I2P dev meeting, January 4, 2022 @ 20:00 UTC

Quick recap

  • Present:

eyedeekay, zzz, zlatinb

יומן IRC מלא

(03:00:39 PM) eyedeekay: Hi everyone and welcome to the January 4, 2022 meeting
(03:00:39 PM) eyedeekay: 1. Hi
(03:00:39 PM) eyedeekay: 2. Meeh services/Outproxy requirements http://zzz.i2p/topics/3218-meeh-services-going-down-soon
(03:00:39 PM) eyedeekay: 3. Promoting Mac DMG JPackage to stable https://geti2p.net/en/download/mac
(03:00:39 PM) eyedeekay: 4. NetDB research (Sim/Model) http://zzz.i2p/topics/3227-netdb-simulator-proposal
(03:01:13 PM) zlatinb: hi
(03:01:23 PM) zzz: hello
(03:01:55 PM) eyedeekay: Hi zzz, zlatinb, anybody else here today?
(03:02:20 PM) eyedeekay: Outproxy operators?
(03:02:59 PM) eyedeekay: OK then on to 1, they can chime in if they want :)
(03:03:38 PM) eyedeekay: So we're losing all of meeh's non-static services, for us that's mostly trac, which we've already replaced, and false.i2p, the default outproxy
(03:04:07 PM) eyedeekay: Neither have worked well for a long time, but not having an outproxy tends to have a bad UX so we need to replace it
(03:04:59 PM) eyedeekay: We discussed some options last week, one of which was  finding a new volunteer-run outproxy to make it the default.
(03:06:30 PM) eyedeekay: If that is to happen, then we'll need it to meet some basic requirements. Obviously a privacy policy is important.
(03:06:30 PM) eyedeekay: If they aren't passing everything to Tor directly, we'll also need a description of what filtering, if any, occurs
(03:07:34 PM) eyedeekay: Other options are to run one ourselves, which is of limited appeal at best, another is local SOCKS support, which requires a local SOCKS proxy to be available.
(03:08:07 PM) eyedeekay: If anyone wants to volunteer to be an outproxy, the place to reach out is at zzz's forum
(03:09:05 PM) zzz: right
(03:09:12 PM) eyedeekay: I have not been able to get statistics about false.i2p usage from meeh yet. EOT
(03:09:40 PM) zzz: first of all we want to be very careful not to publicly recommend an alternate or set a new default without approval, as that might impose a large burden, financial or otherwise
(03:10:06 PM) zzz: as a tiny baby step, we're looking to get alternates registered on stats.i2p and get approval to add them to our default hosts.txt
(03:10:16 PM) zzz: as that at least makes it easier for people to configure and use them
(03:10:55 PM) zzz: we have approval for that step for two outproxies, however only one is currently registered on stats.i2p
(03:11:05 PM) zzz: so I'm awaiting for that second registration
(03:11:05 PM) zzz: eot
(03:11:35 PM) eyedeekay: Thanks zzz, I'll remind him btw
(03:11:47 PM) dr|z3d: I've had some preliminary discussions with you, zzz, and on that basis I think it should be fairly clear that I'm not interested in offering an official service. Informal is fine, but nothing official.
(03:12:52 PM) zzz: thanks for making it clear, and also thanks for your time and expertise in thoes discussions
(03:12:59 PM) dr|z3d: I'm also of the view that a single outproxy is doomed to failure the moment it becomes official.
(03:14:13 PM) eyedeekay: Certainly can see why that would be the case. Anything else on 1?
(03:14:28 PM) zzz: perhaps. it certainly requires adequate capacity and competent administration
(03:14:29 PM) zzz: eot
(03:14:59 PM) eyedeekay: 3. Promoting Mac DMG Jpackage to stable
(03:15:15 PM) eyedeekay: s/1/2/
(03:15:42 PM) eyedeekay: So this topic was added by zlatinb, take it away
(03:16:21 PM) zlatinb: thanks.  Basically, the non-DMG (jar) package is not notarized and the last 2 (or 3?) versions of Mac OS refuse to execute it unless the user changes some advanced settings
(03:16:52 PM) zlatinb: at the same time, the DMG jpackage has been very stable and in-net updates have been tested end-to-end
(03:17:10 PM) zlatinb: so I propose we make it the default mac download and get rid of the .jar option altogether
(03:17:55 PM) zlatinb: since we already have a nice /mac page dedicated to the package, I suggest we keep that, removing the BETA modifier
(03:18:14 PM) zlatinb: and on the Mac div in /download, just have a link to the /mac page, nothing else
(03:18:19 PM) zlatinb: eot
(03:19:22 PM) eyedeekay: I support it.
(03:20:42 PM) eyedeekay: Do we need to wait for release or just do it now?
(03:21:51 PM) anonymousmaybe is now known as Irc2PGuest95343
(03:22:35 PM) mode (+ao zzz zzz) by ChanServ
(03:22:37 PM) zlatinb: no need to wait for a release
(03:22:54 PM) zzz: did you see my 3 questions before I dropped?
(03:23:01 PM) eyedeekay: No I missed them
(03:23:03 PM) zlatinb: no, sorry, pls repost
(03:23:11 PM) zzz: I'll ask a few easy questions:
(03:23:15 PM) zzz: 1) The user base is sufficient that we are confident it's been well-tested?
(03:23:17 PM) zzz: 2) The release process is documented and in theory somebody else could build it if they had the right keys?
(03:23:20 PM) zzz: 3) We're confident we have the resources to continue maintaining it and doing releases?
(03:23:23 PM) zzz: eot
