Re: not using table aliases in where clause slow-down?
От | Richard Huxton |
---|---|
Тема | Re: not using table aliases in where clause slow-down? |
Дата | |
Msg-id | 00d901c09a56$778bf5e0$1001a8c0@archonet.com обсуждение исходный текст |
Ответ на | not using table aliases in where clause slow-down? ("Mark Cowlishaw" <markc@ot.com.au>) |
Список | pgsql-general |
From: "Mark Cowlishaw" <markc@ot.com.au> > > I noticed that running queries that do -not- use declared table aliases in > the 'where' clause seem to run a hell-of-a-lot slower than when aliases are > used. Is there a valid reason for this? It started out as a typo but now I'm > curious. > > (7.0.3) > > eg: > > select [snip] > from > releases as rel, subsystems as subs, > functions as func, purposes as purp, procedures as proc > where > rel.project_id = 53 [snip] > > (runs pretty much instantly) > > -versus- > > > select [snip] > from > releases as rel, subsystems as subs, > functions as func, purposes as purp, procedures as proc > where > releases.project_id = 53 [snip] > (I get sick of waiting after > 30sec and ^C the query (in psql)) > > -- > Mark Cowlishaw <markc@ot.com.au> I think the second where refers to a different "releases" so it's equivalent to doing a join without a joining condition (set product? can't remember - need more coffee). I think it's the same as putting another "releases" in "from" is what I'm trying to say (badly) - Richard Huxton
В списке pgsql-general по дате отправления: