Hi Calvin

Take a look at KB293638. It has just been upgraded to from PRB to BUG status, so hopefully it might be fixed by SP2.

I doubt that checking the header is likely to help much, unless the table(s) are not in use. My workaround is to append the blank record, insert the primary key (and any candidate key), tableupdate() it, and read it back before completing the update. It does flush to disk, and the read will fail if the table header is corrupt, and return an error each time until the header is repaired.

In reporting such a read failure, at least you get to know immediately if there is a problem. A user will try again, and it will continue to error out.

While I use SDT to repair files, the chances are in a multi user environment that it does not get to do its checking very often as part of the app. In the meantime, data may continue to 'get lost'.

HTH

Geoff



>I am having trouble with records disappearing and I need a quick way to verify that the reccount in the table header is accurate on networks. Is the best way to do this to get the physical size of the table and compare it to the computed size on the basis of reccount. I am using Stonefield and open most of the tables in the forms dataenvironment. Where would I put this code in the dataenviroment events? Any code snippets and suggestions greatly appreciated. Thanks.