>>Hi Peter,
>>
>>I did some testing with your code and found that if the 'noupdate' clause of the 'use cdxbug' statement was removed, the cdx was updated as expected generating a primary key error on the second iteration. So I'm curious about the true nature of the bug. Why did you use the noupdate clause and then try an update? Do you think that the noupdate clause is the real bug?
>>
>>Just a curious fellow developer.<s>
>
>
>Hi, Kendall.
>
>Yes, that's right. If the tables are opened normally in data session 1 they can be updated correctly in data session 2.
>
>I had opened the tables in DS 1 as "noupdate" because I didn't intend to do any updates there, only in the forms which use a private DS. Specifying "noupdate" seemed like the diligent thing to do.
>
>As I understand the nature of DS, each one should be independent of any other. What you do with a table in one DS should have no effect on what happens in another. (Other than locking records or saving changes in common tables, of course.) In my mind it should be like two different workstations opening the tables.
>
>I'm pretty curious about this myself. And thanks for taking the time to replicate this on another machine.
>
>Peter

Alex Bulgakov had a similar problem last year, and now tried it with 6.0 - there was a message from him recently about this. The layout seems to be similar, and I've been able to reproduce the bug in 5.0 last time. He tried to draw attention to this.