Re: Another angle on this....

Heinz Kiosk <no.spam_at_ntlworld.com> wrote in message news:e22d8.1376$Ah1.115361_at_news2-win.server.ntlworld.com...
> And Oracle doesn't anywhere? Just curious. The notation for outer joins
eg.
> I'm not aware of the standard, just common practice which differs from
> Oracle in that respect.

The standard was set a very long time after Oracle was produced and is now supported I believe.

> > And they will return rows where col1 IS NULL.
> I can't see the mangling of = to IS where rhs is a constant null and lhs
> some kind of expression as the worst bug in the world. The intention of
the
> code is completely clear. Unlikely to introduce errors or problems on
that
> platform but allows the uninformed to create code thats a bitch to port
> because a correct handler will reject the lines.

Taking your outer join example, move between ss and oracle and you will get an immediate error message. The difficulty with the null / empty string problem is that it fails silently, so unless you test your code thoroughly you may not find out until there is a corrupt database.

Of course, we all test properly, so this is unlikely to be a real-world problem. Received on Thu Feb 21 2002 - 15:10:25 CST