How to overcome "iif" limitations with subqueries? - sql-server

Question

I have 2 tables (A and B). I want to implement such logic to table B:
[Bank] AS (iif( Some_Statement,
(SELECT [Money] FROM [ATM] WHERE [B].[Currency] = [A].[Currency] ),
[NoMoney]
)
)
I get an error: Subqueries are not allowed in this context. Only scalar expressions are allowed.
Is there a way to implement such logic on creation of the tables? It doesn't look hard.

Answer 1

After reading the discussion in comments I get the feeling, that this was going the wrong direction. Might be because of your very direct question about IIF()...
If I get this correctly you try to add a computed column to you table. Something along this:
DECLARE #tblTest TABLE (ID INT IDENTITY
,SomeValue VARCHAR(100)
,[Test] AS IIF(SomeValue IS NOT NULL, CONCAT(SomeValue,'Blah'), 'Default if null'));
INSERT INTO #tblTest(SomeValue) VALUES('Test');
INSERT INTO #tblTest(SomeValue) VALUES(NULL);
SELECT * FROM #tblTest;
The result
ID SomeValue Test
1 Test TestBlah
2 NULL Default if null
Now you want to get the value for the computed column not just from some simple scalar computation, but you want to pick it from a table.
Here I will try to simulate your issue. Next time it is up to you to do this yourself. Providing DDL, sample data and the expected output together with your own attempts is the best chance to get the answer you are waiting for.
This is the table B from where you want to get the value.
Later we will ask for 'Test' and not for 'xyz'.
CREATE TABLE tblB (ID INT IDENTITY
,SomeResultColumn VARCHAR(100)
,SomeConditionColumn VARCHAR(100));
INSERT INTO tblB(SomeResultColumn,SomeConditionColumn) VALUES('not wanted','xyz')
,('wanted','Test');
GO
--This is what you are trying to do, but I out-commented it because of the error you've got.
--A computed column does not allow for a SELECT. This is not bound to the IIF()
--CREATE TABLE tblA (ID INT IDENTITY
-- ,SomeValue VARCHAR(100)
-- ,[Test] AS IIF(SomeValue IS NOT NULL,(SELECT b.SomeResultColumn FROM tblB b WHERE b.SomeConditionColumn=SomeValue),'Default if null'));
--GO
--But what we can do - and the error message is telling so - provide a scalar expression:
--A Scalar Function is exactly this:
CREATE FUNCTION dbo.GetMyComputedColumn(#Condition VARCHAR(100))
RETURNS VARCHAR(100) AS
BEGIN
RETURN (SELECT b.SomeResultColumn --You might use `TOP 1` to ensure a scalar result
FROM tblB b
WHERE b.SomeConditionColumn=#Condition);
END
GO
--We can use this function in the IIF():
CREATE TABLE tblA (ID INT IDENTITY
,SomeValue VARCHAR(100)
,[Test] AS IIF(SomeValue IS NOT NULL,dbo.GetMyComputedColumn(SomeValue),'Default if null'));
GO
INSERT INTO tblA(SomeValue) VALUES('Test');
INSERT INTO tblA(SomeValue) VALUES(NULL);
SELECT * FROM tblA;
The result
ID SomeValue Test
1 Test wanted
2 NULL Default if null
Should you do this?
The question, if this is a good idea, is something completely different.
Besides the fact, that scalar functions are known as bad performers, the main question is: WHY?
Do you need a persistant default value? In this case a trigger will be a better choice... Or you can use an insert statement with the computed value directly. I'm afraid you are mixing the concepts of computed columns and a default constraint. The computed column cannot be changed...
If you want to compute this value whenever you fetch data from this table, it was much better to use a VIEW or an iTVF, where you simply join the needed value to your result set.

Answer 2

If what you are saying is that you want a value from table "B" IF it matches on a value from your main table "A" and if there is no matching value in "B", use a literal as a default, try this:
select
Bank,
coalesce( (SELECT Money FROM ATM WHERE Currency = A.Currency), 'NoMoney' ) as 'Type'
from
ATM A
where
....
If the select of [Money] returns null, the literal 'NoMoney' will be used.
If need be, you can have a select on both sides of the coalesce, the right need not be a literal.