Re: Trigger question

James Lampert wrote:

Seriously, I've also been sorely tempted to put surrogates in on not only my workflow trigger but "inbound interface" triggers, but worry about the overhead of an additional call, particularly when the trigger program itself is sufficiently big and complex to affect performance.

I have not had complaints about poor performance. My guess is that in the grand scheme of things, a few ms 'overhead' for a program call are dwarfed by the cost of actual I/O. It's a trade-off between a tiny but constant cost vs the rare but large cost of turning the company off while re-compiling a trigger program. Of course the only way to know for sure is to benchmark it on your system with your dataset.

Incidentally, with the surrogate approach, does setting the activation group to *CALLER (usually for performance, but sometimes for the sake of persistent statics) by any chance defeat the whole exercise of using a surrogate?

I may be misunderstanding you here. The activation group of the surrogate won't affect the ability to re-compile the actual trigger.