Subj : NNCP
To   : Oli
From : deon
Date : Thu Oct 06 2022 08:53 am

  Re: NNCP
  By: Oli to deon on Wed Oct 05 2022 01:40 pm

Howdy,

 >  d> I dont think so. A packet today is addressed to a node, so you cant send the same packet to multiple nodes :( A change in
 >  d> packet format could
 >  d> allow that though...
 >
 > That would be a trivial change in the FTN packet format. But I guess it's simpler to replace only the binkp transport with NNCP.
 > And we also could use Netnews format (RFC 5536) with NNCP.

So while I'd happily write a new format - I could do it in php, since that is the language I know best, and I can already read current packets with it. I could probably fumble my way to get it into go, and possibly in javascript for synchronet.

Writing a packet conversion tool would be fairly easy, but no good if the receiver cannot convert it back to a packet. (I guess a utility in go could be the strategy, since you can complile it for multiple targets.)

 > Maybe. I don't have any experience with NNCP and only read the documentation on the website and some blog posts. NNCP with
 > Yggdrasil might be a good replacement for binkp over Tor. I did compile the latest version on the Raspberry. So maybe we could
 > test it with some file transfers.

I havent played with Yggdrasil, but interested in looking at it too..

 > I wonder why the git repository is down. Can we trust the tar ball from Russia with all dependencies included or might it be
 > compromised? ;)

Perhaps if you compared it to the debian src tar that would leave some confidence in the code? I'm assuming all the dependencies are public utilities?

So I see a few challenges with using nncp as a transport:
* It doesnt negate the usage of binkp (or emsi) - those utilities enable anonymous transfers (to support direct/crash netmails, where a link definition doesnt exist). With NNTP you need the remotes public keys in your configuration to initiate a transfer. (So that is OK for known netmail/echomail transfers.)

I dont see that as a bottleneck per se, since it forces routing of netmails, but if there is a broken link in the chain, you dont know that your netmail was delivered.

* There doesnt seem to be a way to trigger an event after sending with nncp-file. That could be implemented with nncp-exec, optionally with the contents being piped in via STDIN, however, I see that as being problematic. (The exec command, to toss the contents, would be implementation specific.) Alterantively, I regular cron could look for stuff in the inbound and process it.

* I like the idea of the multicast transfer, but a new fido packet technique would be required.

All in all, if there were several folks keen to play with this, then the effort may be worth it.



....лоеп
--- SBBSecho 3.15-Linux
 * Origin: I'm playing with ANSI+videotex - wanna play too? (21:2/116)

.