RE: Persistent Search - Question on A.J.Martin

[jukka.aakula@nokia.com]

Hi Alex,

You wrote:

> There is a transistion log and when a change is made to an entry 
> it is checked against the requirements of each client. 
...
> I my case I want to have a 
> distributed hierarchy of sites across Europe,  and I want to maintain 
> caches...the alternative to psearches would be a polling loop.
>  

I suppose you have say n sites which are all making one Persistent Search.
Thus going through the set of Persistent Searches ( per modification )
is O(n) and n is propably not too big.

But if each site would do say a set of persistent searches with size
m going through all the searches would mean O(nm). 

Why then would there be need to have many persistent searches ?
At least if we don't know the subset which we want to cache in advance.
Like in DNS.

cheers Jukka
> --------------------------------------------------------------
> ----------------
> |                                                             
>                |
> |  Dr. Alex Martin                                            
>                |
> |  e-Mail:   a.j.martin@qmw.ac.uk         Queen Mary, 
> University of London,  |
> |  Phone :   +44-(0)20-7882-5033          Mile End Road,      
>                |
> |  Fax   :   +44-(0)20-8981-9465          London, UK   E1 4NS 
>                |
> |                                                             
>                |
> --------------------------------------------------------------
> ----------------
>