I2P dev meeting, April 4, 2017 @ 20:00 UTC
Quick recap
- Present:
atoi, echelon, manas, orignal, randomrng, str4d, z3r0fox, zzz
Registro completo do IRC
20:00:05 <zzz> 0) Hi
20:00:05 <zzz> 1) 0.9.30 update (zzz)
20:00:05 <zzz> 2) UI branch status - (str4d)
20:00:05 <zzz> 3) I2P Summer Dev plans - (str4d)
20:00:05 <zzz> 4) EdDSA update - (str4d)
20:00:09 <zzz> 0) Hi
20:00:12 <zzz> hi
20:00:28 <manas> Hello
20:00:33 <eche|on> hi
20:01:00 <zzz> 1) 0.9.30 update (zzz)
20:01:24 <zzz> ok, things going well, testers are finding some problems which is great. Jetty 9 going pretty smoothly so far
20:01:35 <z3r0fox> Hi! o/
20:01:54 <zzz> I've identified the plugins that need updating on zzz.i2p... for best results those need to be rebuilt before the release
20:01:56 <eche|on> mostly yes
20:02:04 <eche|on> except for old jetty config files
20:02:24 <zzz> glad that we are doing it now so we will be ready for stretch and zesty
20:02:36 <eche|on> yes
20:02:56 <zzz> thanks to echelon and others for testing. Will probably mark a dev build as -rc sooner than usual to get more testing
20:03:12 <eche|on> ok
20:03:20 <orignal_> hi
20:03:34 <zzz> I have set the checkin deadline for friday april 28, release first week of may
20:03:54 <zzz> anything else on 1) ?
20:04:15 <atoi> when will Java have GOST signatures?
20:04:41 <eche|on> 2019
20:05:01 <zzz> atoi we haven't agreed to the proposal yet, if we do, best guess late next year due to priorities
20:05:02 <eche|on> bu fine with 1, good we have postponed UII
20:05:54 <zzz> but I've made some other proposals (136 and 137) to make the introduction of new sig types easier
20:06:00 <zzz> anything else on 1) ?
20:06:02 <atoi> can't wait to have GOST working
20:07:22 <zzz> 2) UI branch status - (str4d)
20:07:35 <zzz> str4d, what's the latest on your ui branch?
20:07:53 <str4d> UI branch has been relatively stable for the last few weeks
20:08:28 <zzz> plan is to prop in early may for .31 ?
20:08:57 <str4d> There's some theme updates I need to integrate and push addressing some feedback, but in terms of structure it's basically there
20:09:08 <str4d> Yeah
20:09:13 <zzz> ok great
20:09:19 <zzz> anything else on 2) ?
20:09:24 <eche|on> can´t wait to test it^^
20:10:00 <str4d> If anyone wants to test but doesn't want the hassle of building, I'll throw up an i2pupdate.zip once these next theme changes are in
20:10:18 <eche|on> I wait for prop...
20:11:02 <zzz> 3) I2P Summer Dev plans - (str4d)
20:11:19 <zzz> str4d what are the plans?
20:11:47 <str4d> Okay, sadie and I threw around some ideas, and what shook out was that for this Summer Dev, we should focus on speed
20:12:37 <str4d> Because a) it's a logical extension of last year (we made it easier for apps to use I2P, but now we need to make them *want* to), and b) it covers about half of our existing roadmap already
20:13:37 <zzz> ok, where might we find these plans, and how do we find people to do them?
20:14:06 <str4d> sadie took notes from my rambling, so I'll pester her for them :)
20:14:18 <str4d> I'm basically thinking this means:
20:14:51 <str4d> - Push forward the various proposals that affect speed (LS2, newer E2E encryption, massive multihoming)
20:15:24 <str4d> - Get something running similar to Tor's bwauth that we can use to start collecting metrics specifically about network speed
20:16:27 <manas> I was thinking of writing some code to test I2P speeds with different programs (rsync, torrents, sftp etc)
20:16:34 <str4d> - Expose tunnel selection through I2CP (a la psi's lua stuff)
20:16:51 <manas> I can share that if it will be useful/interesting
20:16:51 <randomrng> massive multihoming <3
20:17:00 <zzz> sounds good. should we put this on the agenda for next month's meeting to get an update?
20:17:00 <str4d> - ElGamal speedups for the short term
20:17:17 <str4d> Yep.
20:17:35 <zzz> I note that most of what's on the .30 roadmap, set only 3 months ago at CCC, is total fiction and will be pushed out. I'll be updating it shortly.
20:18:10 <zzz> anything else on 3) ?
