Dmitry Kuzmenko wrote:

>
> Hello, Ann!
>
> Ann Harrison wrote:
>
> > > FILE_FLAG_OVERLAPPED |
> > > FILE_FLAG_NO_BUFFERING |
>
> > One of the characteristics of the development of InterBase
> > at Borland is that some directions were explored and abandoned,
> > leaving the code in place, but conditionalized out. SUPERSERVER_V2
> > was such an effort. The one consistent factor among these features
> > is that they never made it into the QA system. Some are more complete
> > than others. Expression indexes and scrollable cursors are two more
> > such features.
>
> I can't believe that eliminating unneccesary OS buffering will
> decrease performance. It is simple to look into TaskManager when
> select count(*) is made on a big table:
> While IB is reading pages into it's cache, they are also being read
> (before IB) into OS File Cache. So, availavle memory decreases, which
> can reduce OS performance (and IB's too). Isn't it?
>

I think the point you are trying to make is valid. It is quite evident that
InterBase and NT write every read back out to NT virtual memory.

However, what Ann has said should not be taken lightly. The code includes
unfinished attempts to implement a new improved SS architecture. This much we
know - it was publicly acknowledged that these attempts were being made for
v7.0. It is probably only Charlie Caro who knows just how much of the work was
completed, what the dependancies were and what was left to be done.

Still, I bet you can't resist seeing what happens if you add the define and do a
rebuild, and I would love to hear about if you do.


Paul
--

Paul Reeves
http://www.ibphoenix.com
taking InterBase further