XSF Discussion - 2018-02-03

Tobias, how/where did you learn about those two I-Ds?

The Twitter realms

Will see if I get around reading them on the 🚆

Tobias, kk :)

Failed outbound s2s EXTERNAL authentication zinid.ru -> jabber.org (208.68.163.218): Authentication failed: Peer provided no SASL mechanisms; most likely it doesn't accept our certificate

does anybody know when SASL EXTERNAL will be fixed on jabber.org?

zinid, yeah...DNS will be fixed.

zinid, currently a lot people are busy doing FOSDEM things though

I don't think this is DNS problem

this is a problem in certificate verification

oh

jabber.org's server doesn't accept LE certificates for example

Flow, https://tools.ietf.org/html/draft-omara-mls-architecture-00#section-2.2 sounds tricky to do in a federated world, unless the "Authentication Service" can federate with others

zinid, i'm using LE on my server and it can talk to jdev, so it supports them but there seem to be issues

Tobias, that's because it authenticate your server via dialback

could be

but not every server is running dialback module, and more servers will stop using it in the future ✏

Tobias: so if the authentication service is trusted this is not e2e and thus not like omemo/otr, right?

that's the question

Tobias, not sure if it is tricky in a federated system, but if you don't have a central authentication authority and can't cryptographically bind your identity with your key, your only option is probably something like fingerprint verification and/or TOFU

right, in the end there are still people who want end to end verification when they meet

"By definition, the AS is invested with a large amount of trust. A malicious AS can impersonate - or allow an attacker to impersonate - any user of the system. This risk can be mitigated by publishing the binding between identities and keys in a public log such as Key Transparency (KT) [KeyTransparency]."

Flow, they just bolt heavy KT on it and be done with it

wonder how that scales, these KT logs become quite large

Tobias, have you found the corresponding mailing list discussion already?

there doesn't ssem to be one, currently it's just a personal draft

Yeah, I can't find anything either.

Flow, also https://tools.ietf.org/html/draft-omara-mls-architecture-00#section-3.1.6 :) "should be able to interoperate"

at least not a MAY :)

Okay, from what I can find they don't even state how this is supposed to work so one can check.

And the reference to the protocol has no link. :p

Federated authentication is tricky

Is that this: https://medium.com/netflix-techblog/message-security-layer-a-modern-take-on-securing-communication-f16964b79642 ?

Tobias, but for proper e2e you need e2e verification anyway...

You can get away without it if you have decent ratcheting, I suppose.

I did a federated authentication thing for iot. But there I hard linked the crypto to the network layer (IPv6 address)

That's not directly possible with free choice DNS names so we would need a different system for XMPP

"hard linked"?

So if a device is compromised you throw it away because the key cannot be changed?

(Seems to be one of the few cases where this might not be problem.)

If you change the key you change the address

Tobias, I axed: https://mailarchive.ietf.org/arch/msg/cfrg/KN9-Dy26tBGMx4YkzY98ZDkpRTg

I linked to proto earlier. See history

https://datatracker.ietf.org/doc/draft-barnes-mls-protocol/

Oh, cool, I'll add that!

Don't know how future proof that is. It's not mentioning block chain at all

-_-

Tobias, https://twitter.com/paul_pearce/status/959138611617673216

😀

AES CBC is literally a block chain.

!

Is "nano" still a cool word?

Zash: sorry, nope

"nanoblockchain"

"microblockchain" then?

pico?

"nano" was a thing when I wen to school. Which is now almost 20 years ago

I was in school 20 years ago too

Zash: I propose "eco" or "bio" because it comes without PoW

Yeah, enough time has passed to move to pico now

PicoBlockChain - PBC

Ge0rG: Eco has been around for too long too

I haven't head of Bio that much in tech however… XEP-xxxx: Bio Chains

Hyperblockchains?

nano is a cool word, I use it all the time

Something something hypercube routing

Noob question, is there a way to never send the plain passwd to the server and only do scram, from account creation to the end. (Thinking about ibr)

Hyper cube, wasn't that the movie full of deadly traps?

pep.: you need to send the password during ibr

pep., no

Wasn't there some draft/protoxep on uploading SCRAM stuff?

oh?

Bunneh: xep scram

Zash: Sorry, I couldn't find a match

Zash: would be nice

Also, scram for the web when

Hah, good luck

Re the talk of device / client registration that was mentioned on the summit

In't that something like the model of OAuth, where instead of the user signing into their account, they authorize the clients to access their account.

In that model, password login don't as much sense

I want an Auth mechanism called SCUMM.

What like oauth with Facebook?