-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
>
> Well, updates get propagated in more or less the same way that inserts do.
> If we accept the premise that performing an insert to only 10-15 nodes is
> sufficient to propagate data (in conjunction with the spreading performed
> by requests), then why shouldn't we accept the premise that updating 10-15
> nodes is sufficient as well? (in conjunction with the spreading performed
> by follow-through requests)
This was the point that I so horribly failed to put in understandable
language. :)
>
> Perhaps the difference is that you want data to be updated everywhere as
> quickly as possible to minimize the number of out-of-date copies. However,
> no mechanism can provide even approximate guarantees that you will not get
> an out-of-date copy. Faced with that reality, it seems more in the Freenet
> "style" to allow updates to diffuse naturally with requests (Oskar's
> proposal) than to try to force the updates out everywhere (your proposal).
> Although updates may propagate a little more slowly, I don't think it will
> make much of a difference, and Oskar's proposal is certainly more
> consistent with the rest of Freenet, and uses fewer messages.
Yes, but if you were going to get the original document, then you're
almost guaranteed to get an up-to-date one.
The instance when you wouldn't is that a document was popular for a while,
causing it to propogate far into freenet (via requests). Then it falls
into disuse, so the update doesn't propogate as far. This is only an
issue if a *very* small number of people (like 1 or 2) are requesting the
data after an update. If a sufficient amount spread over freenet (this
means only a few people, say 4 or 5 widely distributed) are requesting,
then the update propogates as well as the original did.
Scott
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.1 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iD8DBQE5GakBpXyM95IyRhURAnJ8AJ9xmo1uLO0qXwUcKk4Orw2gN3I1HQCgpP01
K0q/sKZbiRty8xt1c1vpIxQ=
=eSB7
-----END PGP SIGNATURE-----
_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
