Re: Update and Insert in a View Insert Rule

Поиск
Список
Период
Сортировка
От Stephan Szabo
Тема Re: Update and Insert in a View Insert Rule
Дата
Msg-id 20020521151829.T20839-100000@megazone23.bigpanda.com
обсуждение исходный текст
Ответ на Update and Insert in a View Insert Rule  (Tom Jenkins <tjenkins@devis.com>)
Ответы Re: Update and Insert in a View Insert Rule  (Tom Jenkins <tjenkins@devis.com>)
Список pgsql-general
On 21 May 2002, Tom Jenkins wrote:

> Hello all,
> I have a table jobs that holds both historical and current jobs:
> jobid  SERIAL
> jobemployee  INT4
> jobiscurrent INT2
> etc
>
> the users manipulate two views: historicaljob and currentjob.  These
> views are simply defined by the value of jobiscurrent (0 for historical,
> 1 for current - yes i know it should be a boolean but erwin won't
> generate a postgres boolean may it rot in hell)
>
> now i have the insert rule working fine for historical jobs.  however
> for currentjob, there is a small twist.  First the old current job must
> be set to historical, then the new current job inserted.
>
> my insert rule is:
>
> CREATE RULE insert_current_job AS
>    ON INSERT TO currentjob
>    DO INSTEAD
>        UPDATE job set jobiscurrent=0, lastuser=New.lastuser
>           WHERE jobemployee = NEW.jobemployee and jobiscurrent=1;
>        INSERT INTO job (
>             jobemployee,
>             jobagencybureau,
>             jobbranch,
>             blah, blah, blah,
>             jobiscurrent
>        ) VALUES (
>             NEW.jobemployee,
>             NEW.jobagencybureau,
>             NEW.jobbranch,
>             blah, blah, blah,
>             1
>        )
>
> unfortunately this gives me an error when I attempt to load the rule I
> get:
> ERROR:  NEW used in non-rule query
>
> I don't understand why this wouldn't work.  Unless it has something to
> do with NEW getting "lost" in the update call?  should i move the update
> out to a function and calling it from the rule?

I think you forgot to use the () around the multiple actions.
It should probably be:
DO INSTEAD (
 ...
);



В списке pgsql-general по дате отправления:

Предыдущее
От: "Gaston Micheri"
Дата:
Сообщение: Graphical Tool
Следующее
От: "Miguel Carvalho"
Дата:
Сообщение: Re: Mysterious locking problem