From: | "Dave Page" <dpage(at)vale-housing(dot)co(dot)uk> |
---|---|
To: | "Andreas Pflug" <Andreas(dot)Pflug(at)web(dot)de>, <pgadmin-hackers(at)postgresql(dot)org> |
Subject: | Re: contribute pg_get_viewdef2 et al |
Date: | 2003-05-07 14:03:48 |
Message-ID: | 03AF4E498C591348A42FC93DEA9661B83AF076@mail.vale-housing.co.uk |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgadmin-hackers |
> -----Original Message-----
> From: Andreas Pflug [mailto:Andreas(dot)Pflug(at)web(dot)de]
> Sent: 07 May 2003 14:57
> To: Dave Page; pgadmin-hackers(at)postgresql(dot)org
> Subject: Re: contribute pg_get_viewdef2 et al
>
>
> That's really a matter of philosophy. Taking MSSQL as an example, the
> view wouldn't be runnable any more, if tables or columns are
> renamed. On
> the other hand, tables can be dropped and recreated, and the
> view will
> still be runnable because the saved plan is dropped and will
> be created
> from source the first time it is used again.
I can't see many people voting for a change to that behaviour. It's too
big, with too little gain.
> As a solution in pgsql, there are two ways (combinable)
> - Preventing table and column rename, if referenced by rules or views
> (ALTER TABLE xx RENAME TO xx2 RESTRICT), just as DROP does
> - invalidating the source, so only the reverse-engineered node
> representation is available (ALTER TABLE xx RENAME TO xx2 CASCADE)
Possibly. You'd have to raise it on the hackers list, and see what the
response is.
Regards, Dave.
From | Date | Subject | |
---|---|---|---|
Next Message | Adam H. Pendleton | 2003-05-07 18:35:51 | Pgadmin III and wxWindows |
Previous Message | Andreas Pflug | 2003-05-07 13:56:56 | Re: contribute pg_get_viewdef2 et al |