On Sat, Aug 01, 2020 at 10:18:00PM +1200, martin f krafft wrote: > Regarding the following, written by "tomas@tuxteam.de" on 2020-08-01 at 09:39 Uhr +0200: > >Hm. For Tox I can't say very much (besides that they do have a > >Client and a Core components, which seems to suggest that you need > >some well-known instance out there where the cliens do a > >rendez-vous. > > > >Jami uses SIP, and that implies there is some "SIP routing > >machinery" (i.e. at least one well-known address) out there to > >coordinate [1]. > > Both Tox and Jami use a technology called distributed hash tables to > map names to numbers, and the technology would be suitable for the > OP's needs, but I don't know of any implementation for SSH. DHTs are nice and cool, but they don't solve the problem of "where is that box I want to talk to", i.e. its IP address. What they /do/ is to keep this "resolution table" somewhere out there (think DNS, but not hierarchical, rather "randomly" shuffled). When "your" client goes out there, it has to tell "someone" "hi, I'm Sue and my IP address is currently so-and-so". To be able to tell "someone", you need someone's IP address. This someone is most probably one of the nodes taking part in the DHT party. So "... to connect to a remote home-machine (dynamic IP) through SSH session but without using a third-party server (free or paid)...", as the OP put it still seems a tall order. The only thing built into IP (v4 or V6) is multicast. There, it's the IP router's job to forward the information "hi, this guy/gal back there is interested in that multicast group", but my impression is (I'd love to be corrected on that) that consumer-grade ISPs don't want people to even know this exists (let alone offer that without billing for an arm and a kidney). Cheers - t
Attachment:
signature.asc
Description: Digital signature