"Joe Conway" <joe(at)conway-family(dot)com> writes:
> While thinking about how to implement TABLE_PRIVILEGES view per SQL92, I
> realized that a function is needed to dynamically return the current
> database name, and current database owner.
Just out of curiosity, why are these needed for TABLE_PRIVILEGES?
> current_dbname() - returns name of current database
> current_dboid() - retuns oid of current database
> current_dbownername() - returns owner name of current database (based on
> current_dbownerid() - returns owner usesysid (well, more accurately datdba)
One thing to bear in mind is that the existing notion of databases
may need to be rethought when we implement SQL-style schemas (which
I hope will happen for 7.2). It might be best to refrain from inventing
more functions until we see how the schema work settles out.
I haven't yet heard any serious discussion about how to integrate
the SQL-defined schema concepts with Postgres' traditional concept of
independent databases within an installation. It's about time to have
that discussion though ...
regards, tom lane
In response to
pgsql-hackers by date
|Next:||From: Bruce Momjian||Date: 2001-05-15 04:45:08|
|Subject: Re: Re: bug in pgcrypto 0.3|
|Previous:||From: Bruce Momjian||Date: 2001-05-15 04:40:54|
|Subject: SCHEMA idea|
pgsql-patches by date
|Next:||From: Joe Conway||Date: 2001-05-15 06:19:01|
|Subject: Re: Re: [HACKERS] contrib |
|Previous:||From: Bruce Momjian||Date: 2001-05-15 04:27:43|
|Subject: Re: Remove columns pg_index.haskeytype and pg_index.indisclustered|