Registering User-Defined Types in SQL Server
In order to use a user-defined type (UDT) in Microsoft SQL Server, you must register it. Registering a UDT involves registering the assembly and creating the type in the database in which you wish to use it. UDTs are scoped to a single database, and cannot be used in multiple databases unless the identical assembly and UDT are registered with each database. Once the UDT assembly is registered and the type created, you can use the UDT in Transact-SQL and in client code. For more information, see CLR User-Defined Types.
The easiest way to deploy your UDT is by using Microsoft Visual Studio. For more complex deployment scenarios and the greatest flexibility, however, use Transact-SQL as discussed later in this topic.
Follow these steps to create and deploy a UDT using Visual Studio:
Create a new Database project in the Visual Basic or Visual C# language nodes.
Add a reference to the SQL Server database that will contain the UDT.
Add a User-Defined Type class.
Write code to implement the UDT.
From the Build menu, select Deploy. This registers the assembly and creates the type in the SQL Server database.
The Transact-SQL CREATE ASSEMBLY syntax is used to register the assembly in the database in which you wish to use the UDT. It is stored internally in database system tables, not externally in the file system. If the UDT is dependent on external assemblies, they too must be loaded into the database. The CREATE TYPE statement is used to create the UDT in the database in which it is to be used. For more information, see CREATE ASSEMBLY (Transact-SQL) and CREATE TYPE (Transact-SQL).
Using CREATE ASSEMBLY
The CREATE ASSEMBLY syntax registers the assembly in the database in which you wish to use the UDT. Once the assembly is registered, it has no dependencies.
Creating multiple versions of the same assembly in a given database is not allowed. However, it is possible to create multiple versions of the same assembly based on culture in a given database. SQL Server distinguishes multiple culture versions of an assembly by different names as registered in the instance of SQL Server. For more information, see "Creating and Using Strong-Named Assemblies" in the .NET Framework SDK.
When CREATE ASSEMBLY is executed with the SAFE or EXTERNAL_ACCESS permission sets, the assembly is checked to make sure that it is verifiable and type safe. If you omit specifying a permission set, SAFE is assumed. Code with the UNSAFE permission set is not checked. For more information about assembly permission sets, see Designing Assemblies.
The following Transact-SQL statement registers the Point assembly in SQL Server in the AdventureWorks2008R2 database, with the SAFE permission set. If the WITH PERMISSION_SET clause is omitted, the assembly is registered with the SAFE permission set.
USE AdventureWorks2008R2; CREATE ASSEMBLY Point FROM '\\ShareName\Projects\Point\bin\Point.dll' WITH PERMISSION_SET = SAFE;
The following Transact-SQL statement registers the assembly using <assembly_bits> argument in the FROM clause. This varbinary value represents the file as a stream of bytes.
USE AdventureWorks2008R2; CREATE ASSEMBLY Point FROM 0xfeac4 … 21ac78;
Using CREATE TYPE
Once the assembly is loaded into the database, you can then create the type using the Transact-SQL CREATE TYPE statement. This adds the type to the list of available types for that database. The type has database scope and the type can only be used in the database in which it was created. If the UDT already exists in the database, the CREATE TYPE statement fails with an error.
The CREATE TYPE syntax is also used for creating native SQL Server alias data types, and is intended to replace sp_addtype as a means of creating alias data types. Some of the optional arguments in the CREATE TYPE syntax refer to creating UDTs, and are not applicable to creating alias data types (such as base type).
Beginning with SQL Server 2005, on a SQL Server database with a compatibility level of "80" you cannot create managed user-defined types, stored procedures, functions, aggregates, or triggers. To take advantage of these CLR integration features of SQL Server, you must use the sp_dbcmptlevel (Transact-SQL) stored procedure to set the database compatibility level to "100".
For more information, see CREATE TYPE (Transact-SQL).
The following Transact-SQL statement creates the Point type. The EXTERNAL NAME is specified using the two-part naming syntax of AssemblyName.UDTName.
CREATE TYPE dbo.Point EXTERNAL NAME Point.[Point];
The DROP TYPE statement removes a UDT from the current database. Once a UDT is dropped, you can use the DROP ASSEMBLY statement to drop the assembly from the database.
The DROP TYPE statement does not execute in the following situations:
Tables in the database that contain columns defined using the UDT.
Functions, stored procedures, or triggers that use variables or parameters of the UDT, created in the database with the WITH SCHEMABINDING clause.
The following Transact-SQL must execute in the following order. First the table which references the Point UDT must be dropped, then the type, and finally the assembly.
DROP TABLE dbo.Points; DROP TYPE dbo.Point; DROP ASSEMBLY Point;
Finding UDT Dependencies
If there are dependent objects, such as tables with UDT column definitions, the DROP TYPE statement fails. It also fails if there are functions, stored procedures, or triggers created in the database using the WITH SCHEMABINDING clause, if these routines use variables or parameters of the user-defined type. You must first drop all dependent objects, and then execute the DROP TYPE statement.
The following Transact-SQL query locates all of the columns and parameters that use a UDT in the AdventureWorks2008R2database.
