At 9:44 am +0100 2005/01/10, David wrote: >Hi !!
>
>>First of all in this case the order does not matter (conditions in
>>a deny are ANDed, and AND is commutative); second these conditions
>>have no side-effects, so you can place them as you like.
>
>well, the result is the same, but order could matter, i would expect
>that the first condition is evaluated first and that evaluation stops
>as soon as a condition fails, so i always put the more time consuming
>or cpu intensive conditions at the end and the fasters ones first. This
>is not the case with this acl, but when you mix db lookups, dns lookups
>and simple comditions and *if* exim works as i expect, is always
>better to put simple conditions first and lookups and other expensive
>conditions at the end
>
you are absolutely correct! I was thinking of the result, but I
indeed do the same as you do, by putting the most expensive tests
last. My statement was very badly worded reading it again...
