Re: Warning about null and open question to Oracle

From: Scott Urman <surman_at_oracle.com>
Date: 1995/06/14
Message-ID: <3rlcc8$1uk_at_inet-nntp-gw-1.us.oracle.com>#1/1


mccusker_at_fast.net (Jim McCusker) wrote:
>In article <Pine.NXT.3.90.950613090306.17269C-100000_at_iluvatar>,
> John Jones <john_at_iluvatar.tip.duke.edu> wrote:
>>I appreciated everyone's response to this, but the majority of responses
>>have been to just wrap a nvl around everything. That is ok, but when you
>>are comparing around 50 or more columns that is a lot of typing and as a
>>programmer I look for ways to cut down on typing as much as possible. I
>>just think that NOTHING should be equal to NOTHING. I have heard that
>>other databases do this and was really just sounding off hoping Oracle
>>would do the same. Oh well, can't have everything I guess. Thanks for
>>listening.
>
>There's a very academic feel to their arguments don't you think? Essentially,
>it is argued that since a value is NULL it was not defined and therefor Oracle
>doesn't know how to deal with it. But lets be real, what really happened is
>that when the record was created the field was not initialized to anything.
>This allows us to make a distinction between a zero-length string and a string
>that wasn't assigned a value. The argument that NULL <> NULL is purely
>academic in my opinion and doesn't help matters.
>
> -- Jim
>
>

Actually, Oracle made an intentional decision to have NULL be the same as a zero length string. This is because there are a lot of applications written which concatenate strings together. This way every expression x || y doesn't have to have NVL calls around x and y. Received on Wed Jun 14 1995 - 00:00:00 CEST

Original text of this message