SQL Server 2005 drop column with constraints

Question

I have a column with a \"DEFAULT\" constraint. I\'d like to create a script that drops that column.

The problem is that it returns this error:

Msg 50         


        

Answer 1

I also think it's a shortcoming in SQL server to not have a cascading drop available. I worked my way around it by querying the system tables in the same way as other people described here:

• INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE only lists foreign keys, primary keys and unique constraints.
• The only way to look for default constraints is to look for them in sys.default_constraints.
• what hasn't been mentioned here yet, is that indexes also make dropping a column fail, so you also need to drop all indexes that use your column before you can proceed with dropping a column.

The resulting script is not pretty, but I put it in a stored procedure to be able to reuse it:

CREATE PROCEDURE DropColumnCascading @tablename nvarchar(500), @columnname nvarchar(500)
AS

SELECT CONSTRAINT_NAME, 'C' AS type
INTO #dependencies
FROM INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE WHERE TABLE_NAME = @tablename AND COLUMN_NAME = @columnname

INSERT INTO #dependencies
select d.name, 'C'
from sys.default_constraints d
join sys.columns c ON c.column_id = d.parent_column_id AND c.object_id = d.parent_object_id
join sys.objects o ON o.object_id = d.parent_object_id
WHERE o.name = @tablename AND c.name = @columnname

INSERT INTO #dependencies
SELECT i.name, 'I'
FROM sys.indexes i
JOIN sys.index_columns ic ON ic.index_id = i.index_id and ic.object_id=i.object_id
JOIN sys.columns c ON c.column_id = ic.column_id and c.object_id=i.object_id
JOIN sys.objects o ON o.object_id = i.object_id
where o.name = @tableName AND i.type=2 AND c.name = @columnname AND is_unique_constraint = 0

DECLARE @dep_name nvarchar(500)
DECLARE @type nchar(1)

DECLARE dep_cursor CURSOR
FOR SELECT * FROM #dependencies

OPEN dep_cursor

FETCH NEXT FROM dep_cursor 
INTO @dep_name, @type;

DECLARE @sql nvarchar(max)

WHILE @@FETCH_STATUS = 0
BEGIN
    SET @sql = 
        CASE @type
            WHEN 'C' THEN 'ALTER TABLE [' + @tablename + '] DROP CONSTRAINT [' + @dep_name + ']'
            WHEN 'I' THEN 'DROP INDEX [' + @dep_name + '] ON dbo.[' + @tablename + ']'
        END
    print @sql
    EXEC sp_executesql @sql
    FETCH NEXT FROM dep_cursor 
    INTO @dep_name, @type;
END

DEALLOCATE dep_cursor

DROP TABLE #dependencies

SET @sql = 'ALTER TABLE [' + @tablename + '] DROP COLUMN [' + @columnname + ']'

print @sql
EXEC sp_executesql @sql

Answer 2

The answer from pvolders was just what I needed but it missed statistics which were causing and error. This is the same code, minus creating a stored procedure, plus enumerating statistics and dropping them. This is the best I could come up with so if there is a better way to determine what statistics need to be dropped please add.

DECLARE @tablename nvarchar(500), 
        @columnname nvarchar(500)

SELECT  @tablename = 'tblProject',
        @columnname = 'CountyKey'


SELECT CONSTRAINT_NAME, 'C' AS type
INTO #dependencies
FROM INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE WHERE TABLE_NAME = @tablename AND COLUMN_NAME = @columnname

INSERT INTO #dependencies
select d.name, 'C'
from sys.default_constraints d
join sys.columns c ON c.column_id = d.parent_column_id AND c.object_id = d.parent_object_id
join sys.objects o ON o.object_id = d.parent_object_id
WHERE o.name = @tablename AND c.name = @columnname

INSERT INTO #dependencies
SELECT i.name, 'I'
FROM sys.indexes i
JOIN sys.index_columns ic ON ic.index_id = i.index_id and ic.object_id=i.object_id
JOIN sys.columns c ON c.column_id = ic.column_id and c.object_id=i.object_id
JOIN sys.objects o ON o.object_id = i.object_id
where o.name = @tableName AND i.type=2 AND c.name = @columnname AND is_unique_constraint = 0

INSERT INTO #dependencies
SELECT s.NAME, 'S'
FROM sys.stats AS s
INNER JOIN sys.stats_columns AS sc 
    ON s.object_id = sc.object_id AND s.stats_id = sc.stats_id
INNER JOIN sys.columns AS c 
    ON sc.object_id = c.object_id AND c.column_id = sc.column_id
WHERE s.object_id = OBJECT_ID(@tableName)
AND c.NAME = @columnname
AND s.NAME LIKE '_dta_stat%'

