Hi Thomas! Thanks.

Involved tables are not buffered. I tried different ways - also record pointer moves - without good results. I did not try locking. If a short lock would work maybe it would be the solution of my problem.


>Hi Josef,
>
>>The behaviour I can see also on the same computer when the application has
>>two instances running simultanously.
>>The sql select command is placed in init method of report dataenvironment
>
>That sounds more like a vfp-buffering problem to me.
>check the buffering status of the table where the "document" (record ? set of records?) will be written first.
>if there is no buffering or the relevant tableupdates() have been issued,
>the buffer of the reading process isthe problem.
>
>If this process opens the table for the first time,
>and all records areas you expect(should be),
>you must work with locks and/or record pointer moves
>to refresh vfp's read-buffers.
>
>hth
>
>thomas