IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#ls2
/2022/09/26
zzz 0) Hi
zzz what's on the agenda for today?
eyedeekay go-i2p as always
zzz 1) will be go-i2p
orignal what is next?
dr|z3d this has got a distinct smeel of dejavu about it :)
zzz 2) brief update on ssu2 blog post status
dr|z3d ready to share more about RI compression, zzz? :)
zzz "what is next" is a topic?
orignal probably
zzz ok 3) is whatis next
orignal what is out next project
zzz no, I have no update on RI compression. maybe next week.
zzz thats a good list anyway
zzz 1) go-i2p
eyedeekay OK got my weekly copypasta incoming
dr|z3d be prepared for the monsoon...
eyedeekay Started writing tests for noise transport and things started to get more real, for instance:
eyedeekay after reversing the too-many-sockets issue I realized I had left myself without a way to configure the actual underlying socket
eyedeekay like I couldn't tell it a port to use or what host to listen on, so it would pick a random local one every time and so I'd have to get the port from the socket structure itself
eyedeekay that kind of thing bubbled to the surface
eyedeekay then I had to spend time over into some contexts that aren't part of this meeting
eyedeekay thulium's willingness to second-guess my refactoring decisions continues to be valuable, he's finding things to simplify about twice a week days and running them by me, right now he's looking at a refactor of the Certificate common interface and some of the things that implement it
eyedeekay looking forward to seeing what comes of that
eyedeekay we're both working on branches right now with merges planned for "Soon," optimistically we should start them this week
eyedeekay Other news: Progressing on godoc fixes, making more decisions on when and where it's appropriate/helpful to depend on a library vs writing something ourselves
zzz have you started work on ntcp2 data phase, or you just have the handshake so far?
eyedeekay Just the handshake so far still
eyedeekay No data yet
zzz yeah that was my guess. Still, sounds like good progress
eyedeekay I figure it will be hardest when it's new and this is all still pretty new to me
eyedeekay When I switch from Noise to NTCP hopefully I work much faster because I've already done it once
zzz thanks for the update
zzz anything else on 1) ?
eyedeekay you're welcome, nothing else on 1 from me
orignal are you able to recieve something back?
orignal e.g. with an existsing router
orignal your connect then send SessionRequest
eyedeekay Not with an existing router, that I've tried, yet
orignal do you receive SessionCreated back?
orignal what was the problem?
zzz I believe he's still working with an unmodified noise lib, he has to go back and do the aes-encrypt of the keys and add the padding
eyedeekay But again this is pure-noise and not NTCP2 so I'm not sure it should work at all, I've been talking to myself only for a while
zzz remember we advised him to start with unmodified noise first and then add the mods so he didn't get too messed up
orignal I would suggest to start with connection with somebody else
orignal and see how it should lokke like in reality
zzz sounds like he'll be there soon
orignal idk what "unmodifuied noice" mean
eyedeekay I'm sort of diving into the deepest of the deep end here, I've got no network engineering experience that didn't come from homelabs and I2P so I think I ended up taking some in-between steps to help understand things better
zzz unmodified noise == using a noise library unchanged (no AES encrypt of keys or added padding)
eyedeekay Trying to do an I2P SessionRequest over Noise without modifying it to behave as NTCP2 using the spec is what I've been trying to do in as specific terms as possible
orignal is there a noise library in go?
eyedeekay github.com/flynn/noise
orignal but noise itslef is veru simple
orignal just MixHash and MixKey
orignal two function
eyedeekay And I've also been using the Go noiseSocket implementation as examples
orignal that's all you need
zzz sure, but a library is a reasonable choice also.
zzz anything else on 1) ?
eyedeekay It's what all the examples I read are based on, and I'd rather use a lib than not for now
eyedeekay Maybe that changes someday
eyedeekay But nothing else for 1 now
zzz 2) ssu2 blog post update
zzz I've done a few edits
zzz added a mention of chacha2 for the headers
zzz added a performance section
zzz I now plan to finish by late next week, after our Oct. 4 monthly meeting
zzz because I want to have a decision on what we call the next release first
zzz EOT
zzz any more comments, let me know, thanks
zzz anything else on 2) ?
eyedeekay Just a sec I want to pull up the post here
zzz still working on fixing all the links
orignal what's your plan for the next release?
orignal regarding SSU2
zzz enable it for everybody
zzz that's been the plan for at least 6 months, no change
orignal please remning about the next release
zzz what about it?
