From: "Brandon" <[EMAIL PROTECTED]>
>
> > What's the current consensus about requesting data with a HTL of 1, as
> > these people plan to do (as I understand it)?
>
> Yep, HTL 1 is quite a problem. I highly recommend changing HTL decay to be
> probabalistic instead of arithmetic so that HTL of 1 still has some chance
> of moving data onto the node.
>
I read this on the web site quite a while ago, and it seemed to make sense:
Instead of having an HTL counter that is decreased every time the request is
relayed, randomly or not, attach to each query a probability-of-spreading
x -- whenever a node gets a request, it forwards it to other nodes with x %
probability. This probability need not be modified along the way, and it's
simple to calculate the expected number number of hops (and the standard
deviation or some such) that a request will survive for a given x.
The advantage of this is that when there is no HTL counter to decrement, the
request can be passed on in the exact same form it was received, eliminating
the HTL 1 problem. There would be no way to see whether a certain node was
originating or just relaying the request.
And no, you couldn't kill Freenet by making requests with a
probability-of-spreading of 100% -- naturally, nodes can and will change
this number if it is too high, just like they can change a ridiculously high
HTL (they _can_ do this, right?).
// Kalle
_______________________________________________
Freenet-chat mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/freenet-chat
