Edgar Friendly wrote:
> As for changing this to try to take into account what the node
> percieves as its specialization, I will insist on the inherent
> wrongness of any such procedure and leave it at that.
I don't understand why this is inherently wrong. I presume the
"histogram of keys in the local datastore" available in the nodestatus
servlet represents the specialization of keys on the local node. Why
not use it?
> >When writing data to the cd it might be necessary to use some random
> >factor in addition to most recent date in order to get more generic
> >data onto it. That way there's more likely to data of interest to
> >other nodes on it.
>
> one could include the most popular data from your node, but BOTWFRW
> (because of the way freenet routing works), that data is unlikely to
> be requested of another random node.
Which is why I suggested adding some sort of randomness to it.
Jeff Teitel
_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl
