Mark,
BINTOC() can only potentially reduce the size of an index on fields of type N. If you have an N(5) field where values don't exceed 32767 you can use a 2 byte bintoc result. Since N are really stored as characters with leading spaces in the dbf file, I'd be a little suprised that the Run Length Encoding compression technique used in the CDX didn't result in smaller CDX file sizes without the BINTOC(). You also have to weigh the time that function takes to execute while building indexes to see if it's worth it.
>I am experimenting with using BINTOC() on some integer indexes. I have a table I expect to be very large and would like to make the indexes as compact as possible.
>
>I added 15,000 dummy records to my test table and checked the CDX file. It was 279,552 bytes for just integer keys.
>
>I changed all four index expressions to BINTOC( ... ), reindexed, refreshed the view in file explorer and expected the CDX file to be smaller, but it wasn't--exactly the same # of bytes.
>
>I am missing something? Thanks.