From: <[EMAIL PROTECTED]>
> This causes requests to fail, so people try to fix it
> by inserting onto more nodes, thereby making the problem worse.
Hmmm, so if 'multiple insertion points' would cause worse problems for the
network...
Maybe an idea might be to place occasional FNP requests to different random
nodes, as a way of 'sucking' new files through the network.
I note your point (Ian) about (did I hear you right?) network 'heuristics',
and the process whereby new nodes 'grow into' the network.
So then, much of the problem can be fixed in the client documentation (!) -
warning users that newer nodes, with less history of inserting and
requesting, enjoy lower success rates.
Lastly, I wonder if part of the problem could be a high ratio of transient
nodes to permanent ones.
If this is true, then there might be a case for implementing a 'dynamic
IP'-type system, whereby all hosts without static IPs/hostnames *must* be
addressible by a static name.
Thinking left field, the user can be asked to choose a name, then that name
is sent to inform.php with a "?name=myhostname" arg appended to the
inform.php req. This way, inform.php could also serve as a kind of 'dynamic
IP' sevice.
If one has a static IP or DNS hostname, that would be acceptable (since node
can be contacted by that name or IP).
But if user has no static IP or DNS hostname, then maybe that user should be
forced to choose a permanent name.
I know it's a bit irresponsible to be proposing new functionality for
inform.php, when that facility is to be phased out.
But hey, there's always going to be the need for out-of-band means for new
nodes to enter Freenet - when inform.php dies, such functionality could move
to its replacement.
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
