How to phrase a T-SQL horizontal join? - sql-server

Question

I have a database which is including two tables, Labs and LabUsers
How do I join a column from LabUsers into a cell in Labs.
To be specific, I would like to have the user names of a lab to be concatenated and separated with a bullet ( • is Alt+0149), and to have these usernames sorted alphabetically left to right.
Here is an example.
The Labs table looks like this:
LabID LabName LabStudents
----- ---------- -----------
1 North NULL
2 North East NULL
3 South West NULL
and the LabUsers looks like this:
LabUserID LabUserName LabID
--------- ----------- -----
1 Diana 1
2 Paul 2
3 Paula 2
4 Romeo 1
5 Julia 1
6 Rose 2
7 Diana 2
I would like to get this outcome in the Labs table:
LabID LabName LabUsers
----- ---------- ---------------------
1 North Diana•Julia•Romeo
2 North East Diana•Paul•Paula•Rose
3 South West NULL
Here is the script to create the tables:
USE [tempdb];
GO
CREATE TABLE [dbo].[LabUsers]
(
[LabUserID] [int] PRIMARY KEY CLUSTERED,
[LabUserName] [nvarchar](50) NOT NULL,
[LabID] [int] NOT NULL
);
GO
INSERT [dbo].[LabUsers] SELECT 1, N'Diana', 1;
INSERT [dbo].[LabUsers] SELECT 2, N'Paul', 2;
INSERT [dbo].[LabUsers] SELECT 3, N'Paula', 2;
INSERT [dbo].[LabUsers] SELECT 4, N'Romeo', 1;
INSERT [dbo].[LabUsers] SELECT 5, N'Julia', 1;
INSERT [dbo].[LabUsers] SELECT 6, N'Rose', 2;
INSERT [dbo].[LabUsers] SELECT 7, N'Diana', 2;
CREATE TABLE [dbo].[Labs]
(
[LabID] [int] PRIMARY KEY CLUSTERED,
[LabName] [nvarchar](50) NOT NULL,
[LabUsers] [nvarchar](max) NULL
);
GO
INSERT [dbo].[Labs] SELECT 1, N'North', NULL;
INSERT [dbo].[Labs] SELECT 2, N'North East', NULL;
INSERT [dbo].[Labs] SELECT 3, N'South West', NULL;

Answer 1

SELECT l.LabID, l.LabName, LabUsers = STUFF((SELECT N'•' + lu.LabUserName
FROM dbo.LabUsers AS lu
WHERE lu.LabID = l.LabID
ORDER BY lu.LabUserName
FOR XML PATH(''),
TYPE).value(N'./text()[1]', N'nvarchar(max)'), 1, 1, N'')
FROM dbo.Labs AS l;
I see absolutely no reason to store this in the table, since you can always generate the code at runtime when you run a query. If you store it in the table, then you have to update it every single time you change any row in the table.
However, if I can't convince you not to do this (it really is bad to store redundant data like this), you can try this way:
;WITH x AS
(
SELECT l.LabID, l.LabName, x = STUFF((SELECT N'•' + lu.LabUserName
FROM dbo.LabUsers AS lu
WHERE lu.LabID = l.LabID
ORDER BY lu.LabUserName
FOR XML PATH(''),
TYPE).value(N'./text()[1]', N'nvarchar(max)'), 1, 1, N'')
FROM dbo.Labs AS l
)
UPDATE l
SET LabUsers = x.x
FROM dbo.Labs AS l
INNER JOIN x ON l.LabID = x.LabID;
As for the performance tests, I'd compare the above version with this variation:
SELECT l.LabID, l.LabName, LabUsers = STUFF((SELECT N'•' + lu.LabUserName
FROM dbo.LabUsers AS lu
WHERE lu.LabID = l.LabID
ORDER BY lu.LabUserName
FOR XML PATH('')), 1, 1, '')
FROM dbo.Labs AS l;
On my system I see the initial version at the top of this answer to be far more expensive. Also note that stuffing (no pun intended) these approaches into a user-defined function will bring it closer to the concatenation method #RThomas proposed.

Answer 2

Give this a try
SELECT LabName ,
STUFF(( SELECT ',' + LabUsers.LabUserName
FROM dbo.LabUsers
WHERE LabUsers.LabID = Labs.LabID
ORDER BY LabName
FOR
XML PATH('')
), 1, 1, '') AS Labusers
FROM dbo.Labs
ORDER BY LabName
The FOR XML PATH('') concatenates your strings together into one XML result and the STUFF puts a "nothing" character at the first character, e.g. wipes out the unneeded first comma.

Answer 3

Another way to do it is set up a UDF that returns all the lab users as a single string like this:
CREATE FUNCTION LabUserString
(
#pLabId Int
)
RETURNS NVarChar(Max)
AS
BEGIN
Declare #pResult NVarChar(Max)
SELECT #pResult = COALESCE(#pResult + N'•', '') + [LabUserName]
FROM LabUsers WHERE LabId = #pLabId
Return #pResult
END
And then a query like this:
Select LabID, LabName, dbo.LabUserString(LabID) AS LabUsers FROM Labs