As these are now system data types we needed to determine which tables and columns would be affected by the upgrade.
Come to find out, this was a pretty simple task. Using the system tables; Objects, Columns and Types we were able to pull a list together very quickly of the tables and columns that needed to be updated:
SCHEMA_NAME(t.schema_id) AS SchemaName
,o.Name AS TableName
,o.Type_Desc AS TableDescription
,o.Type AS TableType
,c.Object_id AS ColumnObjectID
,c.name AS ColumnName
,t.name AS TypeName
FROM sys.columns AS c
JOIN sys.types AS t ON c.user_type_id = t.user_type_id
JOIN sys.objects AS o ON c.Object_id = o.Object_ID
Where t.name = 'Date' -- Type Name here
and t.is_user_defined = 1 -- Yes, we want User Data Types only
and o.Type = 'U' -- User Table Types only
ORDER BY o.name,c.name;
Well, we had a gotcha here... The system not only saves which tables and columns that link to the UDT but all Indexes, Statistics, Stored Procedures, functions and views point to the UDT. To see all affected types just comment out the last AND statement in the above script.
What we ended up having to do was
- Create new UDT
- You can not just change a UDT, it has to be created and old one dropped
- Also means new UDT can not have same name, of course you could get creative here and add a couple of steps to the process so as to retain the same name.
- Dropping the Statistics
- Disabling and in some cases having to drop the individual indexes involved
- Then Altering the table
- Rebuilding or recreating the indexes
- Recreate the statistics
- Recompile the Stored Procedures and UDF's
- Recreate any affected Views
- Then you can drop the old UDT
It was strange, even though they still point to the 'date' UDT the functions still worked.
Seems to me, that the the UDT's aren't accomplishing their main objective and that is making it easier to maintain consistency across tables.
You have a Social Security Number of 11 digits (9 digits plus the two dashes). You created a UDT for SSN of char(11). This would allow you to have any tables with a SSN in it to be marked with this UDT and make sure that all are a consistent 11 digits.
Now, the government decides that they need to add a couple more digits to the mess. So we are tasked with finding and updating all SSN to this new length format. The UDT makes this very simple to locate all affected columns.
But, updating this column isn't as simple as changing the UDT. Instead, you have to some major changes through out the database(s).