BY if you're using SQL Server 2000.
--
Hugo Kornelis, SQL Server MVP
Time to get a new boss?
What she is asking for is not possible. You would have to package the
user's SELECT statement somehow, so you can modify to add the row-number
column. As Hugo pointed out, this is the same on SQL 2005.
--
Erland Sommarskog, SQL Server MVP, esq...@sommarskog.se
Books Online for SQL Server 2005 at
http://www.microsoft.com/technet/prodtechnol/sql/2005/downloads/books.mspx
Books Online for SQL Server 2000 at
http://www.microsoft.com/sql/prodinfo/previousversions/books.mspx
Cant you make use of Recordnumber feature such as the one available in
Crystal reports?
Madhivanan
That's what I thought. I just wanted to be extra sure that there
wasn't some tricky way to do this before I went back to her and said
that it simply could not be done the way that she was asking.
There is no guarantee that it will work any version of SQL Server. In fact
for a result set of any size, I would not expect it to work.
What is guaranteed to work, at least in SQL 2005, is if you have a
table with an IDENTITY table, and perform an INSERT with an ORDER BY.
Note that this does not apply to SELECT INTO with the IDENTITY function
and ORDER BY. In that case, there is *no* guarantee.
But, at least we "solved" the problem!!!!