orignal
weko has discovered something interesting
orignal
NTCP2 sessions got didconnected quiclky with his mobile operator
orignal
while SSU2-only sessions stable
zzz
is he in a censorship country?
R4SAS
some operators just blocks torrent traffic
R4SAS
so they think that NTCP2 is bittorrent )))
orignal
in Russia
orignal
it's not about censorship
orignal
it's about p2p traffic I guess
orignal
so my patricular question is if SSU2 should have priority over NTCP2 for mobile
zzz
what we do is if there's a very high failure rate on one transport we prefer the other one
orignal
good to know
dr|z3d
what about this RI/dest compression, orignal? are you contemplating getting that into i2pd? see: git.idk.i2p/i2p-hackers/i2p.i2p/-/merge_requests/66
orignal
what's this?
dr|z3d
zzz said you're not sold on the idea..
dr|z3d
would be good to get it into the next release.. zzz's contemplating a rekeying event, so maybe it would be good to make that happen around the same time SSU2 shift happens..
zzz
dr|z3d, I'll bring it up when I'm ready, keep your pants on
dr|z3d
orignal's all about performance, so it seems like a win :)
orignal
I don't use a compressor
orignal
usually I use mmecpy only ))
dr|z3d
I don't think we're talking about a gzip event here :)
orignal
then what's that?
dr|z3d
the gzip you're referring to is related to on disk storage, that's a secondary benefit. the main thrust is this: Replaces the 256-byte ElG key in dests with padding.
dr|z3d
Make all padding in dests and router identities be a repeating random 32-byte pattern.
orignal
yes I know what you are talking about
zzz
0) Hi
zzz
hi
orignal
hi
zzz
what's on the agenda for today?
orignal
not too much and I need to run to another meeting at 3:30
eyedeekay
fuck there's a dog attacking my cat outside, I'll be late for the meeting
zzz
lol
zzz
I'll add a brief 1) SSU2 blog post update
dr|z3d
not funny, unless eyedeekay's been on the acid again :)
orignal
yes please
zzz
we'll put go-i2p on as 2) if the cat lives
zzz
anything else?
dr|z3d
:)
orignal
that Identity compression dr|z3d meantioned before
zzz
ok
zzz
1) SSU2 blog post update
zzz
and I've been getting some reviews and comments on my forum
zzz
need to add something about performance
eyedeekay
OK I'm back
eyedeekay
He's fine
orignal
you have missed Chacha2 crypto from the list
dr|z3d
it's good, readable and pitched at a level mere mortals can understand, zzz.
zzz
ok thanks for the feedback
zzz
chacha is in there somewhere
zzz
All I2P Noise protocols use the following standard cryptographic algorithms:
zzz
X25519
zzz
ChaCha20/Poly1305 AEAD
zzz
SHA-256
zzz
should I put it somewhere else?
orignal
Just Chacha20
orignal
not chacha/poly
zzz
oh, ok
zzz
will do
orignal
that's my only note
orignal
everything else looks good
zzz
will keep banging away at it, hope to be done by end of the month, but feel free to steal whatever you or acetone want
zzz
anything else on 1) ?
orignal
acetone is MIA
orignal
you know why, ritgh ? ))
zzz
no, don't know him well at all
zzz
2) go-i2p
orignal
he has got released from prison redcently
orignal
hence he doesn't have plenty of free time anymore
zzz
better than still in
zzz
eyedeekay, what's the latest on go?
zzz
30 sec and we'll skip to 3)
eyedeekay
re: go-i2p transports: I fixed the thing with the creating sessions on different ports and have been able start writing real connection tests, which I actually just started checking in along with the noise transport itself.
eyedeekay
I would say it doesn't work right, but it's getting much closer
zzz
oh great
eyedeekay
also got to give it up to thulium, who has become a consistent contributor and I really like where his head is at about this.
zzz
I saw the mapping fix patch
orignal
zzz my point is that I might take care about SSU2 article on habr
eyedeekay
Super helpful to have him around, many, many thanks to him
eyedeekay
MVP of the week
eyedeekay
EOT
zzz
re: habr, yea I figured, maybe what I wrote will help you, maybe not
zzz
you were assuming all integers were 8 bytes or something...
orignal
definitly it will
eyedeekay
Not on purpose, but yes I was
zzz
ok, anything else on 2) ?
eyedeekay
No
orignal
no
zzz
go ahead dr|z3d
orignal
yes, what's your pruposal and what's a goal?
dr|z3d
well, it's really yours, zzz, but I'll say a couple of words by way of introduction..
dr|z3d
the idea is to reduce the size of RIs and dests which should make for faster exchanges when they're used.
dr|z3d
I think you mentioned something about the major benefit being for streaming, zzz?
zzz
the MR is only WIP
orignal
let me clarify
orignal
originally we were trying to fit SessionConfirmed message
orignal
but then we have found that fragmentation is still rteqeuired in some cases
orignal
beside it I don't see any motivation to implemente it
zzz
I'm working on a proposal. I'll have more to say when I'm ready. I'm not prepared to do any discussion at this time.
zzz
it will be proposal #161
dr|z3d
can you just summarize in a single line what you think the main benefits will be, zzz?
zzz
the background is at zzz.i2p/topics/3294
zzz
It seems like you're pressuring me to immediately try to convince others
zzz
I'll state my case in the proposal
zzz
you also have misquoted me when you said I was targeting the next release. I explicitly told you that was not true in #i2p-dev the other day
dr|z3d
no pressure, but you obviously thought it was at a state where it could be included in I2P+, so a simple summary of your ideas wouldn't hurt.
zzz
if you want to lobby somebody about something, go ahead
dr|z3d
sorry, that wasn't my intention. I was hoping it _might_ be considered for next release in the event orignal was convinced to get on board, to sync with the SSU2 rollout, but that's all me. :)
zzz
I'll do it when I'm ready
orignal
it does't require a release
orignal
because no protocol changes
zzz
that's right, but for fingerprinting reasons it would be better to do the same thing
dr|z3d
sure, but it will flag your router as having adopted the changes, which is why zzz mentioned possible rekeying. and what he said ^ :)
zzz
anyway, I'll be leaving a big space in the proposal for a discussion of future changes and where we might be going, so I'll need help when I get there
orignal
rekeing?
orignal
why?
orignal
you only care about first 32 bytes
dr|z3d
to prevent fingerprinting, in essence.
dr|z3d
because new routers will have a compressed dest, old ones won't.
orignal
and we don't need rekeing for it
orignal
just use existsing keys
dr|z3d
then you have 2 classes of router on the network. if that's not a major issue, then fine.
zzz
I'll let you all know when I get a draft up. maybe a week, maybe more
zzz
anything else on 3) ?
orignal
no
eyedeekay
no
zzz
anything else for the meeting?
zzz
ok, thanks everybody, have a great week