best practices stored procedures and triggers

Using SS2000 SP4. We have a .NET app that accesses the sql server database.
I've tried to keep all access via stored procedures and views and the user
only has execute permissions on the sp and select on the views.
The contractors have put a trigger on one of the tables.
CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
FOR INSERT, UPDATE
AS
IF (Select bitCompPrimaryContact From Inserted) = 1
Begin
Update tblCompanyContacts Set bitCompPrimaryContact = 0
Where numCompanyId = (Select numCompanyId From Inserted)
And numContactID <> (Select numContactId From Inserted)
End
So, the user needs select and update permissions on the table. Since I
didn't want to give insert, update or delete permissions on any tables to the
user is there a way around it? What is a best practice approach. Do I just
not use update triggers?
Thanks,
--
Dan D.As long as you are using stored procedures you do not need to use this
trigger. Tell the contractors to put this code within the stored procedure
that updates the table tblCompanyContacts.
--
Keith Kratochvil
"Dan D." <DanD@.discussions.microsoft.com> wrote in message
news:D9A15C06-49C3-4F1A-8EB1-884159C0184F@.microsoft.com...
> Using SS2000 SP4. We have a .NET app that accesses the sql server
> database.
> I've tried to keep all access via stored procedures and views and the user
> only has execute permissions on the sp and select on the views.
> The contractors have put a trigger on one of the tables.
> CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
> FOR INSERT, UPDATE
> AS
> IF (Select bitCompPrimaryContact From Inserted) = 1
> Begin
> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> Where numCompanyId = (Select numCompanyId From Inserted)
> And numContactID <> (Select numContactId From Inserted)
> End
> So, the user needs select and update permissions on the table. Since I
> didn't want to give insert, update or delete permissions on any tables to
> the
> user is there a way around it? What is a best practice approach. Do I just
> not use update triggers?
> Thanks,
> --
> Dan D.|||That sounds good. Thanks Keith.
--
Dan D.
"Keith Kratochvil" wrote:
> As long as you are using stored procedures you do not need to use this
> trigger. Tell the contractors to put this code within the stored procedure
> that updates the table tblCompanyContacts.
> --
> Keith Kratochvil
>
> "Dan D." <DanD@.discussions.microsoft.com> wrote in message
> news:D9A15C06-49C3-4F1A-8EB1-884159C0184F@.microsoft.com...
> > Using SS2000 SP4. We have a .NET app that accesses the sql server
> > database.
> > I've tried to keep all access via stored procedures and views and the user
> > only has execute permissions on the sp and select on the views.
> >
> > The contractors have put a trigger on one of the tables.
> >
> > CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
> > FOR INSERT, UPDATE
> > AS
> > IF (Select bitCompPrimaryContact From Inserted) = 1
> > Begin
> > Update tblCompanyContacts Set bitCompPrimaryContact = 0
> > Where numCompanyId = (Select numCompanyId From Inserted)
> > And numContactID <> (Select numContactId From Inserted)
> > End
> >
> > So, the user needs select and update permissions on the table. Since I
> > didn't want to give insert, update or delete permissions on any tables to
> > the
> > user is there a way around it? What is a best practice approach. Do I just
> > not use update triggers?
> >
> > Thanks,
> > --
> > Dan D.
>
>|||I don't believe you don't need to grant any additional permissions on
the base table to users to execute a trigger. The "grant execute" on
the update or insert stored procedures should be enough.
Dan D. wrote:
> That sounds good. Thanks Keith.
> --
> Dan D.
>
> "Keith Kratochvil" wrote:
> > As long as you are using stored procedures you do not need to use this
> > trigger. Tell the contractors to put this code within the stored procedure
> > that updates the table tblCompanyContacts.
> >
> > --
> > Keith Kratochvil
> >
> >
> > "Dan D." <DanD@.discussions.microsoft.com> wrote in message
> > news:D9A15C06-49C3-4F1A-8EB1-884159C0184F@.microsoft.com...
> > > Using SS2000 SP4. We have a .NET app that accesses the sql server
> > > database.
> > > I've tried to keep all access via stored procedures and views and the user
> > > only has execute permissions on the sp and select on the views.
> > >
> > > The contractors have put a trigger on one of the tables.
