That is kind of what I was suspecting. I have some files with 12 million and 14 million records and the temp files seem to get cleared between each set of generating/indexing cycles. It has always been too fast to read the incrementing count on the Indexing, even with FPW 2.6.

Maybe on smaller files the indexing engine is able to make do with one pass...which might make an improvement on how the cdx file is actually stored on the disk.

>
>On the test I did, I used a table with nearly 2 million records, and 7 index tags. The table is 1.5 GIGs in size. The reindexing operation took nearly
>1.5 hours, and it did make multiple passes through the table creating the keys. the indexing then was less than 2 seconds. As each succeeding key was added, the operation took measurably longer (a few seconds) to complete. The temp files grew quite large during the key creation part of the operation, were emptied, then grew again on each pass.
>
>As a result, I am as lost as ever before <eg>