On May 1, 2009, at 9:35 AM, Matthew Toseland wrote:

> We could halve this again but it would mean that the
> average failing request is diverted to nodes which might have the  
> data but
> don't 2.5 times (24 hops * 20 peers = 480 bloom checks, vs 0.005%  
> false
> positives); each time the request has to wait.


Ok. I get it! That is why you want it only on a failing request...

Rather than only checking 24 nodes for the data, we will effectively  
be checking 480 (or a higher multiple w/o the 20 peer limit).... !!!

--
Robert Hailey

-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20090501/28fc21d2/attachment.html>

Reply via email to