> On Sun, Dec 28, 2008 at 9:29 AM, Bernd Helmle <bernd(at)oopsware(dot)de> wrote:
> i would say check for negative attnums and deny that view to be
> updateable because of SQL92 says in 11.19 <view definition> syntax
> rule 6:
> 6) If the <query expression> is updatable, then the viewed table
> an updatable table. Otherwise, it is a read-only table.
> wich i understand as deny updatability in any view that constains non
> updateable <query expression> in the target list
Yes, but afaiks SQL99 allows partial updatable column lists, so we could
argue that we can follow this. However, it seems your approach is better
> yes. if we didn't do that we will be against spec. syntax rule 12
> (again in 11.19 <view definition> ) says:
> 12)If WITH CHECK OPTION is specified, then the viewed table shall
> be updatable.
I'm currently travelling (visiting my parents during turn of the year),
checking my mail sporadically. I'll provide an updated patch ASAP.
In response to
pgsql-hackers by date
|Next:||From: Bruce Momjian||Date: 2008-12-29 21:02:04|
|Subject: Re: Idea - fallback mode for psql backslash
commands using information_schema|
|Previous:||From: Jeff Davis||Date: 2008-12-29 19:59:16|
|Subject: Re: TODO items for window functions|