Philip Warner <pjw@rhyme.com.au> writes:
>> What do people think of my implicit-GROUP-BY-ctid idea?
>> That would basically say that the aggregate is computed over all the
>> tuples that join to a single target tuple.
> Sounds perfect to me...
Note that it would not meet your expectation that
update t1 set f2=count(*) from t2 where t1.f1=2 and t2.f1=t1.f1 ;
means the same as
update t1 set f2=(Select Count(*) from t2 where t2.f1=t1.f1) where
t1.f1 = 2
... at least not without some kind of outer-join support too. With
an inner join, t1 tuples not matching any t2 tuple wouldn't be modified
at all.
regards, tom lane