Fabio,

I think my problem does not have solution. BINARY INDEXES just work fine if you have all indexes in MACHINE collate.

My problem is... The user wants to find someone that has the first name like "Jo" (it means Jose, José, João, João, Joana, etc)... So, I have to list it for him in the right order (GENERAL collate)

If I add this new index, and use it in a network, the query runs too slow.
If I try to change the binary index from machine to general, I get an error message.
If I change my indexes to machine, my results is a little bit confused.

So, I think I will not use DELETED() index. I want to improve a little bit, but I think it is not possible.

By the way: When I have to list all names that start with "Jo", I use it:

Select * From MyTable Where Name Likes "Jo%" Into Cursor Resultado
<\pre>
and I think it works fine, I mean, rushmore!

Thanks!

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Rodolfo Duarte