Question

I have a temporary table that hold all items available

#TableA

ItemId  ItemName    ItemVer
1       Name1       2
2       Name2       9
3       Name3       5
.
.
.
500     Name500     3

I then have another that holds all the dates a Snapshot was taken along with the SnapshotId

CREATE TABLE #tmpSnapshot_Dates(
[SnapshotId] [uniqueidentifier] NOT NULL,
[DateTaken] datetime NOT NULL) 

INSERT INTO #tmpSnapshot_Dates SELECT SnapshotId, DateTaken From Snapshot_Info Where DateTaken <= @EndDate

Now for each date in the #tmpSnapshot_Dates table I want to get a count that relates it each Item

SELECT ItemId, Count(*) From Items Where SnapshotId = @SnapshotId

And then update #TableA by adding a column

DECLARE @ColName VARCHAR(100)
SET @ColName = 'Installed ' + CONVERT(Varchar, @DateA, 6)

exec('ALTER TABLE #TableA ADD ['+@ColName+']  int NOT NULL CONSTRAINT [DF_#TableA_Installed on] DEFAULT 0')

And then insert the data into the new column. I then need to repeat this for each date in #tmpSnapshot_Dates

giving a resulting #TableA that looks something like

#TableA

ItemId  ItemName    ItemVer    Installed 01 Jan 12    Installed 07 Feb 12 
1       Name1       2          34                     33
2       Name2       9          56                     59
3       Name3       5          12                     26
.
.
.
500     Name500     3          98                     106

My questions are

1. Am I creating a nightmare here by approaching it the wrong way?
2. How am I best to loop through #tmpSnapshot_Dates and then get the data to create the next column? I always try to avoid cursors but is this one of those situation I need to use one?

There could be 12 columns to add and not every item will be returned for each date.

Answer 1

Definitely it won’t be nice, but yes, you need to use cursors and iterate over the #tmpSnapshot_Dates, but if you have the option i suggest you to use a reporting software with the capability to create pivot grids.