On 03-May-2000 Hans-Peter Zorn wrote:
> Jochen, DG6VJ, did talk about his ideas at the PR-Convention in
> Darmstadt last month. It is in the scriptum, but I think he can
> make it available in postscript (German, though). To summarize the
> ideas: The Network should allow cycles to avoid net-splits, which are
> very annoying especially in the EU converse network which consists of
> about 90-100 convers nodes, all of them being linked over _radio_
> links.
>
> To have a non-hierarchic structure, each message needs to get a serial
> number (seperatly for each host) and a ttl. To avoid unnecessary
> transmissions, the system should be able to (temporarily) suspend
> a link when it receives old data over this route. When a supplying
> link failes he can re-activate the "sleeping" renundant link.
>
> We want to use serial numbers, not hashes or something, because we
> need low memory consumption. Gunter of FlexNet is interested in having
> this system in his new rmnc master, which will not have a harddisk or
> virtual memory.
>
> Another question is how to route personal (/m call) messages through a
> network of this style.
>
> We don't have a protocol specification and no working code yet.
>
> Other details are IRC-Like descriptive aplhanumeric channel-names and
> perhaps some kind of user-authentication.
All this sounds horribly like my new DXCluster protocol which is also
multicastable (an important requirement in my view).
I am currently working on exactly such a thing and there has been a working
copy of a sort of specification put out for discussion in the past which I am
now implementing (in perl as it happens - but don't let that put you off).
I will tart it up and stick it on the web site sometime this week.
Dirk G1TLH
http://www.dxcluster.org
--
Dirk-Jan Koopman, Tobit Computer Co Ltd
At the source of every error which is blamed on the computer you will find
at least two human errors, including the error of blaming it on the computer.
