Cursor size and memory/speed

[Keith DeLong]

Running a number of experiments on 50,000 record sets, Ruslan confirms my
suspicions as to how Valentina obtains and manages the cursor.

After thinking about my particular problem for some months, it is good to
hear Ruslan say these types of algorithms are much deceptively hard!

Thanks for the discussion Guys,

Keith

> On 2/18/05 3:13 PM, "Robert Brenstein" <rjb at robelko.com> wrote:
> 
>> What about sql option to return only n top records?
> 
> If looks deeply, this option really speed up thins or optimise something in
> very rare cases. In most case you still must build full joins, then do
> sorting of all records and so on.
> 
> To really get speed up it needs to write quite special algs for major
> operations. Not easy task.
> 
> In 2.0 we support syntax, and probably semantic of this,
> Just I think it will not give speed up, just correct result.
>