>First off: I believe if you set ansi off, the results should be different, so any comparison might be unfair just on the number of returned values for index expressions and the resulting data set - but I am a bit blurry at the moment from not enough sleep<g>.
>
>

>select ... from ... where cl_name =="GREE" and cf_name == "MAR" and cM_Initial == ""

>should be equal to

lcName = padr("GREE",25,"_") + padr("MAR",20,"_")+padr("",1,"_")
>select ... from ... where cl_name + cf_name + cm_Initial == m.lcName
>

Hi Tomas,

First of all, you're a little bit mistaken. If I use underscore "_", the two expressions are different. They would we the same, if I use space.

Secondly, I wanted to search by part of the name, that's why I used SET ANSI OFF. But since LIKE clause is fully optimized in VFP9 with % I may want to use it and this way my view would be SQL Server compatible...

We found few bugs in VFP9, but we hadn't chance to test, if they are VFP9 bugs or Mere Mortals related. So I am not sure when we're going to switch to 9 but hopefully soon.

>If your table is non-trivial, always include the MDot: it gets searched for every record.

I personally always use mdot out of the habit I got from my prev. work by adapting Mike's Asherman coding style, which I liked very much. I probably didn't put it here because it was just an example.

>For measuring "targeted" speed differences I'ld recommend putting the table and index onto another computer with a bottleneck connection: either use an old 10 MBit hub or serial connection to simulate a busy LAN - Probably then the second version will be faster if the table has many records. If speed is of utmost importance for this, also try out seek / scan while.
>
>my 0.02 EUR
>
Thanks again.