The [name]
field in sys.objects
will contain only the actual name (i.e. trg
), not including the schema (i.e. dbo
in this case) or any text qualifiers (i.e. [
and ]
in this case).
AND, you don’t specify the table name for DROP TRIGGER
since the trigger is an object by itself (unlike indexes). So you need to remove the ON
clause (which is only used with DDL and Logon triggers).
IF EXISTS (SELECT * FROM sys.objects WHERE [name] = N'trg' AND [type] = 'TR')
BEGIN
DROP TRIGGER [dbo].[trg];
END;
Please note that you should prefix the object name string literal with an N
since the [name]
field is a sysname
datatype which equates to NVARCHAR(128)
.
If you did want to incorporate the schema name, you could use the OBJECT_ID()
function which does allow for schema names and text qualifiers (you will then need to match against object_id
instead of name
):
IF EXISTS (SELECT * FROM sys.objects WHERE [object_id] = OBJECT_ID(N'[dbo].[trg]')
AND [type] = 'TR')
BEGIN
DROP TRIGGER [dbo].[trg];
END;
And to simplify, since the object name needs to be unique within the schema, you really only need to test for its existence. If for some reason a different object type exists with that name, the DROP TRIGGER
will fail since that other object is, well, not a trigger ;-). Hence, I use the following:
IF (OBJECT_ID(N'[dbo].[trg]') IS NOT NULL)
BEGIN
DROP TRIGGER [dbo].[trg];
END;