4 ms·
Seems to be using WebRTC, so presumably it's using STUN servers to do the NAT hole-punching. So not really "no server involvement". But I suppose that would be
by kelnos 5y ago
Seems to be using WebRTC, so presumably it's using STUN servers to do the NAT hole-punching. So not really "no server involvement". But I suppose that would be basically impossible on today's internet.
Hopefully it's not also using TURN servers if STUN fails...
- capableweb 5y agoNot even, the application is making GET, POST and PUT requests to send the file to a /link endpoint which also the receiving side is hitting. How is this even close to being "no server involvement"? It's misleading, because the server is very much involved here. > But I suppose that would be basically impossible on today's internet. P2P is possible on the internet (100% without centralized servers), but it's not possible without STUN/TURN on the web.
- iCarrot 5y ago>the application is making GET, POST and PUT requests to send the file Have you looked at the actual payload of the POST/GET/PUT requests? It only contains the WebRTC offers/answers. Actual file transfer is via local WebRTC. The site also claims that it only work for "devices on the same network", so there's no need for any STUN/TURN because there's no wall to punch.
- meheleventyone 5y agoEven on the same network you might need TURN. Browsers use mDNS addresses for local connections rather than leaking internal IPs and if your network doesn’t work with it then a direct connection will fail. You can disable this but it’s not ideal. Getting P2P connections working reliably is an effort in edge case discovery and either fixing it or falling back.
- remram 5y agoIf those POST/GET requests are coordinating the transfer and exchanging metadata, "no server involvement" is still wrong. It's like saying you can connect to a domain name like google.com with no DNS involvement, because you only get the address and then the connection is over HTTP.
- Kiro 5y agoNo-one said anything about the "no server involvement" claim here. What's disputed is capableweb's comment that the app is sending the file itself to the server and not even using WebRTC, which is incorrect. They even say "checking how it works is trivial if you open up the Network inspector" but WebRTC traffic does not show up in there. So much for it being "trivial".
- remram 5y agoYes, they did. kelnos said (4 posts up): > So not really "no server involvement". and capableweb said (3 posts up): > How is this even close to being "no server involvement"? It's misleading, because the server is very much involved here. No one said the file was being sent through the server, that's your own reading of it. You are entitled to your opinion but don't tell people off about "no one said this" or "what's being discussed", it's just rude.
- fastball 5y agoThis thread is devolving rapidly but: > Not even, the application is making GET, POST and PUT requests to send the file to a /link endpoint Capableweb definitely said the file is being sent to the server. Not sure how much more literally that could be said, in fact. You somehow managed to quote the second line of that comment (some might say "cherry pick") while completely missing the first line where they explicitly said what it was claimed they said. Genuinely not sure if you're being deliberately misleading or have some of the worst reading comprehension I've seen on the internet.
- remram 5y agoYes, in order to send the file, the application is making GET, POST, and PUT requests. He didn't say the file was sent over the requests. We obviously understood this differently. If you and Kiro could just recognize that different people can understand that sentence in different ways, we wouldn't be there, but instead you like to proclaim that "no one is saying anything about" and "what's being disputed is" and other "that other person definitely said" I can see how I'm expecting too much of you. You have a nice day now, though I'm sure it will be full of people not understanding things exactly as you do.
- iCarrot 5y agoYou can inspect the source, there seem to be STUN servers only. "No server involvement" part is probably just regarding the actual file transfer.
- the8472 5y ago> But I suppose that would be basically impossible on today's internet. Bittorrent works just fine here.
- dboreham 5y agoNo it doesn't. (you're relying on someone's public server for it to work).
- npteljes 5y agoOriginally we did, and we still do, as it's way more convenient. But the bittorrent protocol has been improved so that it doesn't need servers anymore. The Distributed Hash Network can handle peer discovery and search queries too.
- input_sh 5y agoYes, that's the point of peer-to-peer communication, you need peers to communicate with.
- spookthesunset 5y agoEven in the land of IPv6 there would have to be something to punch a hole in the end user's firewall rules. Any good end-user router will default deny incoming IP traffic. Assigning public IPv6 addresses to every device in your home just pushes the "deny" rule to a slightly different part of the network stack. We will still require something similar to STUN to get around firewall rules.
- nybble41 5y agoThe only default-deny rule for incoming traffic for general-purpose, up-to-date hosts should be the one built into the OS which rejects incoming connections to ports where nothing is listening. (If you enable UPnP, as most routers do by default, this is the policy you actually have in practice… just with more steps.) If you can get around the firewall rules with something like STUN then those rules aren't really doing you much good, are they? By all means, put random IoT devices on a separate VLAN and block all incoming connections directed there (without UPnP). And be sure to block all the obvious workarounds like STUN for that VLAN or there's no point in having the firewall. But for PCs, smartphones, and the like it should be possible to connect to their open ports from outside the network using their public IP addresses without any tweaks at the router. These devices are perfectly capable of securing their own listening ports.