Overview and Logs for the tini2p Dev Meeting Held on 2019-03-28
Project infrastructure, Current status, Roadmap, I2P proposal implementation, and miscellaneous
dev diaries
i2p
crypto
el00ruobuob / oneiric
Logs
<oneiric_> 0. greets <oneiric_> hi <Corklander> Hey! <endogenic> o/ <oneiric_> 1. Project infrastructure / DevOps <kinghat> hi all <oneiric_> recently setup CircleCI, Docker and Bitbucket CI pipelines <oneiric_> going to try out Drone CI (integrates w/ GitLab), and decide on a final CI provider <oneiric_> will keep alternative configs in contrib/ <oneiric_> got a basic Docker image setup with everything needed to build tini2p <oneiric_> need to fix net tests to be able to run in a container for accurate coverage stats <oneiric_> other than that, CI is more-or-less setup <oneiric_> will be experimenting/playing with gitea over the next weeks, and mirror GitLab to GitHub, BitBucket, and gitea <oneiric_> unfortunately lost access to tini2p account on GitHub, though still have push ability <oneiric_> main github repo is now: https://github.com/tnii2p-project/tini2p <Corklander> 404 :( <oneiric_> derp <oneiric_> type <oneiric_> typo <oneiric_> my fingers are terrible today <oneiric_>https://github.com/tini2p-project/tini2p <Corklander> Yea! <oneiric_> -_- <oneiric_> lol <oneiric_> anyway <oneiric_> any questions/comments? i think that's all for infrastructure stuff. <Corklander> Do you see any dependency problems with Drone CI or the other choices? (I haven't used them so I'm not sure about any restrictions/costs.) <oneiric_> most of them i've seen are basically the same <oneiric_> x free hours for builds, then monery <oneiric_> the integrations are the differentiators <oneiric_> circle works with github and bitbucket, drone works with gitlab <Corklander> sry, have to go due to emrf, bbl <oneiric_> think there are webhook integrations on most/all major gitserver providers <oneiric_> anyway <oneiric_> ttyl <oneiric_> anything else on 1? <oneiric_> 2. Current project status / what's been done <oneiric_> spent a lot of time replacing crypto++ with libsodium + tiny-AES-c <oneiric_> then realized, "oh, i need an SSL lib" <oneiric_> so, added LibreSSL as a dependency for potential future Keccak patch, and replaced tiny-AES-c with SSL impl <oneiric_> that's in an open MR: https://gitlab.com/tini2p/tini2p/merge_requests/2 <oneiric_> finished with code cleanup, generic crypto/signature wrappers, and a basic experimental ECIES-X25519-AEAD-Ratchet impl <oneiric_> merged MR: https://gitlab.com/tini2p/tini2p/merge_requests/1 <oneiric_> currently working on generic wrappers for the proposed Blake2b EdDSA variants <oneiric_> which i will discuss in 4 <oneiric_> questions/comments on 2? <endogenic> awesome progress <oneiric_> will also hopefully be able to help zzz with ECIES testing (fingers-crossed) <oneiric_> thanks endogenic :) <oneiric_> 3. Short-term road map <oneiric_> implement the Blake2b sig variants to help other I2P projects with testing <oneiric_> implement tunnels + basic netdb <oneiric_> if lmdb interface is simple enough, will skip past doing blockfile format, and just use lmdb for addressbook/netdb needs <oneiric_> that's where i want to be anyway :) <oneiric_> adding alpha and beta milestones to the gitlab project <oneiric_> projected alpha release for 2019/07/10 <oneiric_> with a beta following 3 months after <oneiric_> will see how this first cycle goes, but will try to keep to that release schedule, or shorten it if feasible <endogenic> good plan <endogenic> the minimum usable set is generally critical to keep in mind to avoid getting lost along the way <endogenic> dont think that's a huge issue here given the name "tiny" though :) <endogenic> good origins <oneiric_> yeah, if shorter, may do alpha (2 months) -> beta (2 months) -> point release <oneiric_> true, as part of 4. i'll talk about the current sig variant proposals, and which of those will end up in tini2p <oneiric_> which is actually a nice transition (smooth endogenic) <oneiric_> 4. I2P proposal implementation <oneiric_>https://geti2p.net/spec/proposals/123-new-netdb-entries <oneiric_>https://geti2p.net/spec/proposals/144-ecies-x25519-aead-ratchet <oneiric_>https://geti2p.net/spec/proposals/146-red25519 <oneiric_>https://geti2p.net/spec/proposals/148-eddsa-blake2b-ed25519 <oneiric_> we went over 123 and 144 last meeting, iirc <oneiric_> one thing that has changed, zzz has said he will be switching focus back to ECIES, and hammering out some spec details <oneiric_> so we are getting closer to a finalized spec, which will reduce code footprint <oneiric_> 146 and 148 are the new sig variant proposals <oneiric_> for eddsa <oneiric_> RedDSA is needed for blinding encrypted lease sets, and Blake2b variants are needed to protect against LEA attacks <oneiric_> Ed25519ctx was also suggested, and is being considered for LEA protection <oneiric_> personally, think Ed25519ctx is gross given the alternatives, but it's in an RFC <oneiric_> RFC 8032 to be specific <oneiric_> the hash function in RFC 8032 isn't specified, other than being a crypto-secure hashing function with 512-bits security <oneiric_> Blake2b satifies all criteria <oneiric_> for full details see the proposals and #ls2 meeting logs <oneiric_> so, my plan is to write generic wrappers for all variants, then keep whichever get accepted <oneiric_> right now, it's looking like RedDSA-Blake2b-EdDSA can be used for all places where a signature is needed <oneiric_> a lot is still up in the air though, and nothing has been finalized yet <oneiric_> any comments/questions on 4? <oneiric_> one more thing actually, EdDSA-Blake2b-Ed25519 could also be used everywhere a signature is used <oneiric_> afaiui EdDSA-Blake2b-Ed25519 can also do blinding needed for encrypted lease sets <oneiric_> 5. Confirm next meeting time <oneiric_> same time two weeks? <kinghat> yup <oneiric_> right on, meeting over. thanks to all for attending