Atomically INSERT+SELECT or SELECT row (based on UNIQUE column) - sql

Question

My table is created as follows:
CREATE TABLE StackFunctionID
(
m_FunctionID int PRIMARY KEY IDENTITY,
m_GroupID int DEFAULT 0 NOT NULL,
m_Function varchar(256) UNIQUE NOT NULL,
);
GO
CREATE TRIGGER INSERT_MakeGroupID ON StackFunctionID AFTER INSERT AS
BEGIN
SET NOCOUNT ON;
UPDATE StackFunctionID
SET StackFunctionID.m_GroupID = INSERTED.m_FunctionID
FROM StackFunctionID
INNER JOIN INSERTED ON (StackFunctionID.m_FunctionID = INSERTED.m_FunctionID)
END;
Now I'm trying to write a query for Sql Server 2008 that will
If m_Function does not exist, create it, and return m_FunctionID, m_GroupID
If m_Function exists, return m_FunctionID, m_GroupID
Perform it atomically
After much trying, this is what worked for the last years:
DECLARE #Results TABLE(m_FunctionID INT, m_GroupID INT); -- trash use
DECLARE #dummy1 int; -- trash use
MERGE INTO StackFunctionID WITH(HOLDLOCK) AS Target -- WITH(HOLDLOCK) for thread safety
USING (VALUES(null)) AS dummy2(dummy3) -- syntax requires USING
ON Target.m_Function = #m_Function
WHEN MATCHED THEN
UPDATE SET #dummy1 = 0 -- syntax requires something in 'WHEN MATCHED', which is needed for OUTPUT
WHEN NOT MATCHED THEN
INSERT(m_Function) VALUES(#m_Function)
OUTPUT inserted.m_FunctionID, inserted.m_GroupID
INTO #Results; -- Having a trigger requires 'INTO'
SELECT * FROM #Results -- Having a trigger requires 'INTO'
However, there was a bug: when a row is INSERTED, OUTPUT returns m_GroupID=0
Questions:
How do I always get correct m_GroupID ?
Is it possible to achieve the goal with a simpler query? (because current one is really hard to understand with all those "syntax requirements")

Answer 1

Relying on UNIQUE m_Function use TRY CATCH to INSERT and then SELECT.
Demo
CREATE TABLE StackFunctionID
(
m_FunctionID int PRIMARY KEY IDENTITY,
m_GroupID int DEFAULT 0 NOT NULL,
m_Function varchar(256) UNIQUE NOT NULL,
);
INSERT StackFunctionID(m_Function)
VALUES ('a'), ('b');
SELECT * FROM StackFunctionID;
DECLARE #m_Function varchar(256) = 'a';
BEGIN TRY
INSERT StackFunctionID(m_Function)
VALUES ( #m_Function);
END TRY
BEGIN CATCH
PRINT 'Already exists';
END CATCH
SELECT m_FunctionID, m_GroupID
FROM StackFunctionID
WHERE m_Function = #m_Function;