ON 128 TYPES OF NULL

From: Paul Cresham

Date: 2-27-2006

Someone by the name of “Keith W. Hare” posted this on a discussion on NULLs on the SQL Server Central forum:

Early on in the development of the SQL:1999 ANSI & ISO standard, there was a concept of user-defined NULL types. The idea was to allow up to 128 different types of NULLs. One then needed a mechanism for specifying which NULL type, and comparing two NULL types to see if they were the same type of NULL. The concept was very powerful from a database design standpoint, but very complex to specify in the standard. There was no indication that any of the vendors were ever likely to implement the concept, so it was eventually weeded out.

What’s that? 130-valued logic!? Not quite, I don’t think, but the mind boggles at the lunacy nonetheless…

From: Fabian Pascal

As I said so many times: when societies dismiss knowledge and reason, lunacy reigns and kills them. It may take a long time, but it's as certain as human death. We are witnessing the death of the west. The database lunacy is hardly the most profound problem by far.

From: Hugh Darwen

Keith Hare is the current (and probably last) convener of ISO/IEC JTC 1/SC 32/WG 3, the committee that drafts the SQL international standard. The proposal on different kinds of NULL was in the SQL3 draft when I joined in 1989 but didn't stay there long enough for me to study it. I don't think it involved >3-valued logic, but I can't imagine how it could have been anything but shaky.

Posted 4/21/06