I would bet that the index size reduction is caused by UPPER() and has nothing to do with ALLTRIM().
IMHO, the leading spaces should be removed in the table not in the index.

>Ah, you are correct, if you are indexing to a single idx. If you are indexing using the tag keywords, the results will be smaller.
>I have a table with 1.2 million records.
>If I do the following:
>INDEX on ALLTRIM(UPPER(last)) TAG last
>INDEX on ALLTRIM(UPPER(first)) TAG first
> The resulting file size is 11,365 kb
>
>INDEX on last TAG last
>INDEX on first TAG first
>The cdx file is 22,731 kb
>
>This also strips out any leading blanks, and forces upper case.
>
>John
>
>>You're mistaken. VFP does not support variable length index keys. As result, an index key after ALLTRIM() will be padded back to the size of the field. Adding ALLTRIM() is not required for index to be it optimazable either because it's optimaziable w/o it already.
>>

>>upper(thiswhatever)) = upper(m.thiswhatever)
>>

>>
>>The bottom line, ALLTRIM() is useless in indexes at least and dangerous at worst.
>>