* Craig Ringer (craig@2ndquadrant.com) wrote:
> > Hmm, the 'gotcha' I was referring to was the issue discussed upthread
> > around rows getting locked to be updated which didn't pass all the quals
> > (they passed the security_barrier view's, but not the user-supplied
> > ones), which could happen during a normal 'update' against a
> > security_barrier view, right? I didn't think that would require the
> > view definition to be 'FOR UPDATE';
>
> It doesn't require the view to be defined FOR UPDATE.
Ok, great, glad I got that correct. :)
> I'll try to write an isolstiontester case to donstrate this on the weekend.
Great, thanks. I'll take a stab at writing up the 'gotcha' note tonight
or tomorrow.
Thanks again,
Stephen