SQL Server 2005 unique key with null value

倖福魔咒の 提交于 2019-12-05 05:25:59
Thomas

This is a long time complaint about SQL Server's Unique constraints/indexes. The best solution is to create a view with schemabinding and then put a unique index on that column:

Create View dbo.MyUniqueColView
With SchemaBinding
As
Select MyColToBeUnique
From MyTable
Where MyColToBeUnique Is Not Null

GO

Create Unique Clustered Index IX_MyTable_MyColToBeUnique On MyUniqueColView ( MyColToBeUnique )

You cannot enforce this with a table level constraint in SQL Server 2005 but you can create a view with definition SELECT Col FROM t WHERE Col IS NOT NULL and then create a unique clustered index on that.

In SQL Server 2008, you could create a filtered index on the values that are not null - unfortunately, that is a new feature in the 2008 release, so in 2005, you're not able to do anything like that, I'm afraid.

Always an annoyance of mine. I prefer the "other" interpretation of the standard.

See Creating a unique constraint that ignores nulls in SQL Server. It covers three different approaches.

  1. Solution 1: Filtered indexes.

    This seems pretty common, judging from other replies. I have not tried this approach -- it does require SQL Server 2008 though.

  2. Solution 2: Constraint on computed column

    I have used this, and it does work. However, relationships may need to be adjusted and/or secondary non-unique covering indexed added (one for unique constraint, one for index cover).

  3. Solution 3: UDF Check Constraint (Not recommended)

    Ahh, the power of TSQL (read: something else to deal with). I have not gone down this route.

Happy coding.

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!