Can anyone point me to an example of a CTE being used to step through a recordset instead of a cursor?
All of the example I've been able to find show recursion. I don't need that. I just want to step through a recordset one a开发者_JAVA百科t a time.
Thanks.
(EDIT)
Perhaps this is best asked another way. I assumed a CTE or cursor might be the best way to get the data I want but here's an example.
If I have a table with userids, codes and dates like the following, where the userid not a primary key.
2011-05-24 11:06:28.080 CODE1 199 2011-05-24 11:06:28.080 CODE2 199 2011-06-08 13:30:14.830 CODE2 209 2011-06-08 13:30:14.313 CODE1 209 2011-06-08 13:30:13.840 CODE2 209 2011-06-08 13:30:13.450 CODE1 209 2011-06-08 13:30:13.050 CODE2 209 2010-05-07 10:45:06.800 CODE3 79 2010-05-07 10:45:04.833 CODE3 79 2010-10-15 07:30:16.193 CODE3 79 2010-01-26 13:51:43.923 CODE3 79 2010-01-26 13:51:43.923 CODE3 79 2010-01-26 13:51:44.427 CODE3 79 2010-01-26 13:51:45.103 CODE3 79 2010-01-26 13:51:45.890 CODE3 79 2010-01-26 13:51:46.733 CODE3 79 2010-01-25 12:40:39.737 CODE3 81 2010-01-25 12:40:40.890 CODE3 81 2010-01-25 12:40:41.627 CODE3 81 2010-01-25 12:40:43.277 CODE3 81 2010-01-25 14:29:08.360 CODE3 81 2010-01-21 19:36:34.660 CODE3 98 2010-01-21 19:36:34.843 CODE3 98 2010-01-21 19:36:35.013 CODE3 98 2010-01-21 22:27:24.317 CODE3 83 2010-01-21 22:31:21.443 CODE2 83 2010-01-22 19:44:28.880 CODE3 83
And I want to select from this table the oldest date and code for each user is it possible to do this using a CTE?
I can do a select where I group by customerid and get the MIN(datefield) but I get can't retrieve the code associated with that earliest date unless I add it to my group by clause... and then of course I get multiple rows for a customer, one with each associated code.
I found a solution that works where I create a temp table and fill it with the MIN(Date) and customer ID by grouping on the customerid. Then I update this temp table by joining it to the original table by customer and the date field to get the code associated with that row... but this seems to be a hack.
I was wondering if looping through the original table (sorted by customerid and date) and each time I see a new customer id inserting that record into a temp table, would be a cleaner solution... especially since the solution I'm finding assumes a unique datetime/customerid combination.
Here's an example of what I'm using now. It just seems like a hack and I'm wondering if a cusor (which I normally avoid) or a CTE would be a cleaner solution.
DECLARE @Table1 TABLE
(
ClaimDate datetime,
VenueCode nvarchar(50),
CustomerID int
)
INSERT INTO @Table1
SELECT sc.CreateDate, v.code, sc.CSFCustomerID
FROM foo join foo1 on (snip)
DECLARE @Table2 TABLE
(
ClaimDate datetime,
VenueCode nvarchar(50),
CustomerID int)
INSERT INTO @Table2
select MIN(ClaimDate), NULL, CustomerID from @Table1 group by CustomerID
UPDATE ft
SET ft.SomeCode = t.VenueCode
FROM @Table2 ft
INNER JOIN @Table1 t ON ft.CustomerID = t.CustomerID
AND ft.ClaimDate = t.ClaimDate
Thanks, and I promise I'll be better about selecting answers in the future.
with qry
(
SELECT Min(Date), userID
FROM TABLE
GROUP BY UserID
)
SELECT Table.*
FROM TABLE
INNER JOIN qry on qry.UserID = Table.UserID AND qry.Date = Table.Date
Though it is quite old post and the question title is not appropriate to illustrate your main task, that is finally: And I want to select from this table the oldest date and code for each user is it possible to do this using a CTE?
The answer is yes. You may try this:
-- CTE version of the SQL
WITH CTE (OldestClaimData, CustomerID)
AS
(
SELECT
MIN(ClaimDate), CustomerID
FROM
UserClaims
GROUP BY
CustomerID
)
SELECT DISTINCT
UC.*
FROM
UserClaims UC
INNER JOIN
CTE ON (
CTE.OldestClaimData=UC.ClaimDate
AND
CTE.CustomerID=UC.CustomerID
AND
UC.VenueCode=(
SELECT TOP 1
VenueCode
FROM
UserClaims sub
WHERE
sub.ClaimDate=UC.ClaimDate
AND
sub.CustomerID=UC.CustomerID
)
)
ORDER BY
UC.CustomerID;
I've put my code and yours into an SQL Fiddle for your convenience.
P.D.: work on your accept rate.
精彩评论