Sergey,

I wondered what Craig might be getting at with "...set up properly...". The only thing I could think of was actually having a table field with the same name in the active program.
So I did a test like yours, but opened a table first and used a memvar name identical to a field name.

The results I got for 1,000,000 executions of each were:

1) With m. on both sides = 1.158 seconds
2) With m. on right only = 1.018 seconds
    - Difference is .140 seconds per 1,000,000

For comparison, the same test WITHOUT a table opened yielded:

1) With m. on both sides = 0.920 seconds
2) With m. on right only = 0.777 seconds
    - Difference is .143 seconds per 1,000,000

This is a small enough difference that I will continue to use m. on both sides.

JimN

SNIP
>>How did you conduct your tests? If not set up properly, you will get your results.
>
>I ran two simple FOR loops with 1,000,000 itterations incrementing variable lnCounter inside. One with m.lnCounter on the left side of assugment, another with lnCounter

>m.lnCounter = 0
>m.Start = SECONDS()
>FOR i=1 TO 1000000
>    m.lnCounter = m.lnCounter + 1
>ENDFOR
>? SECONDS() - m.Start
>
>m.lnCounter = 0
>m.Start = SECONDS()
>FOR i=1 TO 1000000
>    lnCounter = m.lnCounter + 1
>ENDFOR
>? SECONDS() - m.Start

>>
>>>
>>>IMO it's all a metter of personal preferences.
>>
>>I believe I said that.
>
>At least we can agree on that :)