On Mon, Apr 23, 2001 at 10:20:22PM +0200, Stefan Reich wrote:
> From: "Tavin Cole" <[EMAIL PROTECTED]>
> > No, this one's trivial to solve. We've talked about doing selective
> caching
> > in 0.4 or soon after. Nodes should incorporate a simple key closeness
> > judgment as well as a load balancing judgment in the selective cache test.
> > Then if you request all parts of a splitfile, your node will only cache
> > some of them.
>
> Selective caching has its drawbacks too... the data store will lose the
> convenient local caching function it has now. When I open the same file a
> second time, there is a chance I will have to wait for it to load through
> Freenet again - rather than pop up from the local data store instantly.
I'm becoming convinced that the clients should be responsible for caching
frequently accessed data, and doing it in a secure way. There's no perfect
solution here but I like the purism of making the node protect deniability
to the maximal extent when left to itself.
> Is this a convenience versus deniability question? It could actually become
> a runtime option.
Sort of.
--
# tavin cole
#
# "The process of scientific discovery is, in effect,
# a continual flight from wonder."
# - Albert Einstein
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
