Thanks for your info about the web-page about Primary keys. But generating the keys seems to work fine the way I do it. I changed from having a value in a field in another table after the first incident. It did work fine too, but I thought it was somehow safer to pick the last value from the table itself.

The reindexing might be a problem. I run the following on programstart. If it can be opened exclusive, it does reindexing:
Function reindexer()
On error RETURN
Select 0
Use makorn exclusive
Reindex
Use
Use ma50 exclusive
Reindex
Use
On error
Endfunc

I have tried to switch off my computer having the DBF's on another computer and have not lost any data. There has been no GPF's or other serious errors at the lab either as far as I know.

But I will change the program to delete the indexes and recreate them on program start.




>Hi Einar,
>
>The problem you're experiencing maybe a result of corrupted index. Don't use REINDEX, deltete all indexes on your table and recreate them from scratch.
>Also IMO, this way of creating Primary keys is unrelaible and required file locking not record loging. I would recomend that you use technique that's presented in article 'Primary Keys' at http://www.craigberntson.com/Articles/kb006.htm.