Re: So what's null then if it's not nothing?

Hugo Kornelis wrote:

[snip]

> Hi Mike,
>
> I'm not sure if it will survive past all filters, but I just posted an
> empty message in reply to this message.

[snip]

> I did NOT write a reply to Alfredo's post.
>
> If you check your news reader, you should be able to see the difference
> between my empty post to you, and the lack of any post by me to Alfredo.
>
> If a database can't distinguish between these two categories, it's unfit
> to be used for representation of Usenet discussions. (And for a whole
> lot of other applications as well).

Just FYI - on Pick there'd be an empty item in one case (your empty response) and no item in the other (your non-existent response). The non-existent item would not contain any nulls. The empty item would contain just the one null. Some other items, with non-null responses, might contain 2 or more nulls. There could even be items that are themselves non-null but contain nothing but nulls. Getting a bit silly now though.

>
> >You will say that at the logical level something can exist - in that it
> >is part of a set - and yet have an unknown value. You would, presumably,
> >also say that it does not necessarily have a value equal to empty
> >string, because it's unknown. In order to record that "fact" you (or
> >the SQl engine) have/has to store some code at the physical level to
> >say it's not an empty string - it's null. It's not though is it? It's
> >whatever you stored to represent "unknown".
>

[snip]

>
> HOW a Null is represented is irrelevant. WHAT it represents is what
> matters.
>

Fair enough. I simply believe NULL is actually nothing. I think it should be respresented as nothing. Something that is unknown is something else entirely, and can/should be represented in an entirely different way. Whatever it is it's not NULL - not in my book, or my dictionary, or, apparently, in the ANSI standard. Just in SQL. Am I wrong?

> Best, Hugo
> --

Regards
Mike. Received on Fri Nov 18 2005 - 01:57:25 CET