I2P dev meeting, March 2, 2021 @ 20:00 UTC

Quick recap

  • Present:

eyedeekay, zzz, zlatinb

Full IRC Log

(08:01:02 PM) eyedeekay: Hi everyone and welcome to the March 2 Meeting, please let me know if you're here
(08:01:27 PM) eyedeekay: zzz zlatinb eche|on eche|off
(08:01:42 PM) eyedeekay: Agenda
(08:01:42 PM) eyedeekay: 1) Hi 
(08:01:42 PM) eyedeekay: 2) 0.9.49 remaining items 
(08:01:42 PM) eyedeekay: 3) Mac Launcher Status 
(08:01:42 PM) eyedeekay: 5) 0.9.50 release 
(08:01:42 PM) eyedeekay: 6) Trac migration summary
(08:01:46 PM) Irc2PGuest1578 [kilian@xvbemdlawzj2qlt3cgjgaclevziobxvwmipcvecbla4xqkmwjd2q.b32.i2p] entered the room.
(08:01:46 PM) zzz: hi
(08:01:55 PM) zlatinb: hi
(08:01:55 PM) eyedeekay: 4) 1.0.0 vs 0.9.50
(08:03:04 PM) eyedeekay: hi zzz, hi zlatinb, timeout 30s anyone else?
(08:03:39 PM) eyedeekay: Thanks everyone, starting right in with 2) 0.9.49 remaining items
(08:03:51 PM) eyedeekay: The only one I know of is the .dmg version of the Mac installer
(08:04:20 PM) zzz: the others are official debian and ubuntu
(08:04:45 PM) zzz: I'll explain a little more
(08:04:50 PM) eyedeekay: Ok thanks.
(08:05:06 PM) zzz: unfortunately, debian bullseye just hit a freeze
(08:05:29 PM) zzz: our debian maintainer either wasn't aware of the schedule or didn't advise us to hurry
(08:05:51 PM) zzz: so while we did pull in the schedule to for ubuntu hirsute 21.04, the debian deadline was earlier
(08:06:09 PM) zzz: since ubuntu pulls from debian, ubuntu didn't get it either
(08:06:45 PM) zzz: this is a once every two year thing, but still, would have been nice to know
(08:06:54 PM) zzz: as it was, we hurried up about ubuntu, all for nothing
(08:07:25 PM) zzz: so, at some point, debian will unfreeze, and 49 should show up in sid. but bullseye is 48
(08:07:27 PM) zzz: eot
(08:07:59 PM) eyedeekay: Thanks zzz. So for the time being the recommendation for Debian users to get an up-to-date router should be via our repository
(08:08:17 PM) zzz: yup. ditto ubuntu.
(08:08:32 PM) zzz: oh, if I may, a brief report on the network:
(08:08:41 PM) eyedeekay: Sure go ahead
(08:08:48 PM) zzz: 52% updated to 49; 6% rekeyed to ECIES. All looks good so far
(08:09:03 PM) zzz: very few bugs found or reported
(08:09:05 PM) zzz: eot
(08:09:21 PM) eyedeekay: Excellent to hear, thanks for the report
(08:09:49 PM) eyedeekay: And I guess I can work on figuring out what mailing list we need to be subscribed to to get word earlier on when Debian will freeze
(08:10:02 PM) eyedeekay: 3) Mac Launcher Status
(08:10:14 PM) eyedeekay: This is the DMG based installer, not the .jar
(08:10:54 PM) eyedeekay: I dropped the ball on this one, by failing to notify people that the previous maintainer was no longer building the installer
(08:11:15 PM) eyedeekay: As a result I removed the Mac installer from the site
(08:11:39 PM) zzz: iirc the last one built was .45 a year ago, and it probably was a broken link for most of last year
(08:11:41 PM) eyedeekay: I have since aquired a Mac with the intent to take up maintenance of the product
(08:12:00 PM) eyedeekay: zzz you are correct
(08:12:04 PM) zlatinb: there is a problem with the dmg installer - at least on my mac I can't get the router to stop.  Some daemon keeps restarting it
(08:12:09 PM) zzz: so it was actually a longstanding issue. you were correct to remove it, thanks for that
(08:13:16 PM) zlatinb: so if other mac users are in the same situation we should come up with some sort of guide for cleaning up
(08:13:28 PM) zzz: have you figured out if there's some auto-update or notification built in? and if so is that broken also? or is it just the news entry in the console?
