>
> Also, add a check to the trigger so it only performs the copy under
> "normal" conditions. For example, you might perform all your importing
> using a login that employs the IMPORTER role. Then the trigger can check
> whether the current connection is using the IMPORTER role and not
> perform the copying if it is.
>

This is what I am currently doing except that copy is made to the same table under normal conditions. But as I've said earlier - does introducing the IMPORTER role(assuming that with this role records are not copied on update) not kill the original purpouse of data-safety? It is hard to see for me the difference between updating a record by user mistake in application and updating the record by user mistake that he made picking up the wrong(old data) import file.

In first case I have the copy of a record, in a second case I do not have. That is inconsistent.