> With all do respect, i don't think you understand the working of rushmore.
> If only 1 part of your expression is optimizable (combined with AND) rushmore
> will optimize this. I strongly suggest reading 'Optimizing Rushmore
> Expressions' chapter 15 (in the developersguide of VFP 5.0)

Well, believe me. I've read this chapter several times. ;) My example I sent to Jim should give you a good overview over the facts.

I'm sorry for replying to your message. Please excuse my interruption, but my understanding was that if somebody sends a message on the UT, everybody may feel free to answer. Especially if it's such a controversial message as the one you sent. Apparently I was wrong. I promise not to do it again... <s>

> Your misunderstanding of rushmore is exactly the reason i wrote this message.
> An expression does not have to fully optimizable which don't have to mean
> that it would be slower than a fully optimizable expression

Well, this is a rather interesting statement. :) Let's see: 'Partially optimized queries are equally fast as fully optimized ones'. This sounds like a paradoxon to me. :)

The fact of the matter is that even very small and simple queries run essentially faster when they are fully optimized then when they are partially optimized. The simple example I sent to Jim runs twice as fast fully optimized then when it's partially optimized (I just tried it). Even though there are only 92 records and none of them are deleted. The interesting part is that when you remove all indexes and run the query again (which is now non-optimized), you can hardly see a difference to the partially optimized version.

Now you can say that this is only because it's a small table, and you have a point there. However, it already shows that the difference between fully optimized and partially optimized is at least as big as the difference between partially optimized and non-optimized. At least in many cases. And that's enough reason for me to add the Deleted() index.

> OIW the deleted tag is NOT required to take advantage of Rushmore. In
> fact the deleted tag may be much of a burden in a lot of cases.

Quite the contrary! The index on Deleted() is relatively cheap (sorry, I don't have any numbers to back this up...) while the difference between partially optimized and fully optimized often is substantial.

Regards,

Markus