> > >
> > > CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
> > > FOR INSERT, UPDATE
> > > AS
> > > IF (Select bitCompPrimaryContact From Inserted) = 1
> > > Begin
> > > Update tblCompanyContacts Set bitCompPrimaryContact = 0
> > > Where numCompanyId = (Select numCompanyId From Inserted)
> > > And numContactID <> (Select numContactId From Inserted)
> > > End
> > >
> > > So, the user needs select and update permissions on the table. Since I
> > > didn't want to give insert, update or delete permissions on any tables to
> > > the
> > > user is there a way around it? What is a best practice approach. Do I just
> > > not use update triggers?
> > >
> > > Thanks,
> > > --
> > > Dan D.
> >
> >
> >|||On Mon, 22 May 2006 06:52:02 -0700, Dan D. wrote:
>Using SS2000 SP4. We have a .NET app that accesses the sql server database.
>I've tried to keep all access via stored procedures and views and the user
>only has execute permissions on the sp and select on the views.
>The contractors have put a trigger on one of the tables.
>CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
>FOR INSERT, UPDATE
>AS
>IF (Select bitCompPrimaryContact From Inserted) = 1
>Begin
>Update tblCompanyContacts Set bitCompPrimaryContact = 0
>Where numCompanyId = (Select numCompanyId From Inserted)
>And numContactID <> (Select numContactId From Inserted)
>End
Hi Dan,
First, fire the contractors. This trigger will fail on any multi-row
insert or update. No-one should write triggers like that and be paid for
it!
>So, the user needs select and update permissions on the table.
Says who? As long as the table is owned by the same user that owns the
trigger, the user doesn't need any permissions on the table - see the
repro script below.
> Since I
>didn't want to give insert, update or delete permissions on any tables to the
>user is there a way around it? What is a best practice approach. Do I just
>not use update triggers?
In most cases, I would recommend having all objects (tables, views,
stored procedures, triggers, functions, ...) owned by dbo. Then, you can
set permissions exactly as required. Due to all objects having the same
owner, all statements from within stored procedures, triggers, and
functions will work. (Unless you're using dynamic SQL).
Here's a repro to show that a user doesn't need permission on the table
if a trigger changes data in the table.
First, run the script below while connected with an accoount with
sufficient rights to add databases and logins:
USE master
go
DROP DATABASE test
go
sp_droplogin @.loginame = 'test'
go
CREATE DATABASE test
go
sp_addlogin @.loginame = 'test',
@.passwd = 'test',
@.defdb = 'test'
go
USE test
go
sp_grantdbaccess @.loginame = 'test'
go
CREATE TABLE tab (a int NOT NULL,
b int DEFAULT NULL)
go
CREATE VIEW vw
AS
SELECT a, b
FROM tab
go
CREATE TRIGGER trig
ON tab
FOR INSERT, UPDATE
AS
IF @.@.ROWCOUNT > 0
BEGIN
UPDATE tab
SET b = a + 15
WHERE b IS NULL
AND EXISTS
(SELECT *
FROM inserted
WHERE inserted.a = tab.a)
END
go
CREATE PROC ins_tab (@.a int)
AS
INSERT INTO tab (a)
VALUES (@.a)
go
DENY ALL ON tab TO test
DENY UPDATE, INSERT, DELETE ON vw TO test
GRANT SELECT ON vw TO test
GRANT EXECUTE ON ins_tab TO test
go
Then, open a new connection, log in with userid 'test' (use SQL Server
authentication, not integrated authentication) and execute the code
below:
-- Login as user 'test'
-- Fails - no permissions on tab
SELECT *
FROM tab
-- Succeeds - SELECT permission on vw granted
SELECT *
FROM vw
-- Fails - no INSERT permission on vw
INSERT INTO vw (a)
VALUES (3)
-- Succees - EXECUTE permission on ins_tab granted
EXECUTE ins_tab @.a = 3
-- Check that trigger has executed as well
SELECT *
FROM vw
go
Here's the output:
Msg 229, Level 14, State 5, Line 1
SELECT permission denied on object 'tab', database 'test', owner 'dbo'.
a b
-- --
Msg 229, Level 14, State 5, Line 1
INSERT permission denied on object 'vw', database 'test', owner 'dbo'.
a b
-- --
3 18
Hugo Kornelis, SQL Server MVP|||I see it now. Thanks for such a detailed explanation Hugo.
