如果未提供列列表，则对具有RowVersion列的表执行INSERT操作需要一个空值。 为什么？

Question

Working a project where we're considering adding a rowversion column to a bunch of tables. Seems like it wouldn't be a big deal because SQL Server assigns the value, so we wouldn't need to modify existing stored procedures, of which there are many. Not that easy, though, because the INSERT requires either a column list or a null value. So if a rowversion column were added to a table, any stored procedure written without the column list would fail.

Below is sample code from the Microsoft docs with some modifications to show what I mean:

CREATE TABLE MyTest   
(
     myKey int PRIMARY KEY,
     myValue int, 
     RV rowversion
);  
GO  

--Works fine 
INSERT INTO MyTest (myKey, myValue) VALUES (1, 0);  
GO  

--Fails with error: Column name or number of supplied values does not match 
--table definition
INSERT INTO MyTest VALUES (2, 0); 

--Works fine. Weird
INSERT INTO MyTest VALUES (2, 0, NULL);
GO  

I don't understand why it requires a NULL value if a column list isn't supplied. Since SQL Server calculates the value, I don't see why it would throw off the mapping to the table definition, and it makes the code confusing because it looks as if you're giving the field a null value when that's not what happens.

Does anybody know the reason for this peculiar behavior?

Answer 1

The column-order is only predefined, when you provide all values. Otherwise you could also provide the Parameters in the wrong order. So if you want the to skip Parameters, you have to provide a Columnlist, to give a hint, how the values should be interpreted, wether there is a computed value involved, or not.