IRC channel logs
2023-06-14.log
back to list of logs
<Zarutian_iPad>on a credible runtime substrate like an blockchain you pretty much get integrity and repudation for nearly free <KrisKowal[m]>in any case, the on-chain isn’t terribly relevant. it’ll probably need off-chain help for 3ph <Zarutian_iPad>so markm has come up with ‘my name is how you know it is me’ that actually splits an vat identifier into two parts <Zarutian_iPad>how to speak or address the vat, and how to verify utterances of the vat to others <Zarutian_iPad>so a lighthouse or Ethereum event watcher or ither such bridge is needed betwixt on chain vats and many solo, qourum, and smpc vats anyway <Zarutian_iPad>what I want to point out re 3PH in CapTP of E is there is no namespace or noncespace that is vat global <Zarutian_iPad>basically each giver and gift recipiant have a name-/nonce-space between them and per direction of gifting <Zarutian_iPad>this means that you need only to authenticate each vat vice versa when establishing an integrity guranteed connection between them <Zarutian_iPad>so a 3PH where an on chain vat is the vat hosting the Remotable and the others are off chain but authenticated via their signitures eorks <Zarutian_iPad>same with any public key signed message, like aforesaid ActiveCapCerts <Zarutian_iPad>if the object hosting vat is an off chain vat and the giver is an on chain one then the hosting vat must know, via .provideFor() sent by the giving vat to hosting vat that the hosting vat should give the recipiant vat access to the gift. <Zarutian_iPad>KrisKowal: did you write Q-comms way back? something that worked over any port or other such interface? <Zarutian_iPad>so ocapn over tls+tcp, tor, websockets, webrtc, bittorrent uTP, (muxed) serial, and unix stream socket should work or be the goal, no? <Zarutian_iPad>heck stuff each op:deliver or such message plus some counter info + ed2519 signiture into an udp datagram would work-ish <KrisKowal[m]>And yeah, UDP with a sequence number is a thing that I’ve thought about, tho then you get to hold flow control, congestion avoidance, and retries in your own bare hands. <KrisKowal[m]>In a prior life, I got to work with a pod of CalTech queuing theory postdocs and my Dunning-Kruger has been cured. I do like to think of hand-rolling CoDel for funsies but as they say at NASA, any mission that involves a new vehicle is a new vehicle mission. <KrisKowal[m]>So, if we can get everything we need out of TLS, great. That’s decades of not reinventing the wheel I can get behind. <KrisKowal[m]>Self-signed x509s that are only relied upon for secrecy on the wire sounds good to me and plausibly absent over my unix domain socket, but I might have to convince some folks. <KrisKowal[m]>Brian Warner is more inclined to strip away TLS, though. He trusts libsodium much more than he trusts openssl. <KrisKowal[m]>And he has other thoughts about decoupling sessions and connections. I think that’s out of scope for ocapn tho. <Zarutian_iPad>decoupling sessions and connections can be done with a membrane <KrisKowal[m]>Also, for the love of all that is holy, I’m sure the sun will go down eventually in Iceland. Surely this is past your bedtime :-) <KrisKowal[m]>* pronunciation of Æ depends generally on whether the subject is academic or ecclesiastical Latin. I think the academic view matches the old English and Icelandic interpretation. <KrisKowal[m]>(Ecclesiastical Latin Æ is I think closer to German Ä, which Americans can’t do, so more like just E, but there’s so much evidence of confusion since Æ variously reduced to AI for airplane / aeroplane, ether for æther, iso- for æso-, etc, all of which are pronounced differently) <KrisKowal[m]>* / aeroplane, E in ether for, * for æther, I in iso- for <isd>We definitely should not invent our own transport protocol or try to muck with message-level crypto. http://noiseprotocol.org/ might be a good alternative to TLS; simple, cryptographers seem to like it, uses ECC, several available implementations.