orignal I will disable SSU and ennable SSU2 for everybody
orignal SSU2 looks stable for me now
orignal I have another question for eyedeekay
zzz oh yeah, speaking of stable, I did fix my short-packet bug
eyedeekay I don't think I have much to suggest in the way of phrasing/organization/etc, links are important for the sections where it's compared to QUIC but you mentioned tat
orignal when you try to use libi2pd in your go code, did it even work?
orignal because it seems a lot of missing initializations there
eyedeekay Yes it did work, but you have to construct the configuration and pass it in from Go, the example code in the go-i2pd repository should work
orignal did it connect with other routers?
eyedeekay Yeah I used it to build an application
eyedeekay It made a configuration directory, started a SAM bridge and everything
orignal because some guy pointed me that it kept crashing
orignal thanks
eyedeekay Hm. Well I don't know offhand why but I'll take a look at it this week and if there's anything I can do I'll fix it
eyedeekay Do you have a github issue for it? If not, could you have him file one on: github.com/eyedeekay/go-i2pd
zzz I don't think the 'some guy' was using your go code eyedeekay, I think it was a native C++ project?
orignal yes he uses C++ project
orignal but libi2pd
zzz anything else on 2) ?
zzz 3) what is next
zzz your topic orignal, go ahead
orignal this is the qeustion
orignal I would suggest chacha20 encrytpion for tunnels
zzz maybe PQ
zzz maybe nothing for a few months
zzz other ideas?
orignal what is pq?
zzz post quantum
orignal and how is it related to us?
zzz maybe add it to ratchet, or ntcp2
orignal I don't understand what's an advantage
zzz maybe, maybe not
zzz zzz.i2p/topics/3294 for the discussion
orignal yes I read that topic
orignal but it looks nothing by hype for me
zzz you could start a thread on chacha20 tunnels, get the discussion started
eyedeekay I suppose whether one sees an advantage is based on whether one believes a CRCQ is imminent
eyedeekay a couple months would also give go-i2p a chance to catch up a bit
eyedeekay to what extent is human-readable addressing a common feature in the context of meetings we have here in #ls2?
zzz define human-readable addressing?
zzz define CRQC?
eyedeekay SusiDNS, addressbook, jump services, TLS potentially
orignal I see instance advatbnage of chacha20 vs. AES for 1K messages
eyedeekay Cryptographically-relevant quantum computer
orignal it's just faster
zzz sure, "human-readable addressing" is a valid topic
zzz anything else for 3) ?
zzz obviously we're not deciding anything today
eyedeekay OK. I didn't exactly come prepared for it, but I have some ideas about how our addressing might be improvable, in particular I'd like to define a protocol for allowing host providers to share addresses with eachother and under what circumstances they'll accept such a "sync" operation
zzz zzz.i2p threads always good to start discussion or "pre-proposals"
eyedeekay Don't have much for it written down but maybe I come to zzz.i2p with an idea soon
zzz interesting eyedeekay
zzz not clear if there's any middle ground between our current halfassed subscriptions and full-blown GNS
eyedeekay I had one of my toy projects, called "jump-transparency" which was designed to spot names that were registered with different keys on different services that sort of prototyped this but like a lot of things I did, it was an experiment that didn't get much attention after a certain point
orignal another thing
orignal possible disconnect Russia from Internet
orignal next month
zzz that would be something
orignal because on Sept 30 Putin will declare war
orignal to the world
orignal so i2p my split to two segments
orignal *mght
orignal just FYI
zzz you have any RU reseeds?
orignal R4SAS has one
zzz should we add it?
orignal I thought it's in the list already
dr|z3d we could probably get a reseed host up in kyiv, but that's probably not going to help.
zzz I have a reseed-pl and reseed-fr, but maybe that's a lie?
zzz "https://reseed-pl.i2pd.xyz/" + ',' + // r4sas-reseed_at_mail.i2p.crt // CA
zzz "https://reseed-fr.i2pd.xyz/" + ',' + // r4sas-reseed_at_mail.i2p.crt // CA
R4SAS no, it is not public
zzz ok, if you make it public let us know
zzz anything else for the meeting?
eyedeekay Nothing from me
zzz good luck R4SAS
zzz stay safe
R4SAS I will, but I need to think about reachability to my servers in RU )))
orignal ha ha
R4SAS would see in new month
zzz thanks everybody
orignal after 30-th
R4SAS i don't think they will block access for ex-USSR countries
R4SAS so there probably we will find possibility to get outside wall
dr|z3d Belarus will be fine.
orignal not they
orignal other world can do it
orignal Putin is too dumb to be able to do it
R4SAS don't underestimate the dumb