(08:13:57 PM) zlatinb: auto-update does work, strangely enough
(08:13:57 PM) eyedeekay: It can't auto-update, at least not successfully
(08:14:04 PM) eyedeekay: Oh well that's weird
(08:14:09 PM) zlatinb: I just can't kill it and make sure it stays dead
(08:14:28 PM) eyedeekay: Well sounds like some of the behavior is pretty erratic
(08:14:56 PM) zzz: eyedeekay, last we discussed it, there was some debate as to the value of this installer product to our users, compared to the effort required to maintain it
(08:15:15 PM) zzz: how do we nvestigate and evaluate those two factors?
(08:15:58 PM) zzz: and zlatinb do you have any thoughts on the value of a "mac way" installer today?
(08:16:36 PM) zlatinb: I still think a Mac way and Win way installers are far superior than the izpack monstrocity
(08:16:37 PM) eyedeekay: I think zlatinb and I will need to compare notes, I'm seeing different behavior than he is and if I don't know why continuing to build and support it becomes much more intimidating
(08:17:16 PM) zlatinb: but I think we need to re-evaluate the complexity in light of jpackage coming out with Java 14+
(08:18:02 PM) zlatinb: either way, a  Mac-way installer would/should be lower priority than Win-way installer
(08:18:05 PM) zzz: I'm not a mac person, but "far superior" was the consensus at the time we started development of the installer
(08:18:43 PM) zzz: if the consensus is different now, I'd like to understand why
(08:19:24 PM) zlatinb: to my knowledge it's still the same consensus, just the ecosystem has changed (i.e. jpackage exists)
(08:20:26 PM) eyedeekay: IIRC my experience with Mac at the time was basically nil and my favor for the idea was based on the idea that working with familiar packaging systems makes our packages easier to trust
(08:20:39 PM) eyedeekay: jpackage does the runtime image/eliminate the need to install Java thing right? the dmg to my knowledge didn't do that?
(08:20:51 PM) zzz: right
(08:21:18 PM) zlatinb: right
(08:21:30 PM) zlatinb: jpackage builds dmgs supposedly, I haven't tried it
(08:21:38 PM) zzz: so jpackage would be some 100MB thing. since it's only for one OS, it's feasible to do it for mac. 
(08:21:47 PM) zzz: yeah dmgs would have to be tested for sure
(08:22:14 PM) zlatinb: it builds windows installers too, I haven't used that functionality though
(08:22:26 PM) zlatinb: and rpms and debs but I'm pretty sure we don't want those
(08:22:52 PM) zzz: one of our failings as a project is that the dmg was always labeled 'experimental' on our d/l page. We never paid it enough attention to remove the label or even notice that nobody was building it
(08:22:57 PM) Irc2PGuest1578 left the room (quit: Read error).
(08:24:06 PM) zzz: as with all our other official products, if we're going to support it we need enough resources for a competent maintainer
(08:25:15 PM) zzz: at this point I propose that we continue the evaluation of both user demand and effort required, both for existing dmg and jpackage.
(08:25:29 PM) zzz: interim report in one month, final decision in two months, in time for .50
(08:25:52 PM) zlatinb: any thoughts how to go about that?  survey? 
(08:26:32 PM) eyedeekay: I could set up a Reddit survey after the meeting
(08:26:42 PM) zzz: forum posts
(08:27:11 PM) eyedeekay: Works for me, I'll add it to next month's meeting agenda
(08:28:06 PM) eyedeekay: Anything else on 3)?
(08:28:32 PM) eyedeekay: 4) 1.0.0 vs 0.9.50
(08:29:02 PM) zzz: this was my item
(08:29:10 PM) eyedeekay: Take it away zzz
(08:29:28 PM) zzz: I don't feel strongly either way, but I think we should go to 1.0.0 in the next year or so
(08:29:49 PM) zzz: as we don't have a separate stable branch, 1.0.0 is not a particular guarantee of stability
(08:30:23 PM) Irc2PGuest1578 [kilian@xvbemdlawzj2qlt3cgjgaclevziobxvwmipcvecbla4xqkmwjd2q.b32.i2p] entered the room.
