>>>My job as the hired gun is to solve this problem, not tell them that a decade or more
>>>of development is based on a framework that some don't like.
>>
>>
>>FWIW, in code that works correctly otherwise, the suggestion about a resource timing out and losing its connection or handle after 45 to 60 minutes seems to me the most likely cause.
>
>Yeah, but I wouldn't think it would fail on the same record each time, and it is.
>
>Tamar

Could it be a bad index? Does the client workstation have very little ram? The bitmaps of the different tables can be quite large...