March 11, 2013 at 9:18 am
Let's just say I have a table with columns ID and Name that looks like:
[font="Courier New"]ID Name
01
02
03 Bill
04
05
06
07
08
09 Ted
10
11
12 [/font]
I want to update rows 1-2 & 4-6 with the Name Bill
and 7-8 & 10-12 with the Name Ted.
the update will always be an update to two prior and three subsequent rows where there is a non-null entry in column [Name]
This is easy enough to do in an RBAR fashion but not efficient but I'm not interested in that.
How could I achieve this using ROW_NUMBER()
Thank you in advance
Sal.
(Updaning??? must be a new command, meant updating, sorry)
March 11, 2013 at 9:48 am
You don't need ROW_NUMBER here:
declare @table table (ID int, Name varchar(100))
insert @table select 1,null
insert @table select 2,null
insert @table select 3,'Bill'
insert @table select 4,null
insert @table select 5,null
insert @table select 6,null
insert @table select 7,null
insert @table select 8,null
insert @table select 9,'Ted'
insert @table select 10,null
insert @table select 11,null
insert @table select 12,null
UPDATE u
SET u.Name = n.Name
FROM @table u
CROSS APPLY(SELECT TOP 1 t.Name FROM @table t
WHERE t.Name IS NOT NULL
AND ( t.ID - u.ID between 1 AND 2
OR u.ID - t.ID between 1 AND 3)) n
WHERE u.Name IS NULL
SELECT * FROM @table
Please refer to the link at the bottom of my signature. The article provides tips about how to post your question on this forum in order to get most relevant and prompt help.
March 11, 2013 at 10:17 am
Thank you for the quick response. The example you provided works great. So I've tweaked it to update my test table which has about 2.5million rows. It's been running for 10 minutes and counting. The estimated execution plan looked ok and it was making use of available indexes. I'll post up the results when it completes.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply