~dr|z3d
@RN
@StormyCloud
@T3s|4
@T3s|4_
@eyedeekay
@orignal
@postman
@zzz
%acetone
%cumlord
%snex
+FreefallHeavens
+RTP
+Xeha
+ardu
+bak83
+onon_
+poriori
+profetikla
+r00tobo_BNC
+uop23ip
+weko
Arch
Danny
DeltaOreo
FreeB
Irc2PGuest13718
Irc2PGuest19074
Irc2PGuest36180
Irc2PGuest40608
Irc2PGuest41133
Irc2PGuest73814
Meow
Onn4l7h
Onn4|7h
idontpee
maylay
not_bob_afk
qend-irc2p
radakayot
shiver_
simprelay
solidx66
thetia
u5657
user__
username
woodwose
zer0bitz_
zzz
I now have 6 of my 7 MRs merged, after that will turn to littler stuff like this
dr|z3d
testing, testing, are you receiving, snex, over...?
snex
yes i receive
dr|z3d
excellent.
snex
the lyrics i posted when it happened both times were a famous song by The Animals
snex
its definitely those lyrics
snex
postman: please check your shadowban rules for that
fox
it lives
fox
oof 31 names
snex
?
fox
*** passes the rig to RN ***
fox
*** wanders off to find psi and his cat ***
snex
fox i will bet you 100 internet points that you cant sing the first verse to "house of the rising sun" by the animals
fox
THERE ONCE, WAS A MAN, FROM NEW ORLEANS
fox
so did putin push all the russians out windows or are they still here?
fox
*** need to add slow's index if it still exists ***
dr|z3d
*** winks at snex ***
fox
dr|z3d: has i2p had any audits recently? also any talk of switching to pqc?
zzz
changelog MR is up, so eyedeekay doesn't run out of things to do
eyedeekay
Me? Running out of things to do? never.
dr|z3d
yes, fox
dr|z3d
zzz and orignal are working on post-quantum.
zzz
Blinded message
dr|z3d
and a partridge in a pear tree ...
dr|z3d
:)
eyedeekay
Acked changelog MR, likely to same for the logs page MR
zzz
ok those are small beans though, the last one I need soon is the BW params
eyedeekay
Yup will respond by the end of the day, that and the HTTP Server threading one required the most serious thinking
fox
kyber?
fox
or maybe they picked something more original
fox
*** will see himself out ***
fox
so almost kyber
fox
lots of metions of sha256 lol atleast it's not md5
orignal
md5?
orignal
what are you talking about?
zzz
well, that's another question. We could move to blake or sha3, but if sha256 is broken all of i2p is broken
orignal
never heard that sha256 is broken
zzz
if we do want to switch, now is the time
orignal
we should switch to one suppoerted by sha-ni
zzz
MLKEM requires SHA3-256 and -512 already, so I'd prefer one of those over blake
zzz
might take a while to make a decision on things like this
orignal
it's inside it
orignal
you don't have to call it explicitly
orignal
yes sha3 then
zzz
so I'd prefer you didn't include MLKEM in your next release, or at least hide it behind some sort of test config
orignal
no I'm not going to
orignal
people can play with type 5 but it's not offcial
zzz
so sha-ni does sha3?
orignal
not sure
zzz
to be clear, what I think fox is talking about is the hash function for noise mixHash(). If we change it, it would change the Noise initializer strings
orignal
it's hmac if I remeber
zzz
right, the hash function for the hmac
orignal
yes it's sha256
orignal
not problem for me to implement new MixHash
zzz
with a noise lib, you specify what hash function when you set it up, so it's pretty easy to change
orignal
for me any hash is fine as long as it's in openssl
zzz
yup. Just have to do research and then decide
zzz
sounds like fox probably has an opinion
fox
imho switch it all now with the pqc upgrade
fox
sha256 doesn't have any major flaw im aware of but in these times of orwellian governments willing to dump billions into AI and presumably quantum computers something like sha3 would have a better safety margin
zzz
safety margin reference?
orignal
quantum computers have nohing to do with hashes
fox
grover's algorithm
zzz
funny thing, your friend psi used to bug us a lot about switching hash algos
orignal
you mean Jeff?
fox
i haven't been able to get ahold of jeff in years
orignal
are you that fox from Toronto?
fox
think he got a corp job and turned republican somehow
fox
lol no but im somewhat familiar with ca
zzz
he's still alive and kicking on github
orignal
zzz, could you refresh my memory who is this guy?
zzz
dunno
fox
yeah i saw github and iirc yggdrasil in i2pd is his baby
orignal
but you know he is in touch with psi
orignal
what?
fox
*** is an inigma ***
orignal
how is he related to ygg in i2pd?
orignal
this was acetone's idea
fox
*** throws an ignious rock at RN ***
zzz
"* fox wanders off to find psi and his cat"
orignal
lol
fox
i could totally be wrong. i remember psi had some project with a norse name
fox
zzz: yeah never found him or his cat. i could email him but bleh
zzz
lokinet
orignal
lohinet ))
fox
oh yah that's it
fox
come to think of it we used to have a double phd student here trying to make some kind of overlay network in elixir
orignal
for those who doesn't know what lohi mean
orignal
lohi - лохи
orignal
like loosers/victims
fox
*** has returned in expectation of a migration as the great firewall of... probably palantir or oracle goes up ***
orignal
knijka.i2p is all about them))
fox
that wall of russian is long enough to encircle berlin
orignal
and not pnly berlin
fox
suppose i'll take it psi was killed and replaced by the thing. rip
orignal
no. he has grown up ))
fox
not sure up is the right adjective
zzz
so you have a link to a reference supporting your safety margin claim?
fox
grover's algorithm wikipedia page
zzz
ok
fox
tldr cut bit strength in half. doens't break 256 that i remember
fox
last time i talked to a math major about it my takeaway was quantum computers put some hash algorithms in range of other attacks in theory
zzz
that wiki page doesn't explicitly compare SHA3 and SHA2 in any measurable way
dr|z3d
tldr:
dr|z3d
> SHA-2 has proved to be secure over time (minus length extension attacks), and many think that it won't be broken in our lifetime. Thus SHA-3 isn't more secure than SHA-2 in the real world. It also doesn't help that the SHA-3 standards defined more than a dozen different functions and that everybody is confused about what to use and when.
dr|z3d
> Thus, there are virtually no reasons for developers to use SHA-3 instead of SHA-2 other than for key derivation where its poor performance doesn't matter, but it's "stronger" security does.
dr|z3d
> SHA-512 is slower than SHA-256 on both x86_64 and arm64 architecture, but with hardware acceleration, which is becoming ubiquitous, the gap is very small, and, as we've seen earlier, it's certainly "fast enough" for most use cases, as your bottleneck will be I/O. Furthermore, the BHT attack, hypothetically reducing the security of 256 bit hashes from 2128 to 285 may scare more than one checkbox ticker auditor
dr|z3d
in 2030, and you will have to justify using the (theoretically) "weaker" 256 bit version instead of the 512 bit version.
dr|z3d
> So, by eliminating all the other functions I think we have our winner: go for SHA-512.
mareki2p
In SHA-2 family of functions the SHA-256 suffers from length extension attack. SHA-512/256 does not suffer from this (and is faster on 64bit computers).
mareki2p
SHA-512 is "basically" extactly the same as SHA-256 except: It uses 64bit operations instead of 32bit operations, has 2x the size of input block, and 2x the size of output block, has 80 rounds instead of 64 rounds. So yes, SHA-512 is faster than SHA-256 on 64bit hardware.
mareki2p
Hardware acceleration changes that, as SHA-512 comes only with Zen5 and Lunar Lake and Arrow Lake.
fox
sha512 also seems like a perfectly reasonable choice if you prefer time tested
fox
any of you know how to set tunnel hops and ammount i2pd? im only seeing refrence to i2cp and options for exploratory in conf
zzz
that first link is a good one, thx drz
zzz
it would seem to point us to SHA-512
mareki2p
And BLAKE3 could be parallelized, each 1kB of data could be hashed independently resulting in multiple digests which are later combined.
zzz
99.9% of our hashing use cases, including ratchet, are over small data sizes
fox
does i2p still have many routers on 32 bit?
dr|z3d
*thumbs up* zzz
dr|z3d
you might as well ask the percentage of i2p routers running in computer cases colored beige, fox.
dr|z3d
we have no way of knowing.
fox
i had thought that was a tracked metric somehow
fox
*** visits stats.i2p ***