BTW, we did fire the contractors.:)
--
Dan D.
"Hugo Kornelis" wrote:
> On Mon, 22 May 2006 06:52:02 -0700, Dan D. wrote:
> >Using SS2000 SP4. We have a .NET app that accesses the sql server database.
> >I've tried to keep all access via stored procedures and views and the user
> >only has execute permissions on the sp and select on the views.
> >
> >The contractors have put a trigger on one of the tables.
> >
> >CREATE TRIGGER [SetPrimaryContact_Company] ON [dbo].[tblCompanyContacts]
> >FOR INSERT, UPDATE
> >AS
> >IF (Select bitCompPrimaryContact From Inserted) = 1
> >Begin
> >Update tblCompanyContacts Set bitCompPrimaryContact = 0
> >Where numCompanyId = (Select numCompanyId From Inserted)
> >And numContactID <> (Select numContactId From Inserted)
> >End
> Hi Dan,
> First, fire the contractors. This trigger will fail on any multi-row
> insert or update. No-one should write triggers like that and be paid for
> it!
> >So, the user needs select and update permissions on the table.
> Says who? As long as the table is owned by the same user that owns the
> trigger, the user doesn't need any permissions on the table - see the
> repro script below.
> > Since I
> >didn't want to give insert, update or delete permissions on any tables to the
> >user is there a way around it? What is a best practice approach. Do I just
> >not use update triggers?
> In most cases, I would recommend having all objects (tables, views,
> stored procedures, triggers, functions, ...) owned by dbo. Then, you can
> set permissions exactly as required. Due to all objects having the same
> owner, all statements from within stored procedures, triggers, and
> functions will work. (Unless you're using dynamic SQL).
> Here's a repro to show that a user doesn't need permission on the table
> if a trigger changes data in the table.
> First, run the script below while connected with an accoount with
> sufficient rights to add databases and logins:
> USE master
> go
> DROP DATABASE test
> go
> sp_droplogin @.loginame = 'test'
> go
> CREATE DATABASE test
> go
> sp_addlogin @.loginame = 'test',
> @.passwd = 'test',
> @.defdb = 'test'
> go
> USE test
> go
> sp_grantdbaccess @.loginame = 'test'
> go
> CREATE TABLE tab (a int NOT NULL,
> b int DEFAULT NULL)
> go
> CREATE VIEW vw
> AS
> SELECT a, b
> FROM tab
> go
> CREATE TRIGGER trig
> ON tab
> FOR INSERT, UPDATE
> AS
> IF @.@.ROWCOUNT > 0
> BEGIN
> UPDATE tab
> SET b = a + 15
> WHERE b IS NULL
> AND EXISTS
> (SELECT *
> FROM inserted
> WHERE inserted.a = tab.a)
> END
> go
> CREATE PROC ins_tab (@.a int)
> AS
> INSERT INTO tab (a)
> VALUES (@.a)
> go
> DENY ALL ON tab TO test
> DENY UPDATE, INSERT, DELETE ON vw TO test
> GRANT SELECT ON vw TO test
> GRANT EXECUTE ON ins_tab TO test
> go
> Then, open a new connection, log in with userid 'test' (use SQL Server
> authentication, not integrated authentication) and execute the code
> below:
> -- Login as user 'test'
> -- Fails - no permissions on tab
> SELECT *
> FROM tab
> -- Succeeds - SELECT permission on vw granted
> SELECT *
> FROM vw
> -- Fails - no INSERT permission on vw
> INSERT INTO vw (a)
> VALUES (3)
> -- Succees - EXECUTE permission on ins_tab granted
> EXECUTE ins_tab @.a = 3
> -- Check that trigger has executed as well
> SELECT *
> FROM vw
> go
> Here's the output:
> Msg 229, Level 14, State 5, Line 1
> SELECT permission denied on object 'tab', database 'test', owner 'dbo'.
> a b
> -- --
> Msg 229, Level 14, State 5, Line 1
> INSERT permission denied on object 'vw', database 'test', owner 'dbo'.
