On Thu, May 18, 2000 at 12:11:50AM +0200, Amaury JACQUOT wrote:
> Lee Daniel Crocker wrote:
> > Usenet is a perfect example of how to design a push medium
> > intended to bring down as many nodes on the network as possible
> > and destroy the value of the medium in the process.  Broadcast
> > Push media are inherently flawed: information only has value when
> > people want it.  Storage and time wasted on information that no
> > one wants (like over 90% of Usenet) is better spent making the
> > pull mechanisms more effective.  The only "push" medium that
> > works well is targeted push, i.e., e-mail (with mechanisms to
> > volutarily subscribe to limited broadcast groups--i.e., mailing
> > lists).  And even that is abused by spammers to the extent of
> > millions of dollars of wasted bandwidth.
> 
> This is completely wrong :
> Usenet is not push, but pull (you go grab/mirror newsgroups 
> from another server). Thus it's not push, but human routed pull
> (humans are supposed to route smarter (if they have a newsgroup from 
> one server, they don't grab the same group from another server).

In what I quickly proposed, nodes would send netnews by a push method
but clients would download news with a pull method.  The reason for
have nodes use push instead of the normal pull would be due to the
fact that not all nodes would have netnews over Freenet support.
Those that did not support netnews over Freenet (either because the
server itself doesn't know about netnews over Freenet or that the
system administrator has turned off support for netnews over Freenet -
giving the option is a good thing because netnews can easily swamp
machines which don't have large quantites of hard drive space
available) would just pass on the push messages without interpreting them.

Of course, pull would be favorable - provided that nodes aren't just
smart enough to pass on unsupported messages, but to also pass back
replies to unsupported messages. In that case, I would have to support
all pull instead of node-node push and client-node pull.  However, I
don't know whether they have support for this, and if they don't,
server-server communication will have to be push.

-- 
Travis Bemann
Sendmail is still screwed up on my box.
My email address is really bemann at execpc.com.

_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to