Triggers Mutually Exclusive Boolean Fields - sql

Question

I have two boolean (bit) fields that I want to remain mutually exclusive. They can both be false, but if one gets set to true the other must be set to false.
I tried to create 2 triggers but wound up creating an infinite loop (that stopped after 32 iterations). Apparently they trigger each other. Here is the code I tried:
CREATE OR ALTER TRIGGER dbo.TriggerNameA
ON dbo.TableName
AFTER UPDATE
AS
IF UPDATE(FieldA)
BEGIN
UPDATE dbo.TableName
SET FieldB =
CASE WHEN dbo.TableName.FieldA = 1
THEN 0
ELSE 1
END
FROM dbo.TableName INNER JOIN inserted ON dbo.TableName.ID = inserted.ID
END;
;
GO
CREATE OR ALTER TRIGGER dbo.TriggerNameB
ON dbo.TableName
AFTER UPDATE
AS
IF UPDATE(FieldB)
BEGIN
UPDATE dbo.TableName
SET FieldA =
CASE WHEN dbo.TableName.FieldB = 1
THEN 0
ELSE 1
END
FROM dbo.TableName INNER JOIN inserted ON dbo.TableName.ID = inserted.ID
END;
;
GO
Any idea how to avoid the infinite loop?

Answer 1

Using a Check constraint removes the need for a trigger:
SQL Fiddle
MS SQL Server 2017 Schema Setup:
create table MyTable
(
ID int Identity,
FieldA bit,
FieldB bit,
CONSTRAINT Chk_Fields CHECK (FieldA <> 1 OR FieldB <> 1)
)
INSERT INTO MyTable(FieldA, FieldB)
VALUES (0,0),(0,1),(1,0)
Query 1:
select *
from MyTable
Results:
| ID | FieldA | FieldB |
|----|--------|--------|
| 1 | false | false |
| 2 | false | true |
| 3 | true | false |
Query 2:
INSERT INTO MyTable(FieldA, FieldB)
VALUES (1,1),(0,1),(1,0)
Results:
The INSERT statement conflicted with the CHECK constraint "Chk_Fields". The conflict occurred in database "db_18_941bc9", table "dbo.MyTable".
Query 3:
UPDATE MyTable
SET FieldB = 1
WHERE ID = 3
Results:
The UPDATE statement conflicted with the CHECK constraint "Chk_Fields". The conflict occurred in database "db_18_941bc9", table "dbo.MyTable".
EDIT As #CharlieFace mentions in his comment, the equivalent check
constraint of (FieldA = 0 OR FieldB = 0) is more readable

Answer 2

Okay, I apologize for my previous answer, but I'm sure I got it now. Thanks to #DB<>Fiddle for pointing out that it still was experiencing recursion (or nesting, not sure).
So I used the trigger_nestlevel function. Tested and now (finally) it works:
create table MyTable
(
ID int Identity,
FieldA bit,
FieldB bit
)
go
CREATE OR ALTER TRIGGER dbo.Trigger1
ON dbo.MyTable
AFTER UPDATE
AS
IF TRIGGER_NESTLEVEL() = 1
BEGIN
IF UPDATE(FieldA)
BEGIN
UPDATE dbo.MyTable SET dbo.MyTable.FieldB = 0
FROM dbo.MyTable INNER JOIN inserted ON dbo.MyTable.ID = inserted.ID;
END;
END;
;
GO
CREATE OR ALTER TRIGGER dbo.Trigger2
ON dbo.MyTable
AFTER UPDATE
AS
IF TRIGGER_NESTLEVEL() = 1
BEGIN
IF UPDATE(FieldB)
BEGIN
UPDATE dbo.MyTable SET dbo.MyTable.FieldA = 0
FROM dbo.MyTable INNER JOIN inserted ON dbo.MyTable.ID = inserted.ID;
END;
END;
;
GO
INSERT INTO MyTable(FieldA, FieldB)
VALUES (0,0)
SELECT * FROM MyTable;
-- 0 0
UPDATE MyTable
SET FieldA = 1
WHERE ID = 1;
SELECT * FROM MyTable;
-- 1 0
UPDATE MyTable
SET FieldB = 1
WHERE ID = 1;
SELECT * FROM MyTable;
-- 0 1
Then I can go back and forth and they continue to switch the way I need. Whew!

Answer 3

Here's an alternative using an INSTEAD OF TRIGGER to ensure that on an UPDATE only one of the fields is set to 1. I have made a decision in the trigger that if FieldA in the virtual INSERTED table is set to 1 then I set FieldB to zero:
SetUp
create table MyTable
(
ID int Identity,
FieldA bit,
FieldB bit
)
CREATE TRIGGER TriggerNameA ON MyTable
INSTEAD OF UPDATE
AS
IF UPDATE(FieldA)
BEGIN
MERGE [MyTable] AS t1
USING (SELECT * FROM INSERTED) AS t2
ON t1.ID = t2.ID
WHEN MATCHED THEN
UPDATE SET t1.[FieldA] = t2.FieldA,
t1.[FieldB] = CASE WHEN t2.FieldA = 1 THEN 0 ELSE t2.FieldB END;
END
ELSE IF UPDATE(FieldB)
BEGIN
MERGE [MyTable] AS t1
USING (SELECT * FROM INSERTED) AS t2
ON t1.ID = t2.ID
WHEN MATCHED THEN
UPDATE SET t1.[FieldB] = t2.FieldB,
t1.[FieldA] = CASE WHEN t2.FieldB = 1 THEN 0 ELSE t2.FieldA END;
END;
-- Add Else for cases where FieldA & FieldB weren't updated if required
Result 1
INSERT INTO MyTable(FieldA, FieldB)
VALUES (0,0),(0,1),(1,0);
SELECT * FROM MyTable
ID FieldA FieldB
1 False False
2 False True
3 True False
Result 2
UPDATE MyTable
SET FieldB = 1
WHERE ID = 1
SELECT * FROM MyTable
ID FieldA FieldB
1 False True
2 False True
3 True False
Result 3
UPDATE MyTable
SET FieldB = 1
WHERE ID = 3
SELECT * FROM MyTable
ID FieldA FieldB
1 False True
2 False True
3 False True
Result 4
UPDATE MyTable
SET FieldA = 1
WHERE ID = 2
SELECT * FROM MyTable
ID FieldA FieldB
1 False True
2 True False
3 True False
DB<>Fiddle
Edited to Update fieldA or FieldB depending on what column is updated
in the UPDATE statement