(08:30:27 PM) zzz: so my question is what people think, and can the PR team accomplish messaging on what 1.0.0 is or isn't, on some timeline?
(08:30:29 PM) zzz: eot
(08:31:14 PM) zlatinb: so I have two points regarding 1.0.0:
(08:31:41 PM) zlatinb: 1) RED needs tuning and I will die on that hill if I must.  Tunning it properly may require more than one release
(08:32:19 PM) zlatinb: 2) Back to the installers issue - if we can build much smoother installers for the major platforms, a 1.0.0. release will have much greater impact
(08:32:20 PM) zlatinb: eot
(08:33:40 PM) eyedeekay: I think we can devise and accomplish messaging and PR for 1.0.0, if 1.0.0 coincides with migration of cryptography away from Elgamal, and I agree with zab on 2)
(08:34:30 PM) zzz: we can always pick some headline feature to brag about, any release. It's fairly arbitrary. We could pick any release this year and claim it's when we're ditching elgamal. It's happening already
(08:35:44 PM) zzz: as I'm not hearing any strong consensus, I propose that the next release is 0.9.50, and we discuss it again after that release, in 3 months
(08:35:51 PM) eyedeekay: Then 2) remains pretty important to me, installers are a pain point as strange as that seems
(08:36:15 PM) eyedeekay: I agree that the next one should be 0.9.50
(08:36:27 PM) anonymousmaybe left the room (quit: Read error).
(08:36:31 PM) T3s|4 left the room (quit: Read error).
(08:37:36 PM) eyedeekay: Anything else on 4)?
(08:38:16 PM) T3s|4 [~T3s4@573a4z46ixhpfeuej2hggtzg2wvsllq6nurtha5dzpd7l42awaeq.b32.i2p] entered the room.
(08:38:16 PM) mode (+v T3s|4) by ChanServ
(08:38:18 PM) eyedeekay: 5) 0.9.50 release
(08:38:42 PM) anonymousmaybe [anonymousm@zvezcslfl5ndd6ciniqp2ei3cm6kvcovceeu3nzheqe7rqcj3rra.b32.i2p] entered the room.
(08:38:42 PM) mode (+v anonymousmaybe) by ChanServ
(08:39:19 PM) zzz: I'll let you go first, then I'll list what I've been up to
(08:41:28 PM) eyedeekay: It's been about 2 weeks since the 0.9.49 release, in that time I've been working on style bugs, moving the configuration of X-I2P-Location into the application instead of requiring a reverse proxy or specific configuration, and have been working on finding ways to improve gitlab
(08:42:17 PM) eyedeekay: In particular a way to migrate trac tickets to gitlab en-masse and a way to create tickets anonymously are on my gitlab list
(08:43:06 PM) eyedeekay: Those are actually largely accomplished and part of the next agenda item, so I won't waste time on that now
(08:43:56 PM) eyedeekay: EOT
(08:44:06 PM) zzz: super
(08:44:26 PM) zzz: I fixed NTP for the year 2036 issue
(08:44:33 PM) zzz: implemented UPnP for IPv6
(08:44:45 PM) zzz: reduced memory usage by the profiles
(08:44:55 PM) zzz: added support for IPv6 introducers
(08:45:17 PM) zzz: added "4/6" caps support for better tracking of who can connect to who
(08:45:39 PM) zzz: did some work on smaller tunnel build messages (prop. 157), although that work is going a lot slower than the #ls2 team would like
(08:46:26 PM) zzz: and I reported a major SSU bug to i2pd. they've fixed it. I'm hopeful they will cut a release for it this month, as I think it's really affecting network performance for some subset of connections
(08:46:35 PM) zzz: eot
(08:46:44 PM) eyedeekay: Thanks zzz
(08:47:25 PM) zlatinb: I would like to do some experiments wrt tuning RED in the testnet.  Current theory is that it's way too aggressive and slows down single-stream connections to an unnecessary degree.  Will report as usual.  EOT
(08:47:36 PM) eyedeekay: Thanks zlatinb
(08:48:17 PM) eyedeekay: 6) Trac migration summary
(08:48:17 PM) zzz: re: roadmap. I updated it today on the website to reflect what was in .49 and moved other stuff to .50. eyedeekay please do the same for the items you know about
(08:48:32 PM) eyedeekay: Ack zzz, will do that this evening
(08:51:18 PM) wodencafe left the room (quit: Read error).
