Re: Order of where clause matters!

I must be sleeping. the examples are the same, sorry.

Evan wrote:

> Dave Wotton wrote:
>
> >
> > Sorry, not true. The documentation about operator precedence is possibly
> > misleading.
> >
> > Although the Oracle7 docuementation says:
> >
> > "You can use parentheses in an expression to override operator
> > precedence. Oracle7 evaluates expressions inside parentheses before
> > evaluating those outside. "
> >
>
> Oracle 8 documentation says the same thing. However, considering the
> documentation misleading is being a bit courteous. Oracle's license (as does
> most software) has a clause warrenting that the software conforms to documented
> behaviour.
>
> In the past, Oracle has accepted bug reports and provided patches on the
> non-conforming operator precedence issue. Unfortunately, such conformance costs
> processing overhead, so they routinely slide back to the non-conforming state.
> Dba's prefer the hit and miss/workaround in this case rather than accept the
> penalty of processing overhead. Generally, its not too much of an issue.
>
> Your example takes advantage of Oracle's stack oriented parser. This behaviour
> in the parser is well known and such predictability is definitely of value. The
> original poster supplied an example where the divison had to be moved to the top
> of the where clause before it would work, directly conflicting with your
> example. As well, the behaviour changed when an index is mixed into the pot. The
> optimizer stikes again..
>
> Basically, between the original poster's example and your example, Oracle
> demonstrates a distinct error (ie non-conformance to their documentated
> behaviour and lack of predictability and control). This can't be considered
> acceptable/normal. A bug report seems an appropriate course of action. Hope you
> (as a dba) or the dba looking after the original poster will take this on.
>
> Evan
Received on Wed May 31 2000 - 00:00:00 CDT