DECLARE @dep_name nvarchar(500)
DECLARE @type nchar(1)

DECLARE dep_cursor CURSOR
FOR SELECT * FROM #dependencies

OPEN dep_cursor

FETCH NEXT FROM dep_cursor 
INTO @dep_name, @type;

DECLARE @sql nvarchar(max)

WHILE @@FETCH_STATUS = 0
BEGIN
    SET @sql = 
        CASE @type
            WHEN 'C' THEN 'ALTER TABLE [' + @tablename + '] DROP CONSTRAINT [' + @dep_name + ']'
            WHEN 'I' THEN 'DROP INDEX [' + @dep_name + '] ON dbo.[' + @tablename + ']'
            WHEN 'S' THEN 'DROP STATISTICS [' + @tablename + '].[' + @dep_name + ']'
        END
    print @sql
    EXEC sp_executesql @sql
    FETCH NEXT FROM dep_cursor 
    INTO @dep_name, @type;
END

DEALLOCATE dep_cursor

DROP TABLE #dependencies

SET @sql = 'ALTER TABLE [' + @tablename + '] DROP COLUMN [' + @columnname + ']'

print @sql
EXEC sp_executesql @sql

Answer 3

This query finds default constraints for a given table. It aint pretty, I agree:

select 
    col.name, 
    col.column_id, 
    col.default_object_id, 
    OBJECTPROPERTY(col.default_object_id, N'IsDefaultCnst') as is_defcnst, 
    dobj.name as def_name
from sys.columns col 
    left outer join sys.objects dobj 
        on dobj.object_id = col.default_object_id and dobj.type = 'D' 
where col.object_id = object_id(N'dbo.test') 
and dobj.name is not null

[EDIT] Updated per Julien N's comment

Answer 4

> select CONSTRAINT_NAME from INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE
> WHERE TABLE_NAME = '<tablename>' AND COLUMN_NAME = 'IsClosed'

It's not the right solution as it is explained here : http://msdn.microsoft.com/en-us/library/aa175912.aspx that :

Unfortunately, the name of the column default constraint isn't kept in the ANSI COLUMNS view, so you must go back to the system tables to find the name

The only way I found to get the name of the DEFAULT constraint is this request :

select  
    t_obj.name              as TABLE_NAME
    ,c_obj.name             as CONSTRAINT_NAME
    ,col.name               as COLUMN_NAME

from    sysobjects  c_obj
join    sysobjects  t_obj on c_obj.parent_obj = t_obj.id  
join    sysconstraints con on c_obj.id  = con.constid
join    syscolumns  col on t_obj.id = col.id
            and con.colid = col.colid
where
    c_obj.xtype = 'D'

Am I the only one to find it crazy to be unable to delete easily a constraint that only concerns the columns I'm trying to drop ?
I need to execute a request with 3 joins just to get the name...

Answer 5

Just to build on Jeremy Stein's answer, I created a stored procedure for this, and set it up so it can be used to delete a column that has or does not have default constraints. It's not real efficient since it's querying sys.columns twice, but it works.

CREATE PROCEDURE [dbo].[RemoveColumnWithDefaultConstraints] 
    -- Add the parameters for the stored procedure here
    @tableName nvarchar(max), 
    @columnName nvarchar(max)
AS
BEGIN
    -- SET NOCOUNT ON added to prevent extra result sets from
    -- interfering with SELECT statements.
    SET NOCOUNT ON;

    DECLARE @ConstraintName nvarchar(200)
    SELECT @ConstraintName = Name 
    FROM SYS.DEFAULT_CONSTRAINTS 
    WHERE PARENT_OBJECT_ID = OBJECT_ID(@tableName) 
        AND PARENT_COLUMN_ID = (SELECT column_id FROM sys.columns WHERE NAME = (@columnName) 
        AND object_id = OBJECT_ID(@tableName))
    IF @ConstraintName IS NOT NULL
        EXEC('ALTER TABLE ' + @tableName + ' DROP CONSTRAINT ' + @ConstraintName)

    IF EXISTS(SELECT * FROM sys.columns WHERE Name = @columnName  
        AND Object_ID = Object_ID(@tableName))
        EXEC('ALTER TABLE ' + @tableName + ' DROP COLUMN ' + @columnName)       
END
GO

Answer 6

I just ran into this. You can delete the column with constraints using MSSQL design view. Right click on the column you want to be dropped (with or without constraints) and you are able to delete this without any problems. Ha.. I looked stupid already.