[lug] postgres question

[dan radom]

* Hugh Brown (hugh at math.byu.edu) wrote:
> On Thu, 2003-11-13 at 08:51, rm at fabula.de wrote:
> > On Thu, Nov 13, 2003 at 08:09:59AM -0500, Hugh Brown wrote:
> > > [...]
> > > is tedious.  I falsely assumed that 
> > > 
> > > GRANT { { CREATE | TEMPORARY | TEMP } [,...] | ALL .....
> > > 
> > > meant that I could do a grant on DATABASE dbname and have all
> > > tables,views,sequences, etc. automatically get their privileges
> > > updated.  Postgres is version 7.3.2 (built from the redhat rpms for rh9
> > > I believe).
> > 
> > 
> > Yes, that's a nasty bit of postgres' security implementation.
> > An, unfortunately, it's not possible to replace the relation
> > name with a subquery (the query parser will grok on "GRANT ALL 
> > ON (SELECT relname FROM pg_class ...").
> > 
> > Tell me if you find a workarround, i was bitten by that myself
> > a few times.
> > 
> >   Ralf Mattes
> 
> For now my workaround is
> 
> postgres: createuser foo
> postgres: createdb foodb
> postgres: grant all on database foodb to foo
> foo: \i db-dump
> 
> Hugh
> 

from the postgres interactive documentation...

The
"grant all on database mydb to myuser" syntax doesn't work, however the
help above says it to work.
I can't imagine, that if I would like to grant full access over a
complete DB to a new user, then I have to do it from table-to-table and
seq-to-seq...

perhaps this will be addressed in 7.4, which is currently in RC2.  I've
heard very nice things about it, and would love to upgrade to experience
the reported 100-150 % speed increase, but i'll be waiting for the 7.4
release.  i've had to do enough initdb's between rc's and releases to
want to wait for 7.4 final.

mmmm postgres!