Re: TODO item question [pg_hba.conf] - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: TODO item question [pg_hba.conf]
Date
Msg-id 20060421162600.GB7303@surnet.cl
Whole thread Raw
In response to Re: TODO item question [pg_hba.conf]  ("Gevik Babakhani" <pgdev@xs4all.nl>)
List pgsql-hackers
Gevik Babakhani wrote:
> Thank you :)
> 
> > You are missing an ACL_*_CHR symbol and updating the ACL_ALL_RIGHTS_STR
> > symbol.
> 
> That is why I could not see the new permission in pg_database.
> I was actually looking for that for sometime :)
> 
> I have added the ACL_*_CHR 'D' Is this okay?

Hum, you literally added a symbol ACL_*_CHR?  I was actually thinking in
ACL_CONNECT_CHR or something like that ...

While at it, why D?  Isn't 'c' more natural?  (And conveniently unused.)


> > Also, you should know that changing this requires a change in
> > CATALOG_VERSION_NO in catversion.h as well.
> 
> Why is this needed? Is this a functional requirement?

To force an initdb, because you are causing a system catalog change.
Now that I think about it, maybe it's not needed, because the default
state of the system should be the same as if no privilege has changed.

OTOH you need to speficy the interpretation of the initial state of the
ACL for a database.  I think it should mean that PUBLIC has the CONNECT
privilege.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support


pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: Checking assumptions
Next
From: Tom Lane
Date:
Subject: Further reduction of bufmgr lock contention