#34845: Slow query when using exclude
-------------------------------------+-------------------------------------
     Reporter:  Amin Aminian         |                    Owner:  nobody
         Type:                       |                   Status:  closed
  Cleanup/optimization               |
    Component:  Database layer       |                  Version:  4.2
  (models, ORM)                      |
     Severity:  Normal               |               Resolution:  duplicate
     Keywords:  exclude,filter       |             Triage Stage:
                                     |  Unreviewed
    Has patch:  0                    |      Needs documentation:  0
  Needs tests:  0                    |  Patch needs improvement:  0
Easy pickings:  0                    |                    UI/UX:  0
-------------------------------------+-------------------------------------

Comment (by Simon Charette):

 To add to the above, assuming you are using
 `annotate(payment=FilteredRelation("payments",
 Q(payments__payable_type=payable_type)))` or an equivalent, the reason why
 the ORM cannot simply use a `LEFT OUTER JOIN` instead of `NOT EXISTS` when
 doing `exclude(payment__isnull=False)` is that it lacks introspection
 capabilities into `FilteredRelation.condition` to be sure that the
 condition added to the `JOIN` wouldn't interfere.

 In a sense

 {{{#!python
 INNER JOIN "payment" ON ("order"."id" = "payment"."payable_id"
                                             AND
 ("payment"."payable_type_id" = 30))
 WHERE "payment"."id" IS NOT NULL
 }}}

 is a ''tautology''. The `WHERE "payment"."id" IS NOT NULL` part is
 unnecessary given the usage of an `INNER JOIN` already ensures that
 `"payment"."id" IS NOT NULL` since primary keys cannot be null.

 Normally the ORM would be able to ''know'' that `"payment"."id"` cannot be
 null as it's a primary key but I suspect the usage of a `FilteredRelation`
 prevents the ORM from knowing.

 What #34597 did is switch the query from using `"order"."id" NOT IN
 (subquery using  LEFT JOIN)` to `NOT EXISTS(subquery using  LEFT JOIN
 referring to "order"."id" )` so if you can confirm that your query is
 faster with `NOT IN` than with `EXISTS` it would be great to have another
 data point added to #34597.

-- 
Ticket URL: <https://code.djangoproject.com/ticket/34845#comment:8>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.

-- 
You received this message because you are subscribed to the Google Groups 
"Django updates" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-updates/0107018aa8ed1cfc-383fd807-95cd-47aa-a6ea-3c64e465c95e-000000%40eu-central-1.amazonses.com.

Reply via email to