SQL Looping
I'm quite new to T-SQL, so bear with me.
I have a table with columns UserName, PCName, Date. There is no key field and there may be many entries from the same user name. I only want to keep the 10 newest entries for each unique user name. I've written code like below:
This works fine but how do I expand it and run it for each unique user. I can isolate the unique users easily enough with: SELECT DISTINCT UserName FROM dbo.PCs , but how do I loop through the result.
I do understand that looping isn't a favored approach in SQL, but I'm also having trouble figuring our how to do this any other way. I'm not quite grasping JOINs and such. I'm certainly open to alternatives to looping, but it may take some explaining.
I have a table with columns UserName, PCName, Date. There is no key field and there may be many entries from the same user name. I only want to keep the 10 newest entries for each unique user name. I've written code like below:
DELETE FROM dbo.PCs WHERE UserName = 'MyUserName' AND
Date NOT IN (SELECT TOP 10 Date FROM dbo.PCs WHERE UserName = 'MyUserName' ORDER BY Date DESC)
This works fine but how do I expand it and run it for each unique user. I can isolate the unique users easily enough with: SELECT DISTINCT UserName FROM dbo.PCs , but how do I loop through the result.
I do understand that looping isn't a favored approach in SQL, but I'm also having trouble figuring our how to do this any other way. I'm not quite grasping JOINs and such. I'm certainly open to alternatives to looping, but it may take some explaining.
Eyeballeth thy article on SQL Server Delete Duplicate Rows Solutions, scroll down to the 'Delete #1 section', and change the query to meet your needs and your definition of '10 newest entries', then change WHERE row_number > 1 to > 10.
always backup tables before whacking them with untested code where data could be lost
& for a delete query I suggest trying it as a select query first
If that looks reasonable then:
note that DESC(ending) order is needed to keep the 10 latest records.
also:
no key on a table = not good
using "reserved words" as a column name = not great; use [Date] to identify that field clearly
& for a delete query I suggest trying it as a select query first
;WITH cte
AS (
SELECT
UserName
, PCName
, [Date]
, ROW_NUMBER() OVER (PARTITION BY UserName, PCName ORDER BY [Date] DESC) AS rn
)
SELECT * FROM cte
WHERE rn > 10
;
If that looks reasonable then:
;WITH cte
AS (
SELECT
UserName
, PCName
, [Date]
, ROW_NUMBER() OVER (PARTITION BY UserName, PCName ORDER BY [Date] DESC) AS rn
)
DELETE FROM cte
WHERE rn > 10
;
note that DESC(ending) order is needed to keep the 10 latest records.
also:
no key on a table = not good
using "reserved words" as a column name = not great; use [Date] to identify that field clearly
Premium Content
You need an Expert Office subscription to comment.Start Free Trial