Well, for raw speed the kludgy i-type index method is faster, however, for
reliability and the ability to know when the trigger was fired (before/after,
update/insert/delete) you will probably need to use the file trigger mechanism.
As someone raised in an issue recently, triggers run in a transaction (which
of course, doesn't allow you to run the debugger through it). However, it
should be simple enough to create a test interface for your trigger program
before you apply it to the file.
Personally, I think there are definite advantages to running triggers within a
transaction (opposed to i-type updates that IIRC won't allow it). One being
that any associated updates/deletes are managed as a single update, so all
updates are synchronised and/or if your trigger decides that the update can't
or shouldn't be happening, then it can roll back all changes automatically via
the transaction mechanism.
Stuart.


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of "Robin Stanley"
<[EMAIL PROTECTED]>
Sent: Tuesday, 2 August 2005 10:02
To: <[email protected]>
Subject: [U2] Real Trigger vs. Kludgy Trigger via an Index


Before UV supported real file triggers, we used a kludgy way of doing it by
creating an index on an i-descriptor dictionary, which then called a BASIC
program to do our 'trigger stuff'. 

Does anyone have any input on what would now be faster? The real trigger or
the 'index' trigger? I'm probably being lazy, because I know how to setup the
'index' trigger, but haven't yet ever tried the 'real' trigger. It would be
worth it if I thought performance would be significantly better. 

Thanks. 

Robin Stanley 
Paciolan, Inc 
------- 
u2-users mailing list 
[email protected] 
To unsubscribe please visit http://listserver.u2ug.org/ 
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to