>>> To prevent this, I changed the
>>> behavior of SET search_path so that it validates its input.
>
>> ... It would be possible to make interactive SET behave differently
>> from the non-interactive case,
>
> Wait a minute --- scratch what I said above; interactive "SET
> search_path" already does behave differently from noninteractive.
> So what did your patch change exactly?
I think (don't know all of the ways there are to SET search_path), the
interactive way to SET search_path. :) It changed assign_search_path
in catalog/namespace.c, iirc. I said in my original email that someone
could still do ALTER USER SET search_path and have that work without
checking.
-sc
--
Sean Chittenden