> a b
> -- --
> 3 18
>
> --
> Hugo Kornelis, SQL Server MVP
>|||The trigger they wrote also contained a rookie mistake. It was designed for
a single row at a time. When you are writing a trigger you always need to
plan on having multi row events occur at some point.
Snip--
Update tblCompanyContacts Set bitCompPrimaryContact = 0
Where numCompanyId = (Select numCompanyId From Inserted)
And numContactID <> (Select numContactId From Inserted)
End snip--
See how the subqueries assume that only one item will be returned. This code
will raise an error if there is more than one row in the Inserted table (in
other words, if a multi row update or insert occurs) and that would cause
the original transaction to be rolled back.
Far better would be the following:
---
Update tblCompanyContacts Set bitCompPrimaryContact = 0
FROM tblCompanyContacts C
JOIN Inserted I
ON I.numCompanyId = C.numCompanyId
And I.numContactID <> C.numContactId
---
This way no matter how many rows have been affected by the update or insert
statement the trigger will now handle them.
All of the advice you received on moving the logic to the stored procedure
has merit. You should understand the following. The trigger will fire for
any update or insert on your table. Some folks like to use triggers to
handle some of their data integrity enforcement and not worry about in their
procs. If you do put it in your procs, you will need to put that logic into
every proc that will update or insert into that table. If this is a large #
of procs then you are setting up a shotgun surgery approach to maintenance.
The other thing to be aware of is that your trigger fires as part of your
original transaction, so an error in it will cause your original action to
be rolled back, as well as the locks from your original action are still
maintained during the trigger. In this case that is not a bad thing, since
you want this to happen as part of the update, and if it fails, you want the
update to fail as well.
--
David Lundell
Principal Consultant and Trainer
www.MutuallyBeneficial.com
David@.MutuallyBeneficial.com|||So, is it probably safer to stay away from subqueries in triggers?
One of the things we have to be careful about with triggers is that we do a
lot of data manipulation on the backend. So we have to be sure that if we
insert 5M rows on the backend and not through the app, a trigger doesn't fire
and do something it shouldn't.
Thanks,
--
Dan D.
"David Lundell" wrote:
> The trigger they wrote also contained a rookie mistake. It was designed for
> a single row at a time. When you are writing a trigger you always need to
> plan on having multi row events occur at some point.
> Snip--
> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> Where numCompanyId = (Select numCompanyId From Inserted)
> And numContactID <> (Select numContactId From Inserted)
> End snip--
> See how the subqueries assume that only one item will be returned. This code
> will raise an error if there is more than one row in the Inserted table (in
> other words, if a multi row update or insert occurs) and that would cause
> the original transaction to be rolled back.
> Far better would be the following:
> ---
> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> FROM tblCompanyContacts C
> JOIN Inserted I
> ON I.numCompanyId = C.numCompanyId
> And I.numContactID <> C.numContactId
> ---
> This way no matter how many rows have been affected by the update or insert
> statement the trigger will now handle them.
> All of the advice you received on moving the logic to the stored procedure
> has merit. You should understand the following. The trigger will fire for
> any update or insert on your table. Some folks like to use triggers to
> handle some of their data integrity enforcement and not worry about in their
> procs. If you do put it in your procs, you will need to put that logic into
> every proc that will update or insert into that table. If this is a large #
> of procs then you are setting up a shotgun surgery approach to maintenance.
> The other thing to be aware of is that your trigger fires as part of your
> original transaction, so an error in it will cause your original action to
> be rolled back, as well as the locks from your original action are still
> maintained during the trigger. In this case that is not a bad thing, since
> you want this to happen as part of the update, and if it fails, you want the
> update to fail as well.
> --
> David Lundell
> Principal Consultant and Trainer
> www.MutuallyBeneficial.com
> David@.MutuallyBeneficial.com
>
>|||Subqueries are not the issue. It is the way the subqueries were written
Update tblCompanyContacts Set bitCompPrimaryContact = 0
Where numCompanyId IN(Select numCompanyId From Inserted)
And numContactID NOT IN (Select numContactId From Inserted)
This would also work, although it might not be as efficient as my earlier
example.
As for backend loading. I usually disable my triggers for the load and
reenable them afterwards. But only if I make certain to perform what my
trigger would perform. In this case setting the bitCompPrimaryContact = 0.
ALTER TABLE dbo.Widget DISABLE TRIGGER ALL
Do the load
ALTER TABLE dbo.Widget ENABLE TRIGGER ALL
David Lundell
Principal Consultant and Trainer
www.MutuallyBeneficial.com
David@.MutuallyBeneficial.com
"Dan D." <DanD@.discussions.microsoft.com> wrote in message
news:21CA8851-BF8B-4DAE-92AB-BBC85A239E11@.microsoft.com...
> So, is it probably safer to stay away from subqueries in triggers?
> One of the things we have to be careful about with triggers is that we do
> a
> lot of data manipulation on the backend. So we have to be sure that if we
> insert 5M rows on the backend and not through the app, a trigger doesn't
> fire
> and do something it shouldn't.
> Thanks,
> --
> Dan D.
>
> "David Lundell" wrote:
>> The trigger they wrote also contained a rookie mistake. It was designed
>> for
>> a single row at a time. When you are writing a trigger you always need to
>> plan on having multi row events occur at some point.
>> Snip--
>> Update tblCompanyContacts Set bitCompPrimaryContact = 0
>> Where numCompanyId = (Select numCompanyId From Inserted)
>> And numContactID <> (Select numContactId From Inserted)
>> End snip--
>> See how the subqueries assume that only one item will be returned. This
>> code
>> will raise an error if there is more than one row in the Inserted table
>> (in
>> other words, if a multi row update or insert occurs) and that would cause
>> the original transaction to be rolled back.
>> Far better would be the following:
>> ---
>> Update tblCompanyContacts Set bitCompPrimaryContact = 0
>> FROM tblCompanyContacts C
>> JOIN Inserted I
>> ON I.numCompanyId = C.numCompanyId
>> And I.numContactID <> C.numContactId
>> ---
>> This way no matter how many rows have been affected by the update or
>> insert
>> statement the trigger will now handle them.
>> All of the advice you received on moving the logic to the stored
>> procedure
>> has merit. You should understand the following. The trigger will fire for
>> any update or insert on your table. Some folks like to use triggers to
>> handle some of their data integrity enforcement and not worry about in
>> their
>> procs. If you do put it in your procs, you will need to put that logic
>> into
>> every proc that will update or insert into that table. If this is a large
>> #
>> of procs then you are setting up a shotgun surgery approach to
>> maintenance.
>> The other thing to be aware of is that your trigger fires as part of your
>> original transaction, so an error in it will cause your original action
>> to
>> be rolled back, as well as the locks from your original action are still
>> maintained during the trigger. In this case that is not a bad thing,
>> since
>> you want this to happen as part of the update, and if it fails, you want
>> the
>> update to fail as well.
>> --
>> David Lundell
>> Principal Consultant and Trainer
>> www.MutuallyBeneficial.com
>> David@.MutuallyBeneficial.com
>>|||We're still working on a solution. We have many clients that we do mailings
for and we have hundreds of databases. I'm not sure it's the best way to try
and keep track of every trigger on every table.
We're an example of a small company that grew very fast without any
standards or best practices. Now we're trying to get everything under control.
We recently had a case where a developer put a trigger on a table for an
application and didn't tell anyone about it. Adn to compound the problem, the
trigger didn't work like he thought it would. I guess he didn't test it
thoroughtly. When someone loaded some new data on the backend, the trigger
fired and updated 150,000+ rows that it shouldn't have. We didn't discovered
it until months later and we're trying to figure out how to roll it back.
Thanks for all of your help,
--
Dan D.
"David Lundell" wrote:
> Subqueries are not the issue. It is the way the subqueries were written
> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> Where numCompanyId IN(Select numCompanyId From Inserted)
> And numContactID NOT IN (Select numContactId From Inserted)
> This would also work, although it might not be as efficient as my earlier
> example.
> As for backend loading. I usually disable my triggers for the load and
> reenable them afterwards. But only if I make certain to perform what my
> trigger would perform. In this case setting the bitCompPrimaryContact = 0.
