Hi dorin,

Of course a very exceptional problem. The bug occurs when there already is a problem in the table which should not have occured in the first place.

I stay with my opinion that with using row and table buffering and transactions you should be able to minimize database corruptions.

Walter,


>It was row buffering. But I don't think had any influence. The bug is documented here:
>http://support.microsoft.com/?kbid=293638

>Dorin.
>
>
>>Do you use buffering within transactions ?
>>
>>Walter,
>>
>>
>>>I use transactions. If you are lucky enough and tables are not corrupted, is OK. Otherwise, you have to handle the BEGIN/END TRANSACTION bug. All inserted records dissapear (I've seen that was fixed in VFP8 though).
>>>If you never had index corruption then you are very lucky.
>>>We even had a "delayed write" error that had as result a 50 MB table filled with chr(0). Because some bug in windows network client.
>>>
>>>>Depends on how you write your programs. Use buffering and transactions as much as possible to minimize the problem. In my ten years of developing and maintaining about two dozen VFP apps I never had a clear index corruption while several hundreds of user are using those apps each day. I remember the index corruption problems mostly from the FPW 2.x days when doing things over an unreliable WAN.
>>>
>>>I would vote for VFP + Firebird, but I'm subjective here :-)
>>>
>>>>
>>>>Other difficulties that come with MSDE.
>>>>
>>>>- More difficult to install
>>>>- More difficult to maintain.
>>>>- More suspectible to mallfunction or virusses.
>>>>- MSDE: a Data limit or 2.1? GB
>>>>
>>>>All with all a good reason to evaluate what you need before you decide whether to use a pure VFP solution or a combination of VFP and MSDE.
>>>>
>>>>Walter,