Malcolm


And how is your post or Codd's rules against the fact that emptystring = null?

Roy Lambert

Roy Lambert wrote:

>
> And how is your post or Codd's rules against the fact that
> emptystring = null?
>


I would expect an empty string to be not null - but a missing string to
be null.   <shrug>

--

Malcolm


Its what I love about this debate, everywhere I've seen posts about it.

My view is that to be consistent NULL = empty.

But I know we'll never agree

Roy Lambert

And the nice thing about this ng is that we can disagree without any
fisticuffs .. er .. can't we.  

--

Malcolm


Well - I'd have to have very long arms to reach you

Roy Lambert

"Malcolm" <malcolm@spam.will.bounce> wrote in message
news:88548CC1-5AC1-485D-B9DD-B36CA35BB784@news.elevatesoft.com...
> And the nice thing about this ng is that we can disagree without any
> fisticuffs .. er .. can't we.  

Whimp
Rita

Rita wrote:

> Whimp

...or even whimper.  

--

Roy,

Here we go again. NULL is a missing value, while '' can be a real value
for the field. For example, some people have middle initials and some don't.
In case we know it's empty, we specify ''. In case we have not entered
this information (the value is not present, so we don't know) - it's NULL.
This is really very simple and is Ansi standard and there can be no argument.
Also, all field types must be treated in the same way and if all of them
allow NULL as well as, say, 0 value, then for strings consider '' same as
0 for natural numbers.

It's a different story about how should we handle it, but my suggestion
stays valid - declare all fields NOT NULL if you don't need that flag,
end of story. The fact that your old databases were not designed correctly
and that DBISAM does not adhere to standard does not mean that people
should continue repeating these same mistakes.

Regards,
Michael

Michael


>Here we go again. NULL is a missing value, while '' can be a real value
>for the field.

Yup - here we go again parrot the standard rather than think.

>For example, some people have middle initials and some don't.
>In case we know it's empty, we specify ''.

Its a good job you're not here, my scream of agony at having that particular nonsense example quoted yet again would have ruptured your eardrums.

>This is really very simple and is Ansi standard and there can be no argument.

So its not possible to criticise or debate a standard! Piffle and twaddle. If that view were correct nothing would change.

>Also, all field types must be treated in the same way and if all of them
>allow NULL as well as, say, 0 value, then for strings consider '' same as
>0 for natural numbers.

I want to treat all fields the same way, but unlike you I do not accept that and empty string is the equivalent of zero.

>It's a different story about how should we handle it, but my suggestion
>stays valid - declare all fields NOT NULL if you don't need that flag,
>end of story.

If you have to do that its basically a statement that the standard is wrong.

The fact that your old databases were not designed correctly
>and that DBISAM does not adhere to standard does not mean that people
>should continue repeating these same mistakes.

They were not designed incorrectly, they are not mistakes.

If the validity of NULL <> empty string was self evident there would be no need for Firebird to produce a 51 page manual nor would there be many people like myself who disagree.

Roy Lambert