p2p monticello - any takers?
Cees de Groot
cg at cdegroot.com
Thu Feb 24 00:13:39 UTC 2005
Ok, the P2P chat thing works, a nice UI is left as an exercise for the
interested reader, on to a next proof-of-concept.
Without doing formal coverage analysis, my gut feeling is that the P2P
chat application hits most of the P2P framework. It looks like it works,
so it's time to move to more interesting stuff that builds on this.
So, the next project is going to be hard. Really hard. Well... compared
with the stuff I cobbled up so far. I've declared myself too stupid to do
it all by myself (or too lazy, you choose) so I need your help.
The idea is to write an MCRepository backend on top of this. You create a
MCP2PRepository, point to a (shadow/cache) file storage area, and from
that point on everything you put there is accessible by everyone else who
has added an MCP2PRepository to his/her image.
'accessible' is one thing, but managing flood searches, ensuring timely
propagation, caching on intermediate nodes(?), reliable transfer of larger
datagrams over the network (it *should* work in theory, right now...), a
workable interface that accepts that opening such a repository will
involve a lot of asynchronous updates, ...
You see, enough juicy stuff to sink your teeth into :)
I'm off to bed now. If you're interested, let it sink in, and start
harassing me with questions on the current architecture if you need to
have more info to solve these little puzzles.
(did I mention authentication etcetera?)
G'night,
Cees
More information about the Squeak-dev
mailing list
|