>Hi Bela,
>
>>Can Rushmore use indexes made under different collate sequence than the current one?
>
>No. Rushmore can only use an index that has the same collate sequence as currently is active, because the collate sequence determines how strings are compared. And any other collate sequence beside MACHINE is usually slower, up to 80%. Beside that, the GENERAL collate sequence has a lot of bugs, for example it doesn't work reliable at all with numeric, especially with integer expressions.
>
>Christof
The non-machine sequences use double-byte internally, and many of these double-byte values are the same for different chars - for one, these sequences are case-blind, and their purpose is to rearrange the characters in a specific non-ascii order. Using them for anything but string data really means calling for disaster, and the results are, as one of my colleagues here said, "technicolor sh*".