Re: SQL Question

Hi,

I've performed a test case like yours right now. It works fine - the both cases. As it expected to be though.
Oracle 8.1.7.1 on SPARC Solaris 8, but the release and platform shouldn't affect of course.
Could you show us table desc and the whole query too?

Regards,

--
Roman Mirzaitov
Brainbench MVP for Oracle Administration
www.brainbench.com

<meblahyoublah_at_eatme.net> wrote in message
news:3dacc9e0.201937_at_news.efn.org...


> Hi all,

>

> I am running a simple select from SQL*Plus to create a view on an 8i

> database on NT. The last statement in the query reads as follows:

>

> and (date_termination is null OR date_termination >=

> to_date('01-jun-2002','dd-mon-yyyy')

>

> If I run it this way it returns the records with date_termination null

> but not the records with date_termination >= 01-jun-2002. If I reverse

> the ordering of the clauses around the OR as in

>

> and (date_termination >= to_date('01-jun-2002','dd-mon-yyyy') OR

> date_termination is null )

>

> then it returns both the records containing null term dates and the

> records with term dates >= 01-jun-2002. What I don't understand is why

> the change in position would change the result set if the whole

> statement is enclosed in parens. I know that AND binds more tightly

> than OR but the docs say the parens should overide this. Is it an an

> 'unknown' value form the null that is causing the evaluation to

> terminate befoe the second clause is checked? I can make it do what I

> want but i don't understand why which is annoying.

>

> Thank you,

> mark

>

>