RE: TEST Date in free form

On Wed, 2005-06-15 at 19:44, Christen, Duane J. wrote:
> Joel;
> 
> On the surface your comparison ratios are correct. Under the covers though,
> when the assignment opperation fails it throws an exception and the Monitor
> catches the exception and performs the corrisponding On-Error code.
> Exception processing is expensive, thus my rule of thumb is that if < 30% -
> 40% of the "checked" values are in error use Monitor, otherwise use a TEST
> opcode. 
> I haven't formally tested my percentages, but they seem reasonable. 
> 
> Duane Christen 

I'll agree with both you and Charles that there are circumstances where
the performance may be hindered and that my comparisons were hardly
scientific.  I was trying to point out more that the benefit of using
Monitor is that there far less work to do when no error occurs, and that
given it's benefits I would use it even when the error rate is high.  

And I'm sure you are right that there is probably a "break-even"
percentage somewhere, but are you really going to know what that is at
design time? If you *really* expect such a high rate of errors, then I
suggest there is something more toublesome going on than the performance
of monitor...

Joel
http://www.rpgnext.com