Re: Order-by and indexes

Поиск
Список
Период
Сортировка
От Jean-Yves F. Barbier
Тема Re: Order-by and indexes
Дата
Msg-id 20110629173830.0a63992d@anubis.defcon1
обсуждение исходный текст
Ответ на Re: Order-by and indexes  (Odd Hogstad <odd.hogstad@smartm.no>)
Ответы Re: Order-by and indexes  (Odd Hogstad <odd.hogstad@smartm.no>)
Список pgsql-novice
On Wed, 29 Jun 2011 17:19:00 +0200, Odd Hogstad <odd.hogstad@smartm.no> wrote:

...
>
> The ordering of the fk doesn't matter to me now.

It should: it gives the condition that let you get the latest fk...

> Yes, there might be (and
> are) several ones with the same value for this. I just want the latest added
> one that matches. And I don't understand why this is not always the first
> one matching a forward scan, as new entries are put in front?

Because, according to the docs, such a query as you told about *cannot*
guaranty the order of the rows (logical: you ask for all fk=111 but nothing
except sorting on id can insure you'll have fk rows in the right order.)

> > Also I don't
> > understand why the order by query is scanning backwards, when the record I
> > want is in the other end?

Take a sheet of paper and a pencil, write the whole shebang down, make this
model run'by'hand and you'll see why.

>  Because id is the primary key (I guess:) and ordering DESC puts id latest
> > rows first in list, so limiting select to 1 returns the last one.

Anyway, Tom gave you the answer to speed up your query.

--
<stu> Stupid nick highlighting
<stu> Whenever someone starts with "stupid" it highlights the nick.  Hmm.
        -- #Debian

В списке pgsql-novice по дате отправления:

Предыдущее
От: Mike Thomsen
Дата:
Сообщение: Re: Locking out a user after several failed login attempts
Следующее
От: "Jean-Yves F. Barbier"
Дата:
Сообщение: Re: Locking out a user after several failed login attempts