This method does not appear to work in our collection’s database running Sp7.9.4
Following these instructions produces a recordset larger than the original one, whose contents are the records returned for the second query if run on the whole database, not solely the records within the existing recordset.
Notably, the .csv exported from the subquery is correct. So there is just something wrong with the way the recordset is being produced.
Apologies if this is a known bug. But if not, it should be fixed!
This is the expected behavior at this time. The above instructions specify that you must explicitly select the items to include in the new record set. You can click on one item in the results, scroll to the bottom of the list, and hold Shift while clicking the last item to select all the items in between.
I agree that the ‘Create Record Set’ logic should rely solely on the results returned in the query! I’ve added a GitHub enhancement request: