Re: Schema (namespace) privilege details - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Schema (namespace) privilege details
Date
Msg-id 15706.1019179169@sss.pgh.pa.us
Whole thread Raw
In response to Re: Schema (namespace) privilege details  ("Rod Taylor" <rbt@zort.ca>)
List pgsql-hackers
"Rod Taylor" <rbt@zort.ca> writes:
> [ how it ought to be to support hosting companies ]

I'm not real comfortable with this.  The design I proposed is based
fairly firmly on the Unix directory/file protection model --- which
is assuredly not perfect, but it's survived a lot of use and is not
known to have major flaws.  You're suggesting that we should invent
a protection model off-the-cuff on the basis of the supposed needs
of one class of application.  I think that's a recipe for trouble...

> I'd really like to see a schema owner have full control over all
> objects in a schema, and likewise a database owner have full control
> over their database.   My POV for large systems.

Those things are both easily done: just don't allow anyone else to
create objects in your schema (resp. database).  This is indeed what
SQL99 envisions.  However, in a database where there are multiple
users sharing schemas, I am not convinced that the notion "the schema
owner has ALL rights to objects within the schema" is appropriate.
That seems to me to go way too far; if we are troubling to maintain
distinct ownership of objects within a schema, that should mean
something.  In particular, the guy who is not the schema owner should
be able to have some confidence that the guy who is can't make arbitrary
changes in his table.  Otherwise the schema owner is effectively
superuser, and what's the point of pretending he's not?
        regards, tom lane


pgsql-hackers by date:

Previous
From: "Rod Taylor"
Date:
Subject: Re: Schema (namespace) privilege details
Next
From: Joe Conway
Date:
Subject: Re: Schema (namespace) privilege details