monero-site/_posts/2019-02-07-logs-for-the-tini2p-dev-meeting-held-on-2019-02-07.md
2019-03-29 07:18:14 +01:00

9.3 KiB

layout title summary tags author
post Overview and Logs for the tini2p Dev Meeting Held on 2019-02-07 Project design & goals, Current status, Timeline, Roadmap, Contributors outreach, and miscellaneous
dev diaries
i2p
crypto
el00ruobuob / oneiric

Logs

<oneiric_> 0. Greetings
<oneiric_> hallo
<Corklander> Hello!
<oneiric_> 1. Project design + goals
<oneiric_> the main (somewhat rough) design document is here: https://github.com/tini2p/tini2p/DESIGN.md
<Corklander> I show this as the URL: https://github.com/tini2p/tini2p/blob/master/DESIGN.md
<oneiric_> i2p components will be separated into (mostly) independent modules
<oneiric_> thanks Corklander
<oneiric_> only the minimal set of features for a functioning i2p router will be implemented
<oneiric_> as new protocols come online (LS2, ECIES) old crypto will be deprecated and removed
<oneiric_> any questions?
<oneiric_> comments?
<Corklander> Are there any specific architecture requirements? As in, need an AES-boosted CPU, etc.?
<oneiric_> not that i can tell so far, but i haven't focused on multi-platform too much yet
<oneiric_> need to get it working on a single platform first :)
<Corklander> Yup. :)
<oneiric_> that being said, i'm trying to keep portability in mind, to ease multi-platform suppoort
<oneiric_> support*
<Corklander> A super-slim router would have the distinct advantage of very high portability even to SoCs.
<oneiric_> it would be amazing to run on a super slim board like that
<oneiric_> that may require a port to c which is a potential path to go down once an mvp router is finished
<oneiric_> ^ maybe
<Corklander> Also good to hear.
<oneiric_> a rust impl is also on the table, but we can revisit that in the roadmap section
<oneiric_> are we good to move on?
<kinghat> is there a non dev variant of #tini2p-dev?
<oneiric_> kinghat: absolutely #tini2p
<oneiric_> moving on
<oneiric_> 2. Current project status
<oneiric_> currently building out the ntcp2 transport, and will move to i2np, tunnels and netdb next.
<oneiric_> client modules are completely open to independent, parallel dev
<oneiric_> core components can be developed in parallel with some communication
<oneiric_> the code is still in somewhat high flux, and am just rebasing on a single commit atm
<oneiric_> fairly close to having the networking + session management for the ntcp2 transport
<oneiric_> after that, ntcp2 will be more-or-less finished
<oneiric_> any comments questions?
<Corklander> This is more architectural/design: what license do you plan to release?
<oneiric_> current license is BSD-3 (to be compatible w/ kovri+monero)
<oneiric_> it may change if necessary, currently don't see a need to
<Corklander> Good. :)
<oneiric_> ready to move on?
<oneiric_> 3. Development timeline estimates
<oneiric_> the code should stabilize in the next 1-2 weeks, and i'll change to making PR/MRs against the master branch
<oneiric_> after finishing ntcp2, i2np + tunnels should take ~1-1.5 weeks each to get working
<oneiric_> netdb will be somewhat more involved, and may take 2-3 weeks to get working
<oneiric_> the router context should be fairly easy to implement, ~1 week
<oneiric_> garlic encryption, notably AES+SessionTag management, is fairly complicated, ~2-3 weeks
<oneiric_> total estimated time for core components: ~7-12 weeks
<oneiric_> client components are somewhat easier to implement
<oneiric_> reseed and address book should take ~1.5-2 weeks each
<oneiric_> the most complicated components are i2cp and the proxy interfaces
<oneiric_> atm, only socks + http proxies will be implemented as APIs for external apps (~2-2.5 weeks each)
<oneiric_> i2cp is the interface between the client & router context, ~2 weeks
<oneiric_> client destinations manage the interface b/w proxies & the client context, ~1.5-2 weeks
<oneiric_> total estimated time for client components: ~7-8.5 weeks
<oneiric_> total time for mvp router: ~4-5 months
<oneiric_> the above estimates are conservative, and assume a singular developer
<oneiric_> actual dev time may be much less
<oneiric_> questions comments?
<oneiric_> ready to move on then?
<Corklander> Do you see use of wireframes/mockups that could help make development testing faster?
