Yeah the RECURSIVE vs ITERATIVE is a bit of a red herring here. As you say, the RECURSIVE keyword doesn't specify the processing but marks the fact that the specification of the query is recursive.
Agreed. I started thinking through Fabien's response last night.
I think a syntax that would fit better within the existing framework would be something like
WITH RECURSIVE t AS ( SELECT base case REPLACE ALL -- instead of UNION ALL SELECT recursive case )
I was originally thinking more along the lines of Fabien's approach, but this is similarly interesting.
Obviously I'm very concerned about doing something that the SQL Standard will clobber somewhere down the road. Having said that, the recursive syntax always struck me as awkward even by SQL standards.
Perhaps something like this would be more readable
WITH t AS ( UPDATE ( SELECT 1 AS ctr, 'x' as val ) SET ctr = ctr + 1, val = val || 'x' WHILE ctr <= 100 RETURNING ctr, val )
