Обсуждение: UPDATE keyword

Well, I am making progress in rewriting my MSSQL Server T-SQL triggers and stored procedures in Pl/Tcl.  However, I
haverun into a bit of an issue and I wonder if anyone knows the best way to address it... 

In T-SQL triggers, you can use the UPDATE keyword in conditional expressions as in

IF UPDATE(myfield)
BEGIN
    do something
END

It detects the update of the field.  I have been comparing $NEW(myfield) and $OLD(myfield) which works ok, except where
Iused the UPDATE() test as a crutch.  I sometimes do an update like 

UPDATE mytable SET myfield = myfield WHERE...

to fire an update trigger, and executing only the code in that trigger surrounded by UPDATE(myfield).

Is there an easy way to duplicate this, or should I work around it?

Ian A. Harding
Programmer/Analyst II
Tacoma-Pierce County Health Department
(253) 798-3549
mailto: ianh@tpchd.org

ianh@healthdept.co.pierce.wa.us ("Ian Harding") schreef:
>UPDATE mytable SET myfield = myfield WHERE...
>to fire an update trigger, and executing only the code in that
>trigger surrounded by UPDATE(myfield).

So you update a field by not changing its value, to execute a piece of
code that has nothing to do with the field and the value :-)

>Is there an easy way to duplicate this, or should I work around it?

Isn't it easier (and more poratble) to add a dummy field to the table?
Then you can do:

   UPDATE mytable
   SET dummy = 1 - dummy
   WHERE ...

to execute the piece of code that responds to an update of 'dummy'.

Regards,
René Pijlman

http://www.applinet.nl

True, true... but the next time I need the code to fire I would need ot remember that I had set it to 1 and set it to
somethingelse... 

Ian A. Harding
Programmer/Analyst II
Tacoma-Pierce County Health Department
(253) 798-3549
mailto: ianh@tpchd.org

>>> Rene Pijlman <rpijlman@spamcop.net> 05/26/01 01:56AM >>>
ianh@healthdept.co.pierce.wa.us ("Ian Harding") schreef:
>UPDATE mytable SET myfield = myfield WHERE...
>to fire an update trigger, and executing only the code in that
>trigger surrounded by UPDATE(myfield).

So you update a field by not changing its value, to execute a piece of
code that has nothing to do with the field and the value :-)

>Is there an easy way to duplicate this, or should I work around it?

Isn't it easier (and more poratble) to add a dummy field to the table?
Then you can do:

   UPDATE mytable
   SET dummy = 1 - dummy
   WHERE ...

to execute the piece of code that responds to an update of 'dummy'.

Regards,
René Pijlman

http://www.applinet.nl

---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)

On Fri, May 25, 2001 at 04:16:00PM -0700, Ian Harding wrote:
> Well, I am making progress in rewriting my MSSQL Server T-SQL triggers and stored procedures in Pl/Tcl.  However, I
haverun into a bit of an issue and I wonder if anyone knows the best way to address it... 
>
> In T-SQL triggers, you can use the UPDATE keyword in conditional expressions as in
>
> IF UPDATE(myfield)
> BEGIN
>     do something
> END
>
> It detects the update of the field.  I have been comparing
> $NEW(myfield) and $OLD(myfield) which works ok, except where I
> used the UPDATE() test as a crutch.  I sometimes do an update
> like
>
> UPDATE mytable SET myfield = myfield WHERE...
>
> to fire an update trigger, and executing only the code in that trigger surrounded by UPDATE(myfield).
>
> Is there an easy way to duplicate this, or should I work around it?

how about

    create view
        relationname
        as select * from _real_relation_name_;

    create rule
        on update to relationname
        do instead (
            update _real_relation_name_
                set fld=new.fld,
                col=new.col,
                modified=current_timestamp,
                differencefield=old.something-new.something,
                whatever=some_function()
            ;
            insert into _some_tracking_relation_
                ( field, list, here )
                values
                ( ... )
        );

check out the manuals for 'create rule'. very handy stuff.

--
#95: We are waking up and linking to each other. We are watching. But
we are not waiting.  -- www.cluetrain.com

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

ianh@co.pierce.wa.us ("Ian Harding") schreef:
>True, true... but the next time I need the code to fire I would
>need ot remember that I had set it to 1 and set it to something else...

False, false, I've thought of that :-)

The statement says:

    dummy = 1 - dummy

a) Initially, dummy == 0. Then dummy becomes 1 - 0 which equals 1.

b) So dummy == 1. Then dummy becomes 1 - 1 which equals 0.

c) Goto a.

Dummy changes value from 0->1 or 1->0 on every execution of the
statement.

If you don't believe it, you can also try dummy = -dummy, but then you
have to remember to initialize it to a non-zero value :-)

--
Vriendelijke groet,
René Pijlman <rpijlman@spamcop.net>

Wat wil jij leren?
http://www.leren.nl/