Re: SQL Gurus ...

On Thu, Mar 11, 2021 at 7:26 PM Jon Paris <jon.paris@xxxxxxxxxxxxxx> wrote:

Can someone please explain how I can ever get a lower row count from a left outer join than the number of rows that exist in the left table?

select count(*)
from table1 a left outer join table2 b
on a.email = b.email
where b.event = '2103SM'
order by attend;

That returns query returns a count of 848 rows. There are 858 rows in table1! (left table). And just for fun an inner join also returns 848 rows.

Any ideas anyone? I thought it was an absolute that a left outer always returned all rows in the left table.

I think Peter's explanation is correct. To put it another way: Your
left outer join DID return all the rows in the left table. But THEN
you filtered the result of the join with your WHERE clause.

Surely, if you had instead written

select count(*)
from table1 a left outer join table2 b
on a.email = b.email
where a.event = '2103SM'
order by attend;

(note the a.event instead of b.event) then you shouldn't reasonably
expect that rows where a.event <> '2103SM' are included. Otherwise,
what does it even mean to include a.event in the condition?

Your indentation hints at your intention, and Peter provides the most
straightforward rewording to capture that. I think most people would
indent your original query with the WHERE lined up with the FROM and
ORDER BY, and given the actual workings of it, I'm sure you can
understand why.

John Y.