Thank you. This is exactly the way I see it as well. I don't see this behaviour as a negative, in fact i see it as a positive. "Optimizing 'distinct' away" as ovid puts it is not done well because its not easy to do. And not making queries by default dupecheck means a temporary sort table or hash table doesn't have to be constructed, maintained and then thrown away for every query. The result set can be just piped back. For most well designed properly constrained databases there is no need for dupechecking, and when there is the need 'distinct' or 'group by' are waiting to handle it.

BTW, IMO there are lots of reasons to bitch about SQL, but for me they are mostly syntax related. SQL was IMO not designed by a capable language designer, its syntax is horrible. (I'm not saying I could do better, but Im pretty sure there are folks out there that could.)