<oneiric_> currently using Catch2 as a testing framework
<oneiric_> all code so far is covered by test cases
<oneiric_> currently hammering out some network bugs for ntcp2 sessions, net tests have been extremely helpful here, for example
<oneiric_> does that answer your question?
<oneiric_> or were you talking about something else?
<Corklander> (I'm jumping the gun and asking about how to share workload using wireframes.)
<oneiric_> oh, i have a diagram for component interaction that i'll finish and post after the meeting
<oneiric_> it still contains streaming + SAM components, which likely won't be implemnted
<oneiric_> streaming library may be, but it may turn out to be unnecessary
<oneiric_> any more discussion, or ready for roadmap item?
<oneiric_> 4. Roadmap
<oneiric_> finish ntcp2 transport -> netdb impl -> tunnel impl -> garlic impl -> router context impl
<oneiric_> client destination impl -> address book impl -> socks 4a + http proxies impl -> client context impl
<oneiric_> the above roadmap is assuming singular dev, multiple devs will parallelize efforts
<oneiric_> questions comments?
<Corklander> On roadmap, should there be a list of infrastructure? As in git host, communications info, etc?
<oneiric_> haven't thought too much about infrastructure at this point
<oneiric_> once code is stable (~1-2 weeks), will dedicate more time to things like CI, git host, comms, etc
<oneiric_> right now, the project is hosted on github/lab
<oneiric_> heard about gitea, which also sounds like a great option
<oneiric_> will likely setup a meta meeting to discuss all of that
<oneiric_> thanks for bringing that up Corklander, easy to forget about
<oneiric_> transitions nicely to the next item
<oneiric_> any more discussion before moving on?
<oneiric_> 5. Project management + contributor outreach
<oneiric_> i am a developer, not a management type, and the skillsets are very different
<oneiric_> i can do project management, but this is not my strength
<oneiric_> at the moment, i am the only one contributing, so imho, project management is not that crucial
<oneiric_> the importance will shift once more contributors become involved
<oneiric_> it is good to be forward looking, and some time/effort should be dedicated to reaching out to community members with proven project management experience
<oneiric_> contributor outreach is hugely important, and once core components are in place, i will dedicate more time to looking for developers to help out
<oneiric_> any community help finding project managers + contributors is greatly appreciated
<oneiric_> questions comments?
<oneiric_> ok, almost top of the hour, final item
<Corklander> You've listed time as your only requirement for now. If you have assistance with coding it would likely impact your time to get the current roadmap finished.
<Corklander> What requirements would you like for people to assist you so that you can dedicate your time best?
<oneiric_> absolutely, more contributors familiar with i2p (or somewhat easily brought up to speed) should decrease dev time
<oneiric_> for client components, socks or http proxies should be the easiest to take on
<oneiric_> familiarity with c++ is a req
<oneiric_> doesn't have to be expert level, but novice-intermediate
<oneiric_> i'm still a bit of a c++ greenhorn, so it would take a bit of time for me to train devs
<oneiric_> anyone wanting to contribute to core components should be *very* familiar with i2p, or willing to invest a lot of independent time catching up on docs
<oneiric_> will try to guide people through the mire as best as possible
<oneiric_> for people totally unfamiliar with i2p, socks + http proxies will be the easiest introduction
<oneiric_> socks being the easier of the two
<oneiric_> ok, so we're a little over time
<oneiric_> final item
<oneiric_> 6. Next meeting time
<oneiric_> is this a good time/day for people (know some are in UTC+1, so maybe its a bit late?)
<oneiric_> also, thank you to all that attended/participated!
<Corklander> I'm good with this or later for weekdays.
<oneiric_> ok
<oneiric_> anyone else need a different time/day?
<oneiric_> so next meeting will be 18:00 UTC 21-02-2019 (two weeks from today)
<oneiric_> many thanks again to everyone who attended :)
<Corklander> Thanks oneiric!
<oneiric_> meeting adjourned *gavel strike*