My first SWAG was that there was cursoradapter or view update logic somewhere (else?) in the code involved, which might include automatically set timestamps/lastuser info and if not recalled from server would create problems on next update. Did see nothing of that in the quick scanned later sources, but it is one possible answer ;-)


>Why would one do this?
>
>Scenario:
>
>Data is pulled from SQL server into cursor for an individual
>Cursor is then buffered
>Cursor is updated from input form (there is no update/commit on the cursor, just replaces)
>SQL server is updated from cursor.
>Data is repulled from SQL server into cursor for the same individual
>
>We have instances where (at least) one field is not making it 'back' from SQL server and is causing issues further downstream in the process and, of course, it's only happening on occasion.