Re: ID field as logical address
Date: Sat, 6 Jun 2009 17:21:19 -0700 (PDT)
Message-ID: <f6b766b0-3b71-453e-8af9-e92db13c91d4_at_q37g2000vbi.googlegroups.com>
On May 29, 5:54 am, "Walter Mitty" <wami..._at_verizon.net> wrote:
> In another thread, I suggested that prefixing every tuple with an
> auto-generated field (column) called ID amounted to reference by location
> rather than reference by content. ...
Yes, this is quite a natural question to wonder. If table columns are
labeled by column names, why rows can't be labeled by some sort of
IDs. If table header can be viewed as a hidden row, then why can't one
introduce a hidden ID column? A while ago there was a discussion
thread here on c.d.t where someone (I believe "VC") convincingly
argued that there are fundamental obstacles for such a construction.
The discussion was in the RL context, and my latest understanding of
the issue is documented on page 11 of
http://arxiv.org/abs/0902.3532
Received on Sun Jun 07 2009 - 02:21:19 CEST