On Dec 18, Stephen Frost modulated:
> I agree that it would be a nice addition, as I've said before.
OK, sorry I misunderstood earlier and thought you were dismissing the
idea as redundant with triggers. Thanks for your patience!
> ... We certainly won't be adding it into 9.5 and it's getting pretty
> late for 9.6 too, but I'm anxious to see just how RLS is used in the
> field
If we want to start testing with 9.5 and emulate an environment
supporting NEW and OLD in the CHECK policy for UPDATE, can you
recommend how we should do that? E.g., is there a particular trigger
idiom that would most closely replicate the RLS extension we've been
discussing (so that we'd know that a working policy set + triggers
could be translated to just RLS policies if that feature is added)?
We'd be happy to experiment with these sorts of policies in our
applications. My intuition is that with this extension, we'd be able
to push down nearly all of our important policy enforcement into
PostgreSQL and guard against a slew of potential application
programming errors (compared to handling all this enforcement in our
application data access code)!
Karl