USE Adventureworks2008R2; SELECT o.name AS major_name, o.type_desc AS major_type_desc , c.name AS minor_name, c.type_desc AS minor_type_desc , at.assembly_class FROM ( SELECT object_id, name, user_type_id, 'SQL_COLUMN' AS type_desc FROM sys.columns UNION ALL SELECT object_id, name, user_type_id, 'SQL_PROCEDURE_PARAMETER' FROM sys.parameters ) AS c; JOIN sys.objects AS o ON o.object_id = c.object_id; JOIN sys.assembly_types AS at ON at.user_type_id = c.user_type_id;
You cannot modify a UDT once it is created in a SQL Server database, although you can alter the assembly on which the type is based. In most cases, you must remove the UDT from the database with the Transact-SQL DROP TYPE statement, make changes to the underlying assembly, and reload it using the ALTER ASSEMBLY statement. You then need to re-create the UDT and any dependent objects.
The ALTER ASSEMBLY statement is used after you have made changes to the source code in your UDT assembly and recompiled it. It copies the .dll file to the server and rebinds to the new assembly. For the complete syntax, see ALTER ASSEMBLY (Transact-SQL).
The following Transact-SQL ALTER ASSEMBLY statement reloads the Point.dll assembly from the specified location on disk.
ALTER ASSEMBLY Point FROM '\\Projects\Point\bin\Point.dll';
Using ALTER ASSEMBLY to Add Source Code
The ADD FILE clause in the ALTER ASSEMBLY syntax is not present in CREATE ASSEMBLY. You can use it to add source code or any other files associated with an assembly. The files are copied from their original locations and stored in system tables in the database. This ensures that you always have source code or other files on hand should you ever need to re-create or document the current version of the UDT.
The following Transact-SQL ALTER ASSEMBLY statement adds the Point.cs class source code for the Point UDT. This copies the text contained in the Point.cs file and stores it in the database under the name "PointSource".
ALTER ASSEMBLY Point ADD FILE FROM '\\Projects\Point\Point.cs' AS PointSource;
Assembly information is stored in the sys.assembly_files table in the database where the assembly has been installed. The sys.assembly_files table contains the following columns.
You can use the CAST or CONVERT function to convert the contents of the content column to readable text. The following query converts the contents of the Point.cs file to readable text, using the name in the WHERE clause to restrict the result set to a single row.
SELECT CAST(content AS varchar(8000)) FROM sys.assembly_files WHERE name='PointSource';
If you copy and paste the results in to a text editor, you will see that the line breaks and spaces that existed in the original have been preserved.
When planning your implementation of UDTs, consider which methods are needed in the UDT assembly itself, and which methods should be created in separate assemblies and implemented as user-defined functions or stored procedures. Separating methods into separate assemblies allows you to update code without affecting data that may be stored in a UDT column in a table. You can modify UDT assemblies without dropping UDT columns and other dependent objects only when the new definition can read the former values and the signature of the type does not change.
Separating procedural code that may change from the code required to implement the UDT greatly simplifies maintenance. Including only code that is necessary for the UDT to function, and keeping your UDT definitions as simple as possible, reduces the risk that the UDT itself may need to be dropped from the database for code revisions or bug fixes.
The Currency UDT and Currency Conversion Function
The Currency UDT in the AdventureWorks2008R2 sample database provides a useful example of the recommended way to structure a UDT and its associated functions. The Currency UDT is used for handling money based on the monetary system of a particular culture, and allows for storage of different currency types, such as dollars, euros, and so forth. The UDT class exposes a culture name as a string, and an amount of money as a decimal data type. All of the necessary serialization methods are contained within the assembly defining the class. The function that implements currency conversion from one culture to another is implemented as an external function named ConvertCurrency, and this function is located in a separate assembly. The ConvertCurrency function does its work by retrieving the conversion rate from a table in the AdventureWorks2008R2 database. If the source of the conversion rates should ever change, or if there should be any other changes to the existing code, the assembly can be easily modified without affecting the Currency UDT.
The code listing for the Currency UDT and ConvertCurrency functions can be found by installing the common language runtime (CLR) samples. For more information, see Considerations for Installing SQL Server Samples and Sample Databases.
Using UDTs Across Databases
UDTs are by definition scoped to a single database. Therefore, a UDT defined in one database cannot be used in a column definition in another database. In order to use UDTs in multiple databases, you must execute the CREATE ASSEMBLY and CREATE TYPE statements in each database on identical assemblies. Assemblies are considered identical if they have the same name, strong name, culture, version, permission set, and binary contents.
Once the UDT is registered and accessible in both databases, you can convert a UDT value from one database for use in another. Identical UDTs can be used across databases in the following scenarios:
Calling stored procedure defined in different databases.
Querying tables defined in different databases.
Selecting UDT data from one database table UDT column and inserting it into a second database with an identical UDT column.
In these situations, any conversion required by the server occurs automatically. You are not able to perform the conversions explicitly using the Transact-SQL CAST or CONVERT functions.
Note that you do not need to take any action for using UDTs when SQL Server Database Engine creates work tables in the tempdb system database. This includes the handling of cursors, table variables, and user-defined table-valued functions that include UDTs and that transparently make use of tempdb. However, if you explicitly create a temporary table in tempdb that defines a UDT column, then the UDT must be registered in tempdb the same way as for a user database.