I understood the message to mean that the CPCURRENT()/CPDBF() was intended but undocumented, but I may be wrong. I suggest you send a reply to the answer I got from Aleksey at Microsoft, and ask.

However, is it really necessary to have different codepage for your DBFs than CPCURRENT()? I created a small program to check all my DBFs, and converted all the DBFs with 'wrong' codepage to my CPCURRENT() value (1252), and everything was fine. It turned out that all the DBFs with 'wrong' codepage were old tables, mostly from DOS days, which never had been converted.

If it really is necessasy to have different codepage on some of your DBFs, you can easily SET ENGINEBEHAVIOR 80, run your SELECTs and then SET ENGINEBEHAVIOR 90, can't you? I don't remember, because my problem was solved for good by the codepage change.

>Thank you. Of course, I have read your message. It describes the part of my problem. CPCURRENT() don't match CPDBF() and VFP9 won't use indexes on character expressions for optimization. But in VFP8 it works normally as I expect.
>I would like to know, if it is a bug or the new intended feature.
>In the first case I will wait for service pack 1. In the second case I must rewrite my programs, because the select without optimization is very slow.