(08:51:37 PM) wodencafe [wodencafe@4qx5zjj3rypztq5h4kc2clviwid5cir7cm6iqrqa2l2npvlgt7ta.b32.i2p] entered the room.
(08:51:51 PM) eyedeekay: Re: trac I am in a rock and a hard place here. I'm admin on trac and not the box that trac runs on. I can't do anything to update it or make it better on my own, all I can do is chase time-consuming issues.
(08:51:51 PM) eyedeekay: I really want to get rid of it, but we obviously can't blow away all those tickets or the rest of the information here.
(08:51:51 PM) eyedeekay: I'm proposing that we migrate trac tickets to gitlab tickets and encourage the use of gitlab for issue-tracking purposes
(08:52:51 PM) eyedeekay: Trac tickets don't map 1:1 onto gitlab tickets, tickets for I2P applications will need to be added to the i2p.i2p issue tracker and tagged on gitlab with the corresponding application
(08:54:04 PM) eyedeekay: I've finally figured out how to do it using some of the corresponding material from Tor
(08:54:37 PM) zzz: that's probably the right answer, but we should probably do a quick evaluation of the alternatives, for example just copying over everything to trac on a box we control
(08:54:51 PM) zzz: and again, an estimate of the one-time and ongoing resources required
(08:55:18 PM) zzz: we were going to have a meeting a couple months ago on it, so perhaps now it's time
(08:55:54 PM) lithium left the room (quit: Quit: leaving).
(08:56:02 PM) eyedeekay: Instinctively, running 2 services(Trac and Gitlab) will probably be higher effort over time, but maybe less effort initially
(08:56:05 PM) zzz: i just want to be clear on what we're trying to accomplish
(08:56:05 PM) lithium [lithium@f25fchfdvktukmhg2rkz5es4mlrroyywcou27bpr4mxzfuf3jgya.b32.i2p] entered the room.
(08:56:38 PM) zzz: a full migration to gitlab is an enormous fix for the problem of somebody not responding to emails
(08:56:50 PM) zzz: so the question is what else do we get for it
(08:57:58 PM) zlatinb: tight integration with git, MRs, code review, all that
(08:58:02 PM) zzz: and we need a short list of our requirements, esp. for registration and anti-spam
(09:00:01 PM) zzz: I also think we should take lessons from the git migration last year, and have clear milestones and schedule and status
(09:00:36 PM) eyedeekay: Registration has become a difficult point. I estimate roughly 1/3 of registrations are spam, but it's so difficult to tell the difference because I do not ask for very much information from git users
(09:01:37 PM) eyedeekay: Tor's solution re: anonymous registration is neat, and potentially very useful, but the more I look at it the more I think it may be overkill for us
(09:02:35 PM) zzz: I propose that we find out who wants to be in a meeting about this, and then we'll schedule the meeting later
(09:03:29 PM) eyedeekay: I can work with that. I'll start a new forum thread for the Trac Migration.
(09:04:49 PM) zzz: zlatinb, would you like to be in on it?
(09:05:03 PM) zlatinb: sure
(09:05:21 PM) zzz: super
(09:07:56 PM) eyedeekay: That's everything from the agenda, anything else to add?
(09:08:00 PM) eyedeekay: Timeout 60s
(09:09:32 PM) eyedeekay: That closes the meeting *baffs*
(09:09:32 PM) eyedeekay: Thanks zzz zlatinb for coming, I'll post the meeting log to the site shortly
(09:10:09 PM) zzz: thank you
(09:11:05 PM) devcron left the room (quit: Quit: leaving).
(09:11:11 PM) eyedeekay: no problem zzz