No - that contents.item line can even be removed completely - same
story - but no matter how 'wrong' the SQL is - you'd still expect the
same plan ?

--- In ib-support@y..., Svein Erling Tysvaer
<svein.erling.tysvaer@k...> wrote:

> Nope, what I thought of wouldn't work in this case - I could not

force the

> optimizer to choose the quick plan without explicitly specifying

the plan.

>
> But there was one thing that puzzled me in your query. Shouldn't it

be

> ...
> LEFT OUTER JOIN contents ON (waybill.waybill =
> contents.waybill and contents.item = 1)
> WHERE ...
>
> rather than having contents.item in the WHERE clause? No experience

myself,

> but I once read that the right part of a left outer join could only

be

> limited within the join itself, not in the WHERE clause. I doubt

this would

> be the source of your current problem, but it shouldn't take long

to test

> if you still have access to the slow database.
>
> Set
>
> At 08:23 02.10.2002 +0000, Rod Bracher wrote:
>