On Sep 7, 6:37 am, "Shakespeare" <what..._at_xs4all.nl> wrote:
> "Jeremy" <jeremy0..._at_gmail.com> schreef in berichtnews:MPG.214b41da4e682a7398a49e_at_news.individual.net...
>
>
>
> > In article <46e01453$0$233$e4fe5..._at_news.xs4all.nl>, Shakespeare says...
> >> > select foo.pk, foo.data
> >> > from foo
> >> > where exists ( select null from bar where foo.pk = bar.fk );
>
> >> I would not use select null here... It works, but I have seen some
> >> discussions about null before in this group ;-)
>
> > Can you elaborate on that? Surely it doesn't matter what is selected -
> > it's a case of if the subquery returns 0 rows or not?
>
> > --
> > jeremy
>
> Jeremy,
>
> you're right! But in programs I usually see select '1' from table (and one
> dba consistently used select '8', don't know why, doesn't matter..) . Select
> null from table works (at the moment) but there has been a discussion about
> null and compatibility of the Oracle null with other DB's and languages in
> this group.
> If you have quality assurance in your project, the QA will start asking
> things like 'Are you sure this will keep working?': null is a QA-trigger....
> If you use '1' no one will ask any questions.
>
> Again: I know ths works (even in 11g!) but I would avoid any discussion
> (like this one) and use select '1' ....
>
> Not a big issue however... but the discussion got really out of control...
>
> Seehttp://groups.google.nl/group/comp.databases.oracle.server/browse_thr...
>
> Shakespeare
Thanks Shakespeare. I'll keep that in that back of my mind, but I
don't worry too much about writing cross-platform-compatible select
statements. And, any QA process that simply evaluates my code by
"hitting" on keyword triggers like my code is a google search can
respectfully kiss my behind.
Received on Fri Sep 07 2007 - 09:36:33 CDT
