On Fri, 25 Nov 2011, David Johnston wrote:
Simplistically you load all the data into a staging table that has no natural primary key and then write a query that will result in only a single record for whatever you define as a primary key. Insert the results of that query into the final table.
David, I was leaning toward doing this.
If you only use a subset of columns to uniquely determine the PK than you need to decide how to resolve situations where the PK fields are duplicated but the extra fields are unique. That step is why the first description is simplistic.
Well, that's point of a PK, isn't it? There should be only one row with that specific set of column values. The additional columns are unique to that set of PK values. Thanks, Rich -- Sent via pgsql-general mailing list (pgsql-general@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general