20:18:16 <str4d> We have two months until it will officially start. I'll continue chatting with sadie to flesh out a plan, but what we *really* need is people to pick things they want to work on
20:18:32 <str4d> manas already has a head-start there ;P
20:18:35 <orignal_> what kind of speedup?
20:18:48 <str4d> orignal_, the table precomputation
20:18:48 <manas> :)
20:19:03 <orignal_> mine or something else?
20:19:09 <zzz> if you and sadie document a framework and options that will help people pick things
20:19:22 <str4d> +1
20:19:30 <zzz> anything else on 3) ?
20:19:35 <str4d> orignal_, Won't be yours, because that directly leverages OpenSSL which we don't have access to
20:19:52 <orignal_> I mean basis
20:20:03 <orignal_> not implementation
20:20:28 <str4d> Oh, yes likely based on yours (although I'd want to understand the mathematics myself)
20:21:03 <zzz> 4) EdDSA update - (str4d)
20:21:10 <orignal_> good to know :)
20:21:13 <zzz> str4d, what's going on with EdDSA?
20:21:39 <str4d> I just pushed EdDSA-Java 0.2.0, over a year after the last release
20:22:28 <orignal_> what's a difference?
20:22:37 <str4d> Includes various cleanups, the soon-to-be-standard encoding for PKI, fixed JCA naming, a Security Provider, and should actually be constant-time now thanks to a third party that had it audited
20:22:59 <str4d> I've pulled the code into i2p.i2p.zzz.test2
20:23:26 <orignal_> any speed improvements?
20:23:30 <str4d> But what it now needs is testing to check that everything still works fine
20:23:45 <zzz> how much testing have you done?
20:23:58 <str4d> orignal_, speed decrease in signing due to making it constant-time (but not that much)
20:24:06 <str4d> zzz, the EdDSA code itself is well-tested
20:24:20 <orignal_> constrant time of signing or verifing?
20:24:32 <zzz> I'm still a skeptic about constant-time, but if we don't do it everybody will complain
20:24:39 <str4d> signing (verifying has never been constant-time, as there's no secret info)
20:25:01 <zzz> str4d, you targeting 31 or 30?
20:25:02 <str4d> What I *haven't* tested yet is how it interacts with the rest of I2P, specifically relating to the JCA naming fixes
20:25:39 <zzz> as I posted on zzz.i2p I think it's a fool's errand to chase the naming guidelines from some RFC
20:25:47 <str4d> Probably too close to 30 to pull in as we're using the code directly instead of the library
20:25:50 <orignal_> the performance botlleneck is verification
20:26:22 <str4d> For my library, it made sense to fix things before 0.2.0 because there wasn't a Provider before, so no one was using the JCA names
20:26:26 <zzz> str4d, the prop will also be merge hell due to the javadoc changes you made that I already fixed in .29, wont be fun
20:27:14 <str4d> Okay, sounds like the plan is to merge i2p.i2p into .test2, fix the merge conflicts, then get people testing it
20:27:25 <zzz> but we've always had a provider. your changes in test2 tended to prefer some github PR over what I already did to accomplish the same thing, so I'm a little skeptical about the whole thing
20:27:42 <zzz> you could prop that way if you want, or just defer the pain until the prop the other way
20:28:23 <str4d> Yes, but if we're honest, I doubt anyone was using it
20:28:31 <zzz> as I said, I defer to you, it's originally your code, but I still want to stare at it
20:29:11 <zzz> I have one router with a EdDSA family key to test. can't remember what format. Remember I made all these fixes almost 4 months ago, since then we've been waiting for you and your RFC friends :)
20:29:11 <str4d> (who wasn't already directly importing i2p.jar, and would already have migrations to do when they upgrade)
20:29:32 <zzz> anything else on 4) ?
20:30:08 <str4d> Not from me :)
20:30:13 * zzz looks for the baffer
20:30:19 <zzz> anything else for the meeting?
20:31:20 <str4d> Tor's next tor-dev meeting in September-ish is likely to be in Montreal
20:31:33 <orignal_> when?
20:31:53 <z3r0fox> Neat
20:31:55 <str4d> ------------------------------------^
20:31:56 <zzz> anything else for the meeting?
20:32:14 <str4d> I'm planning on going
20:32:28 <zzz> thinking of switching teams?
20:32:47 <str4d> Just aiming to influence them ;)
20:33:14 <zzz> save yourself the plane ticket money, that's not going to happen
20:33:24 <manas> Ya Montreal is nice :)
20:33:27 <str4d> Also hoping that, as Summer Dev will have just finished, we'll have some nice speed improvements to talk about there
20:34:28 * zzz *BAFS*** the meeting closed