>>>>Hi Naomi,
>>>>
>>>>>The form in question has all complex logic of setting filter, etc. in its Activate method. It makes the debugging of this form much harder.
>>>>
>>>>If every form changes the data session to the same one, then every one operates on the same work areas. Hence, every form would need to restore its environment every time the form is activated.
>>>>
>>>>I wouldn't move it into the Init event. It's better to find a work around for your debugging problem. What exactly is making debugging harder?
>>>
>>>Very interesting thing I found so far. The form uses a ConfigSearch table and in the Load event does some manipulations with this table. I believe it does tableupdate after these manipulations.
>>>
>>>After I moved the code to the Init and switched to command window, I see, that after set filter command there are two same records.
>>>
>>>I opened this table in another VFP session and there is only 1 record there.
>>>
>>>I'll try to determine, may be this extra record is not saved yet - buffered?
>>
>>I'm sure you know that recno() will return a negative number if it's buffered. Is there a relationship set between that table and another?
>
>Attached please find a screen shot of the table.
>
>As you can see, this problem has a long history.
>
>I somehow duplicated it 2 days ago. And in my case with even worse result - same ID for two different customers / sales.

Are you absolutely sure that they are accidental duplicates or fantom records? I ask because sometimes in quoting systems a record is kept to show each quote that was presented to a prospective customer and those records are kept for historical purposes. The changedate makes me wonder...

There are some times when non-normalized data is required.