> ALTER TABLE dbo.Widget DISABLE TRIGGER ALL
> Do the load
> ALTER TABLE dbo.Widget ENABLE TRIGGER ALL
>
> --
> David Lundell
> Principal Consultant and Trainer
> www.MutuallyBeneficial.com
> David@.MutuallyBeneficial.com
> "Dan D." <DanD@.discussions.microsoft.com> wrote in message
> news:21CA8851-BF8B-4DAE-92AB-BBC85A239E11@.microsoft.com...
> > So, is it probably safer to stay away from subqueries in triggers?
> >
> > One of the things we have to be careful about with triggers is that we do
> > a
> > lot of data manipulation on the backend. So we have to be sure that if we
> > insert 5M rows on the backend and not through the app, a trigger doesn't
> > fire
> > and do something it shouldn't.
> >
> > Thanks,
> > --
> > Dan D.
> >
> >
> > "David Lundell" wrote:
> >
> >> The trigger they wrote also contained a rookie mistake. It was designed
> >> for
> >> a single row at a time. When you are writing a trigger you always need to
> >> plan on having multi row events occur at some point.
> >>
> >> Snip--
> >> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> >> Where numCompanyId = (Select numCompanyId From Inserted)
> >> And numContactID <> (Select numContactId From Inserted)
> >> End snip--
> >>
> >> See how the subqueries assume that only one item will be returned. This
> >> code
> >> will raise an error if there is more than one row in the Inserted table
> >> (in
> >> other words, if a multi row update or insert occurs) and that would cause
> >> the original transaction to be rolled back.
> >>
> >> Far better would be the following:
> >> ---
> >> Update tblCompanyContacts Set bitCompPrimaryContact = 0
> >> FROM tblCompanyContacts C
> >> JOIN Inserted I
> >> ON I.numCompanyId = C.numCompanyId
> >> And I.numContactID <> C.numContactId
> >> ---
> >> This way no matter how many rows have been affected by the update or
> >> insert
> >> statement the trigger will now handle them.
> >>
> >> All of the advice you received on moving the logic to the stored
> >> procedure
> >> has merit. You should understand the following. The trigger will fire for
> >> any update or insert on your table. Some folks like to use triggers to
> >> handle some of their data integrity enforcement and not worry about in
> >> their
> >> procs. If you do put it in your procs, you will need to put that logic
> >> into
> >> every proc that will update or insert into that table. If this is a large
> >> #
> >> of procs then you are setting up a shotgun surgery approach to
> >> maintenance.
> >> The other thing to be aware of is that your trigger fires as part of your
> >> original transaction, so an error in it will cause your original action
> >> to
> >> be rolled back, as well as the locks from your original action are still
> >> maintained during the trigger. In this case that is not a bad thing,
> >> since
> >> you want this to happen as part of the update, and if it fails, you want
> >> the
> >> update to fail as well.
> >> --
> >> David Lundell
> >> Principal Consultant and Trainer
> >> www.MutuallyBeneficial.com
> >> David@.MutuallyBeneficial.com
> >>
> >>
> >>
>
>|||On Tue, 23 May 2006 06:07:02 -0700, Dan D. wrote:
>So, is it probably safer to stay away from subqueries in triggers?
Hi Dan,
Only in the same way that not using a database at all prevents errors :)
Use subqueries if you need them - just be aware that a trigger will
always fires exactly once when a corresponding INSERT, UPDATE or DELETE
statement is executed. Design your triggers to work regardless of number
of rows affected. And test them with single-row, multi-row and zero-row
operations.
>One of the things we have to be careful about with triggers is that we do a
>lot of data manipulation on the backend. So we have to be sure that if we
>insert 5M rows on the backend and not through the app, a trigger doesn't fire
>and do something it shouldn't.
Bottom line when deciding if something belongs in a stored proc or in a
trigger is: is in integrity related?
Triggers should be used to enforce the integrity rules that can't be
enforced with standard SQL. For instance complicated constraints that
can't be exppressed in SQL constraints. Or computations that (for
whatever reason) can not be calculated in a view. The execution of a
trigger is about as hard to bypass as the checking of a constraint or
the "updating" of a calculated value in a view. That's because it should
be used for code that really should never be bypassed.
Bypassing the execution of stored procedures is much easier. Exactly how
easy depends on yoour security model as well - but there are always SOME
users able to update a table directly without executing the stored proc
that should be used to update the table.
--
Hugo Kornelis, SQL Server MVP