(03:24:08 PM) zlatinb: 1) - according to matomo it gets 20-25 downloads a day.  We've had users of M1 Macs report that it works as expected.  No negative reports at all
(03:24:28 PM) eyedeekay: 2) and 3) IMO the answers are definitely yes. I can build everything up to the notary part and can teach someone else to
(03:24:53 PM) zlatinb: 2) is yes but requires de-anoning because Apple Inc.
(03:24:54 PM) eyedeekay: So with zlatinb and I there are at least 2 people who know how to build it
(03:25:08 PM) zzz: 2) but it is _documented_ ?
(03:25:33 PM) zlatinb: yes, step by step incl. the notary process in the README.md
(03:25:41 PM) zzz: super
(03:26:16 PM) zzz: 1) does "working" include the update process? have we seen the torrent swarm actually fire off when the news updates?
(03:26:29 PM) zlatinb: http://git.idk.i2p/i2p-hackers/i2p-jpackage-mac
(03:26:43 PM) zzz: (because I think the windows version hasn't actually been seen to work yet)
(03:27:26 PM) zlatinb: 1) yes, the swarm picks up, it's quite small but it's there.  I've seen it both at the 1.5.1 and 1.6.1 releases
(03:27:49 PM) zzz: ok great, thanks for the answers, no objections
(03:28:14 PM) eyedeekay: Cool I'll make the site changes tonight after I post the meeting logs
(03:28:26 PM) zlatinb: awesome thanks!
(03:29:05 PM) eyedeekay: Re: Windows, yes but appears it's an issue in the config file I shipped and not the updater itself
(03:29:05 PM) eyedeekay: So Windows needs 1 more manual update to work with automatic updates, which I think I might have known ahead of time but forgot
(03:30:07 PM) eyedeekay: 4. NetDB research (Sim/Model) http://zzz.i2p/topics/3227-netdb-simulator-proposal
(03:30:44 PM) eyedeekay: Also zlatinb's topic, let us know what you're thinking
(03:31:28 PM) zlatinb: So, in the last two weeks or so there has been a noticeable deterioration in the accessibility of sites like zzz.i2p, postman, event major.acetone.i2p... basically sites I know are up
(03:31:57 PM) zlatinb: I've done some debugging and had situations where the LS of zzz.i2p gets fetched fine, but then the RI of the IBGW must be failing to fetch
(03:32:03 PM) zlatinb: (I'm speculating here ofc)
(03:32:47 PM) zlatinb: in a nutshell, there are many unknowns in the functioning of the NetDB and we have to place our faith in the Kad algorithm, which is fine, but I would rather have more confidence via some research
(03:33:33 PM) zlatinb: just like the testnet helped improve the streaming and ssu and ntcp code a lot, I believe a model or a sim can help improve the functioning of the netDB
(03:34:44 PM) zlatinb: and the reason I want to discuss it here is that if we go with a sim that runs the actual code that will be in the router, some refactoring will be necessary of the NetDB code, possibly some of the *Job classses too
(03:34:52 PM) zlatinb: Note, I haven't looked into the multi-router jvm and JVMCommSystem yet
(03:35:03 PM) zlatinb: maybe those can be an answer up to a given number of nodes
(03:35:37 PM) zlatinb: but my goal is to simulate at least a few thousand nodes, which means abstracting away most of the router components
(03:35:42 PM) zlatinb: eot-checkpoint
(03:36:28 PM) major left the room (quit: Ping timeout: 320 seconds).
(03:36:32 PM) zzz: I'm lukewarm at best, due to how hard it would be, and how long before we got any useful results
(03:37:20 PM) zzz: for starters, I made a series of changes recently to make things more reliable
(03:37:27 PM) zzz: in 1.6.1-6 dev build
(03:37:46 PM) zzz: but I don't know if any of the popular servers (besides mine) are running it
(03:38:17 PM) zzz: I would like to see echelon's and eyedeekay's servers try it. Those are two batches that I've had trouble with for months
(03:38:20 PM) zzz: eot
(03:38:44 PM) eyedeekay: Most of my servers just track the PPA, I'll get git.idk.i2p updated
(03:39:17 PM) zzz: git and gitssh are two that frequently shun me. If they get better, I'll notice
(03:40:07 PM) zzz: echelon's servers are another mystery. they seem to have IP-level issues frequently
(03:42:11 PM) eyedeekay: I can do that soon then
(03:43:37 PM) eyedeekay: My questions about the netDB test harness are probably ill-conceived, for instance, does the existing testnet have the ability to simulate a netDB between the routers within it?
(03:44:08 PM) eyedeekay: And if so, is there a way to use it?
(03:44:44 PM) The account has disconnected and you are no longer in this chat. You will automatically rejoin the chat when the account reconnects.
(03:49:14 PM) The topic for #i2p-dev is: 1.6.1-7 | next dev meeting 8 PM UTC Tuesday, January 4, 2022
(03:49:14 PM) Topic for #i2p-dev set by zzz at 09:08:14 AM on 01/03/2022
(03:49:20 PM) zzz: gosh no not pspice. it was some sort of queueing/behavioral network model
(03:50:31 PM) zzz: EOT and out, gotta run, thanks everybody
(03:50:46 PM) eyedeekay: Later zzz
(03:51:41 PM) eyedeekay: Anything else on 4? zlatinb
(03:51:50 PM) zlatinb: no, that's all
(03:52:08 PM) eyedeekay: All right that covers it for the meeting then, have a good week everybody