That's not how Replicate works.
When an update on the source source row gets committed, an update entry will be in the transaction log, and Replicate will process it unless the row is filtered out. It doesn't matter whether any column was actually changes or not and it certainly doesn't matter that the particular column which had an update was dropped in the table transformation and not forwarded to the target. The update will be processed for all columns which are forwarded whether they changed or not.
It's not immediately clear to me how that would cause duplicate errors unless perhaps the forwarded columns where part of a key? Or maybe there was a foreign key. It would help if you try to explain the duplicate error in further details.
You may be able to define a filter on this, perhaps using column before image values ( use $BI__<columnname> variables) but that feels awfully tricky.
It may be better to fully understand the duplicate error and deal with that.
Hein.
