Conversation

Notices

1. silverpill (silverpill@mitra.social)'s status on Tuesday, 14-Mar-2023 18:22:10 JST silverpill

   https://geti2p.net/en/blog/post/2023/03/13/new_release_2.2.0

   >As you may have noticed, these release notes and the release itself have been signed by idk, and not zzz. zzz has chosen to step away from the project and his responsibilities are being taken on by other team members. As such, the project is working on replacing the network statistics infrastructure and moving the development forum to i2pforum.i2p. We thank zzz for providing these services for such a long time.

   Anyone knows what happened? Why long-time maintainer suddenly steps away?

   • silverpill (silverpill@mitra.social)'s status on Tuesday, 14-Mar-2023 19:35:46 JST silverpill

     in reply to

     • iced quinnsmas :blobcatsanta:

     @icedquinn Yeah that's understandable. But quitting after nearly 20 years of maintainership, without any prior notice, amidst DDoS attack on the network? While some suspicious organizations (DivaChain) are trying to associate themselves with the project?
     This makes me worry about the future of I2P.

   • iced quinnsmas :blobcatsanta: (icedquinn@blob.cat)'s status on Tuesday, 14-Mar-2023 19:35:47 JST iced quinnsmas :blobcatsanta:

     in reply to
     @silverpill a lot of maintainers have decided to move on after dealing with 2020-2022. a lot of projects (like luxcore, appleseed) have lost all or some coredevs. they're just... tired of working for free, i guess.
   • silverpill (silverpill@mitra.social)'s status on Tuesday, 14-Mar-2023 23:41:17 JST silverpill

     in reply to

     • Jeff Martin

     @cuchaz There is a C++ I2P implementation too: https://i2pd.website/

     Why do you want a network client inside of a browser extension? Maybe WebTransport will make it possible. IPFS has been is experimenting with WebTransport and they say it unlocks p2p networking inside a browser:

     >Enable browser nodes (or light clients) as "full" peers in a decentralized network.

     https://blog.libp2p.io/2022-12-19-libp2p-webtransport/

   • Jeff Martin (cuchaz@gladtech.social)'s status on Tuesday, 14-Mar-2023 23:41:18 JST Jeff Martin

     in reply to

     @silverpill whoa, this is the first I'm hearing of I2P. Honestly, it looks awesome for a future project. Will add it to my research list.

   • Jeff Martin (cuchaz@gladtech.social)'s status on Tuesday, 14-Mar-2023 23:41:18 JST Jeff Martin

     in reply to

     @silverpill Finally actually did some research on these overlay networks. I2P looks great, but installing a Java app so your browser can access the network via local SOCKS proxy? Eeek. Bad UX there.

     Honestly, Tor looks like the best option at the moment, especially with the Tor2Web relays. If I understand this correctly, those would let a totally vanilla browser access the overlay network with almost no configuration. Very nice!

     Now I just wish someone could build a real overlay network client inside of a browser extension. No idea how you'd solve Firewalls and NAT punching there, and you'd have to use websockets (or WebRTC?), but it would be very nice UX for my case, since my app is a browser extension too.

   • silverpill (silverpill@mitra.social)'s status on Thursday, 16-Mar-2023 09:12:37 JST silverpill

     in reply to

     • Jeff Martin

     @cuchaz I'm in favor of native processes. The world of web browsers slowly drifts towards the monopoly of one engine, and I don't want to depend on Google in any way. So ideally the browser should only render UI. In that case p2p tech stack could look like this:

     - P2P networking layer (Tor, I2P or Yggdrasil)
     - App servers. Example: locally-run ActivityPub server that federates over Tor network. I hope app-specific p2p networks will become the thing of the past.
     - App clients (browsers, native apps, CLI tools)

   • Jeff Martin (cuchaz@gladtech.social)'s status on Thursday, 16-Mar-2023 09:12:38 JST Jeff Martin

     in reply to

     @silverpill in case you're vaguely interested in this still, libp2p's support for browsers is a bit lacking, but they're done a lot more work on the rust side of things. I'm experimenting with building a (mostly) P2P network using some rust crates.

     Tragically it looks like you can't get true P2P when all the peers are behind NAT, but having just one server on the public net trying to coordinate punching can get you pretty close.

     Tragically this means I'll have to have my browser extension talk to a native process for now (ugh!), but at least I can use browser-to-native comms instead of SOCKS. Plus, there's hope the browser side of libp2p can evebtually catch up using things like WebRTC and maybe WebTransport too.

     Anyway, thanks for the references!

   • Jeff Martin (cuchaz@gladtech.social)'s status on Thursday, 16-Mar-2023 09:12:39 JST Jeff Martin

     in reply to

     @silverpill Java isn't quite the hurdle, so C++ isn't better there. The hurdle is asking users to install local software to do what looks like "visiting a website".

     I'd like to build an app that has the same UX (or nearly the same) as visiting a website, but the servers are hosted on residential networks rather than in data centers. So I want to stick with vanilla browsers, but I can do as much magic on the server as is necessary. It ends up being a weird mix of classic client/server stuff, but also some newfangled P2P stuff.

     So ideally, I want some kind of public overlay network that can do e.g. ICE, routing without static IPs, and is accessible from a vanilla browser. WebRTC seems promising here except it doesn't solve name-based (instead of IP-based) routing as far as I know. Maybe WebTransport will do it? I'll look into it, thanks!

   • Jeff Martin (cuchaz@gladtech.social)'s status on Thursday, 16-Mar-2023 09:12:39 JST Jeff Martin

     in reply to

     @silverpill also, wow. libp2p looks really great. But perhaps not ready for browsers yet.