Path: dp-news.maxwell.syr.edu!spool.maxwell.syr.edu!drn.maxwell.syr.edu!news.maxwell.syr.edu!postnews1.google.com!not-for-mail
From: jcelko212@earthlink.net (--CELKO--)
Newsgroups: comp.databases.theory
Subject: Re: NULL
Date: 1 Oct 2004 06:35:50 -0700
Organization: http://groups.google.com
Lines: 20
Message-ID: <18c7b3c2.0410010535.ab37196@posting.google.com>
References: <415b0d81$0$48933$e4fe514c@news.xs4all.nl> <fPOdnddjv8iFj8HcRVn-sQ@comcast.com> <415c6c73$0$10528$e4fe514c@news.xs4all.nl> <cji0u6$u54$1@news.netins.net> <415c9c59$0$78738$e4fe514c@news.xs4all.nl>
NNTP-Posting-Host: 66.68.34.240
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-Trace: posting.google.com 1096637750 31981 127.0.0.1 (1 Oct 2004 13:35:50 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Fri, 1 Oct 2004 13:35:50 +0000 (UTC)
Xref: dp-news.maxwell.syr.edu comp.databases.theory:26369

>> If somebody could shed some light on the early history of NULL in
databases, I would be grateful. Pure curiosity. <<

I think they first came out of System R at IBM because Dr. Codd had
them in his papers.  The theory says they have no data type and have
no value; the implementations have to reserve storage, so you have
things like "CAST (NULL AS <datatype>)" in SQL.

Before that, Statistic packages had ways of handling missing data -- I
think SPSS was the first one.  But they were often resolved a
fictional value for computations.

I would have to look this up, but I recall that SPARC had a paper in
the 1970's on over a dozen diffrent kinds of missing data -- overflow,
underflow, divide by zero, illegal bit patterns, unknown, missing,
N/A, impossible value, etc.   And of course IEEE floating point has
"Nan" which is like a missing value marker.

In spite of what Fabian Pascal says, applied math spends a lot of time
on "things we don't know"  ...