>
> And Postgres is not very forgiving if it ever figures out that it's been
> tricked...that definitely needs to be worked out.
>
What I'm sort of leaning towards is to catch these attempted writes early
on, and not have to deal with a lot of cleaning up after the fact.
I guess Postgres already supports the concept of a "read-only" database from
a user permissions perspective, right? So maybe take advantage of that
existing functionality?