Question

UPDATE(simplified problem, removed C# from the issue)

How can I write an UPSERT that can recognize when two rows are the same in the following case…

See how there’s a \b [backspace] encoded there (the weird little character)? SQL sees these as the same. While my UPSERT sees this as new data and attempts an INSERT where there should be an UPDATE.

---

//UPSERT
    INSERT INTO [table]
    SELECT [col1] = @col1, [col2] = @col2, [col3] = @col3, [col4] = @col4
    FROM [table]
    WHERE NOT EXISTS
       -- race condition risk here?
       ( SELECT 1 FROM [table] 
       WHERE 
            [col1] = @col1 
        AND [col2] = @col2
        AND [col3] = @col3)

    UPDATE [table]
        SET [col4] = @col4
        WHERE 
        [col1] = @col1 
        AND [col2] = @col2
        AND [col3] = @col3

Answer 1

After hours of tinkering it turns out I’ve been on a wild goose chase. The problem is very simple. I pulled my UPSERT from a popular SO post. The code is no good. The select will sometimes return > 1 rows on INSERT. Thereby attempting to insert a row, then insert the same row again.

The fix is to remove FROM

    //UPSERT
    INSERT INTO [table]
    SELECT [col1] = @col1, [col2] = @col2, [col3] = @col3, [col4] = @col4
    --FROM [table] (Dont use FROM..not a race condition, just a bad SELECT)
    WHERE NOT EXISTS
       ( SELECT 1 FROM [table] 
       WHERE 
            [col1] = @col1 
        AND [col2] = @col2
        AND [col3] = @col3)

    UPDATE [table]
        SET [col4] = @col4
        WHERE 
        [col1] = @col1 
        AND [col2] = @col2
        AND [col3] = @col3

Problem is gone.

Thanks to all of you.