Re: ADD FOREIGN KEY (was Re: [GENERAL] 7.4Beta) - Mailing list pgsql-hackers

From Andreas Pflug
Subject Re: ADD FOREIGN KEY (was Re: [GENERAL] 7.4Beta)
Date
Msg-id 3F797F17.2090404@pse-consulting.de
Whole thread Raw
In response to Re: ADD FOREIGN KEY (was Re: [GENERAL] 7.4Beta)  (Bruce Momjian <pgman@candle.pha.pa.us>)
List pgsql-hackers
Bruce Momjian wrote:

>Fact is, folks are doing it anyway by modifying pg_class.  I know one
>guy who did it in a transaction so he was the only one to see the
>triggers disabled!  The PostgreSQL cookbook page has an example too. 
>People are always asking how to do this.  Why not just make it setable
>only by the super-user.
>
>FYI, TODO has:
>
>    * Allow triggers to be disabled [trigger]
>    * With disabled triggers, allow pg_dump to use ALTER TABLE ADD FOREIGN
>      KEY
>

For practical reasons, I'd prefer the "disable trigger" not to influence 
fk triggers, or at least to have such a default flavor. When restoring a 
database, you might consider the data as consistent and complete, so no 
triggers and ref checks are needed at all. But in the cases of some kind 
of application data import, you might like the data to have fk ref 
checked, but don't want to trigger all user triggers.
The implementation of fk checking by triggers should normally be hidden 
to the user.

Regards,
Andreas



pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: ADD FOREIGN KEY (was Re: [GENERAL] 7.4Beta)
Next
From: Patrick Welche
Date:
Subject: Re: ecpg doesn't compile (datetime.h/dtime_t)