Re: [Exist-open] How to increase performances

[Wolfgang M. <wol...@gm...>]

Hi,

> No change on the fn:count(//LineItem).
> Next, the fn:count(//LineItem[./@ItemNumber]) takes 1300ms avg.
> Next, the fn:count(//LineItem[./@ItemNumber > 0]) takes 2500ms avg.
>
> On a Pentium 2.8 GHz with 1GB of RAM.
>
> Why a simple count takes this "long" time ?

I'm currently doing some similar tests with a 100,000 docs collection
to see if the range index works correct and where we can improve
performance for more data-centric queries. Overall, range index
lookups and comparisons seem to be fast enough now - the time is spent
somewhere else. I planned to continue testing over the weekend. I have
a few ideas where things could be improved...

Wolfgang