Re: Oracle NULL vs '' revisited

On Aug 17, 10:23 pm, "David Portas"
<REMOVE_BEFORE_REPLYING_dpor..._at_acm.org> wrote:
> <zigzag..._at_yahoo.com> wrote in message
>
> news:1187385280.227081.226550_at_g4g2000hsf.googlegroups.com...
>
> > On Aug 17, 10:49 am, Matthew Harrison <m.harris..._at_craznar.com> wrote:
>
> > Any value you put in the field/variable is
> > not same thing as null.
>
> Agreed. Any value is not the same thing as null
>
> > In my view, Oracle's approach is the right approach.
>
> Huh? But Oracle's approach is that the empty string value IS the same thing
> as null!
>
> --
> David Portas

Oracle's approach is that there is no empty string.

While I can sort of see how a second type of emptiness might sometimes be useful, I'm generally glad that Oracle keeps it simple.

There was a long-running debate about this over on OTN forums last year, and I tried to get the poster there to explain why he wasn't concerned about "known no-value" DATEs when "known no-value" strings seemed to drive him crazy, and never got a proper answer there either.

As I understand it, first Oracle came along and handled null strings a particular way; then a couple more rival products appeared that handled them in a different way. Because more than one rival product is different to Oracle, Oracle gets called "non-standard". Is that the basis for "Oracle doesn't adhere to SQL standards for the distinction" or am I missing something?

I must say I have never come across this requirement in any system I have worked on in 18 years so I would be interested to hear the business rule in more detail. Received on Sat Aug 18 2007 - 04:19:37 CDT