Ok, I found this quote:

"For any dataset, KeyLinks establishes the uniqueness of one row when
looking at the server. To get a valid keylink on a single-table query, you
need the table's primary key. If the dataset involves joins or a composite
primary key, KeyLinks will be a list of those column names from the SQL
that, together, guarantee uniqueness of every row."

Now I understand. What it didn't say was whether the format could be <table
name>.<column name> since I have two ID fields from different tables.

Malcolm Smith
MJ Freelancing
ABN: 30 671 763 146
http://www.mjfreelancing.com