&zzz
+R4SAS
+RN_
+T3s|4
+eche|off
+nilbog
+orignal
+postman
+qend-irc2p
+sourceress
Arch
Birdy
Danny
Irc2PGuest30010
Irc2PGuest36077
Irc2PGuest49364
Irc2PGuest51117
Irc2PGuest6564
Irc2PGuest65656
Irc2PGuest67278
Irc2PGuest74235
Irc2PGuest83482
MatrixBot
Onn4l7h
Over
Sleepy
Teeed
Yotsu
__bob_
aargh3
ac9f
acetone_
ahiru
anontor
b3t4f4c3__
cims
dr4wd3_
dr|z3d
duanin2
hababam
hagen_
leopold
makoto
marek
marek22k
n2
noidea
not_bob_afk2
nyaa2pguy
o3d3_
poriori
profetikla
r00tobo
rapidash
solidx66
stormycloud[m]
test7363673
uop23ip
urist_
user_
w8rabbit
zelgomer
zzz
ok orignal give me a couple more days, I've been distracted
orignal
fine
orignal
zzz, I have a question
orignal
about ratchets
orignal
say your server is crypto type 6
orignal
and client comes with type 5
orignal
are you able to obtain client's static key in this case?
dr|z3d
server is 6, client is 5 -> unsupported crypto.
orignal
the question is why
orignal
no real reason to not support it
dr|z3d
sure, it would be nice if we could support all 3 strengths, but we don't right now. there's probably a reason for that.
orignal
my point is
dr|z3d
but...
orignal
if a client supports PQ, PQ type should be defined by server
dr|z3d
that's probably better implemented on the client, not the server.
orignal
I'm talking about client only
dr|z3d
so the client just specifies pq (5,6,7) and uses whatever the server provisions.
orignal
only concern is how a server would react to such leaseset
orignal
clints specifies any of 4,5,6 meaning it supports all of them
orignal
server must specify exact type
dr|z3d
in java land, you can only have 2, and one of them must be ECIES if also using PQ.
dr|z3d
unless there's something I've overlooked.
dr|z3d
right, so we ideally want to support all pq standard instead of a specific strength, on the client.
dr|z3d
then client determines what strength of pq it needs, and switches automatically.
orignal
you can't have 0,4? lol
dr|z3d
read what I wrote.
orignal
client can't decide it's based on what server supprts
orignal
a server doesn't support more than one PQ type for the reason
dr|z3d
and your point is?
orignal
because you will have to try all 3 for every fucking incoming packet
orignal
my point is
orignal
if a client is 4,6 it can connect to server with 5
orignal
using 5 ofc
dr|z3d
we've already established we want multi-strenth support _on the client_ not the server.
orignal
my question is how would a server react
dr|z3d
client on 5, server on 6, server says "nope, not supported, can we fallback to ECIES?"
dr|z3d
the fix is to have the client establish what the server supports and switch.
zzz
if server is 6,4 and gets 5, it tries 6 and 4 and fails (ofc)
orignal
zzz, it's not about session
orignal
it's about LS only
orignal
client connect to server with 6
orignal
but it's LS is 5
orignal
e.g. it's just a verification
orignal
that static key matches
zzz
what are you trying to verify?
zzz
you can only verify static key by going thru the IK handshake
orignal
then a client sends their LS
orignal
do you verify static key?
zzz
yes
orignal
so if a session was created as 6 but the key is in 5 wehat would you do?
zzz
fail, because we look for a key of the correct type
orignal
why can't you take any PQ type?
zzz
to prevent the confusion you're proposing?
orignal
think about NTCP2
orignal
if Alice supports PQ it's based on Bob's type
orignal
or you are also going to verify against Alice's RI?
zzz
my principles when writing prop 169: 1) KISS; 2) a consensus on which PQ type we actually use will quickly form; 3) be flexible in case we were wrong on 2)
orignal
if Alice v=3 connects to Bob with v=4 with PQ what would you do after servering SessionConformed?
zzz
I think that would work? not sure
orignal
would you check version with Alice's RI?
orignal
from SessionConfrimed
orignal
your handshake is 4 but Alice is 3
zzz
looking...
zzz
haven't added any code like that yet, and hadn't thought about it
orignal
then what's diference with ratchets?
zzz
PQ ratchet code is done. PQ transports code is not.
orignal
you care only that a session is eatablished and static key is published in LS
orignal
it can be improved
orignal
I'm going to implement it in i2pd
zzz
I'm focused on PQ transports right now, I have no brainpower left for ratchet changes. The way to convince others to do something is to write a proposal