Iroh is a library for building on direct connections between devices, putting more control in the hands of your users.
How does 0-RTT work in #iroh (and #QUIC, and a little bit of #TLS), a fun exploration: https://www.iroh.computer/blog/0rtt-api
What's even better is that @Frando already landed a PR in #rustls to improve the UX of this (https://github.com/rustls/rustls/pull/2476). Should also land in #Quinn and #iroh soon!
It had to happen: in-chat multi-player #Quake III Arena #webxdc app, running over @n0iroh and using the fine work of https://ioquake3.org/ folks. Quake3 app is 1.2MB but you need to download once a game file of ~50 MB. It's only one level.
Realtime-P2P-networked multiplayer gaming arranged through securely end-to-end encrypted email, from a group chat!
Both Quake3's in-game chat as well as game play messages have forward-secrecy. What a weird future, eh? :)
iroh release 0.35 - prepping for 1.0
- 0.35 is the last release before we move to our release candidate - woo!
- The rc, which we will refer to as 0.99, will have protocol-level breaking changes
- 0.35 will be maintained as we develop 0.99
For details please read the blog post https://www.iroh.computer/blog/iroh-0-35-prepping-for-1-0
And a follow-up blog post about blake 3 internals!
Patch release 0.34.1 for you this week:
* iroh has had some bug fixes and internal polishing
* iroh-blobs now relies on the blake3 crate, removing the need for the dependency on iroh-blake3 fork
* iroh-net-report has been deprecated. The APIs are too closely tied to iroh right now for it to be useful as a general purpose tool. We have pulled it back into iroh to help improve iteration speed and to limit leaking unnecessary structs into the public iroh API.
There is also https://github.com/n0-computer/iroh-willow which is under construction and based on the #Willow protocol. This is also a foundational #p2p technology, which received funding by @nlnet under the EU Horizons #NGI0 programs, to evolve their specifications.
Willow contains among others a specification for support of object capabilities, called Meadowcap, which is very interesting.
https://willowprotocol.org/specs/meadowcap/index.html#meadowcap
A few days late to announce, nevertheless:
iroh@0.34.0 - Raw Public Keys in TLS certificates!
iroh now defaults to using raw public keys in TLS certificates
this is a breaking change and if you need to communicate with endpoints running on older versions of iroh you can using the new EndpointBuilder::tls_x509 method when building the endpoint
the iroh-blobs tag API got a facelift
LocalSwarmDiscovery has been renamed to MdnsDiscovery
For details checkout https://www.iroh.computer/blog/iroh-0-34-0-raw-public-keys
@beaufils Not everyone comes from the same background and has the same strong feelings about the same things. Feel free to communicate with us elsewhere, OTOH I'm sure there are more things to dislike about iroh/number0 (or anyone else). Why are we still on github? Why are we a company? So many tradeoffs to be made everywhere. I'm sorry if we're crossing red lines for you, I guess you don't have to use us. If you can find common ground though, we're still happy to have you.
@beaufils We still have a twitter account? Embarrassing.
Some of us prefer the fediverse :)
And yeah, discord is also not my favourite... we do have a matrix channel on the Next Internet server, but no one seems to use it (but than again, matrix is also not my favourite)
Ever used #sendme to send files around using #iroh? It is built on top of the iroh-blobs protocol, which uses blake3 to do verified streaming. If you do this at scale you also want a store for the blobs, and that's not so trivial to build: https://www.iroh.computer/blog/blob-store-design-challenges
@Junicast @kubikpixel Yes, building magic wormhole with iroh direct connections would be cool! Currently magic wormhole traffic is all relayed via a server AFAIK.
Iroh is open source, and you can self-host a relay. And iroh runs on mobiles. E.g. @delta does all of those things and integrates an #iroh relay in their #ChatMail server. And #DeltaChat runs on mobiles.
Our users are the best, improving the ecosystem for everyone. Check out this gnarly bug somehow triggering for this one use case but never seen elsewhere: https://github.com/conradludgate/futures-buffered/issues/10
Currently playing around with https://github.com/MightyShadow702/v86Emu-webxdc which leverages https://github.com/copy/v86 into DeltaChat. Effectively allowing you to run a 32 bit .iso file inside DeltaChat as a webxdc application.
Right now tcp/ip is wired up with realtime communication over iroh 💥 ðŸ§
And because we couldn't help ourselves: We've also added browser support to one of our protocols, iroh-gossip!
Let's have some fun and jump into an end-to-end encrypted, gossip-based chat together:
https://n0-computer.github.io/iroh-examples/main/browser-chat/index.html?ticket=chato2j7dtcqmvzndvfmzs4c5sa2gfkj4pjeiibqda2uetnsjzoemplqcykrxdrdplzcpurlko3qesbqbpd7xbt52rxir5rm2zg4v3qqvvt6
iroh@0.33.0 - Browsers and Discovery and 0-RTT, oh my!
- iroh can now be compiled into Wasm and used in the browser! We also have some fun examples linked in the blog post, please check them out.
- Discovery now allows for publishing and resolving user-defined data!
- Use 0-RTT connections for one less round trip on connections to endpoints you have already connected to!
For more details checkout https://www.iroh.computer/blog/iroh-0-33-0-browsers-and-discovery-and-0-RTT-oh-my
QUIC implementations need to keep hash tables indexed by the Initial CID chosen by the peer, or by the "reset key" chosen by the peer. These are normally long random numbers, I picked a simple hash function for these tables in picoquic. I was wrong, the keys are really "chosen by the peer", and thus malicious peers could DOS servers with hash collision attacks. I had to switch to using SipHash, a tiny bit slower but much safer.
https://github.com/ncc-pbottine/QUIC-Hash-Dos-Advisory
@smallcircles Ah, thanks for the link tip. Should be deploying a fix.
We use wasm for the browser (https://www.iroh.computer/blog/iroh-0-32-0-browser-alpha-qad-and-n0-future). For use everywhere I suspect it would still be a very bumpy road and probably at least a similar amount of work. You'd need WASI probably at least for real p2p, but that's just my guess. Having native code for p2p networking is still a great advantage currently IIUC.
