At 03:31 PM 6/1/00 -0700, Charlie Caro wrote:

>
>
>Ann Harrison wrote:
>>
>>
>> And perhaps do something clever like actually reserving the rows
>> for update?
>>
>
>If the people wearing SQL decoder rings agree?
>

I don't think reserving the rows is a good idea. If I remember
correctly, proper SQL behavior from the decoder-ring set requires
the FOR UPDATE if there is any possibility for an update. InterBase
used to ignore this completely. Now, apparently, it's an explicit
de-optimization. But if somebody runs a program designed against
a strictly standards compliant dbms against InterBase, reserving
rows (forcing a database change unless somebody is doing something
very clever in SuperServer world) is going to run like a dog, maybe
even like a Basset Hound or a punt-a-pup.

Let's give the right answer and go fast. If somebody wants predictable
results with the unpredicatable results feature turned on, lets point
them to Oracle.

Jim Starkey