Thanks, Aleksey.
I'll give that a shot. Hopefully it will not be a big hit on performance.

I must admit some surprise that it is deemed safe to do this before a BEGIN TRANSACTION. I guess I'd better make sure that it is done only on the first if using nested transactions?

Thanks


>Hi Jim,
>
>This sounds like an issue we already know about. If it is then executing SYS(1104) before BEGIN TRANSACTION should eliminate the error.
>
>Thanks,
>Aleksey.
>
>
>>I've seen 3-4 2065 errors testing an application using VFP9.
>>They have been happening on .TMP files, apparently related to TRANSACTION buffering.
>>
>>On the latest one I was able to keep the 'file' that the error was reported against. To do so I set the file's attributes to be Read-Only before clicking the "OK" button on the error dialogue.
>>
>>When I then tried USEing the file I got the error 2091, "the table... has been corrupted...".
>>Then, after setting TABLEVALIDATE to 0 and trying USE once again I got an error message saying: "Cannot open persistent table buffer directly.".
>>
>>These are happening far too frequently in VFP9. I am hoping that the captured file with the error can be of significant use in finally isolating the problem. I'll be happy to send it somewhere to be looked at.
>>
>>In case it offers any clues the file's name is: "00004U1L00KA.TMP" and was deep in Documents and Settings, in a "Temp" sub-directory.
>>
>>JimN