Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ucbvax!mtxinu!sybase!jeffl@sybase.Sybase.COM
From: jeffl@sybase.Sybase.COM (Jeff Lichtman)
Newsgroups: comp.databases
Subject: Re: Outerjoin implementation?
Message-ID: <11034@sybase.sybase.com>
Date: 24 Sep 90 01:53:34 GMT
References: <1990Sep20.161204.13014@tfic.bc.ca>
Sender: news@Sybase.COM
Distribution: comp
Lines: 21

> 	select qqsv,count(*) from foo
> 	group by qqsv
> produces a separate line for every row in foo that has a null in qqsv...

It's not supposed to.  This is one of those cases where nulls are supposed
to act like they're equal to each other.  Does Informix SQL treat each null
as a separate group?

BTW, I agree that nulls in SQL leave a lot to be desired.  I'm not convinced,
though, that default values are adequate in place of nulls.  There's the
problem of having to choose a default that's guaranteed never to be represented
by known data values.  There's the problem of accidentally using a default
value in a calculation as if it were a true value (calculations with null
always yield null).  There's the problem of storage space - many database
systems don't use any space on disk for null values, which can save a lot
in sparsely populated tables.  There's the problem of naive users incorrectly
interpreting default values as true data.  And on and on....
---
Jeff Lichtman at Sybase
{mtxinu,pacbell}!sybase!jeffl  -or- jeffl@sybase.com
"Saints should always be judged guilty until they are proved innocent..."