Re: Thus spoke SQL3 (on OO)

Marten Feldtmann <marten@feki.toppoint.de>

From: Marten Feldtmann <marten@feki.toppoint.de>
To: Chris Bitmead <chris@bitmead.com>
Cc: Hannu Krosing <hannu@tm.ee>, Chris Bitmead <chrisb@nimrod.itg.telstra.com.au>, PostgreSQL Development <pgsql-hackers@postgresql.org>
Date: 2000-05-22T16:56:02Z
Lists: pgsql-hackers
> Hannu Krosing wrote:
> 
> It's not such a big deal really. When you do an OO model you don't need
> to think about your own primary key.
> 

 Hmm, I see here more and more postings, that do say, the OID (or the
result of a SEQUENCE) is usable for a key to identify an object stored
within a database.

 Though it's true, that SEQUENCE can be used to create unique
identifiers, the function is simply a hack - nothing more for greater
OO software systems and worse than software solutions, which provide
more power and lower traffic.

 The identification of an object has to be based on a unique key and
it does not matter of which type it is.

 The foreign key is of course not useful for the oo-model, but for the
programmer, which produces the object-relational wrapper this is VERY
urgent !

 And here again: if you use SEQUENCE for the OID you use a special
feature of the database ... and that is bad.

 Marten