Tom Lane wrote:
> ... but when it is TRUE, there should be a notice, same as there is in
> SELECT. UPDATE should produce such a notice too, IMHO. Probably we
> omitted the message originally because there was no way to avoid it
> in a DELETE, but now there will be.
Well, my previous message described why I'm not sure that this line of
reasoning is correct. I think the only really proper configuration is
add_missing_from=false and an explicit USING/FROM list. Just about the
only reason to enable add_missing_from would be for compatibility with
previous releases of PostgreSQL -- and that "compatible" behavior is not
to issue a warning for UPDATE and DELETE in this situation. If the user
deliberately enables add_missing_from, I'm inclined to trust them that
they know what they're doing.
-Neil
