Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

False and misleading title.

_serverless_ file sharing says title, yet the technology "requires a HTTP server". "Its completely anonymous as no data is ever stored on the server". Incorrect, only mild protection against sniffing and spoofing is provided.

The underlying code documentations speaks of "This makes it perfect for anonymity": https://github.com/ShirsenduK/WhatAreYouDownloading/tree/mas... No PKI is linked or included. Proxy service or Sender/receiver unlinkability is not provided. The used WebRTC technology limits UDP/TCP listen sockets. Browser constraints mean WebRTC offers a severely limited experience. For instance, the state of the art in UDP NAT traversal using the neighbor-invite method (beyond STUN/TURN) is not possible.

Rant Disclaimer: As an academic working on a real deployed zero-server P2P technology for 7 years, this sort of claims are a bit upsetting. Zero-server file sharing systems, with a proven effective spam/pollution prevention mechanism have been proven to be extremely difficult to build. (e.g. no-spam version of Kazaa,Gnutella) See, https://torrentfreak.com/tribler-makes-bittorrent-impossible... Tribler research group created an upcoming IETF Internet Standard on sharing/streaming which features integrated NAT/firewall puncturing (IETF PPSP work). Compliant IETF PPSP implementations are capable of doing HD-quality streaming, both on-demand and live streaming: https://datatracker.ietf.org/doc/draft-ietf-ppsp-peer-protoc...



Sorry for the misleading title. By serverless I meant direct browser to browser file transfer with no server in between. Files are transferred directly. We need the HTTP Server to host the static webpage which facilitates the bridge. After all, its a browser based solution you need a page to visit. :). Services file dropbox can be used to host the static files and everyone can setup their own file transfer service. With services like WebDHT coming up each of these shares can communicate.

Just read about Tribler, it sounds really interesting. All the best with it.

This, Tribler and countless other solutions will make the internet what it was meant to be, a decentralized, fault-tolorent network for information exchange. Thanks!


Indeed interesting demo of browser-to-browser downloading, very light. Anybody can start a Napster-style service: it only needs a webserver+JavaScript (or trust a tamper-free copy). WebDHT is fascinating, is it leeching of a KAD overlay or can they also fully serve all incoming requests? (due to listen socket limit)

WebRTC defines a FTP-like 1-to-1 transfer. Would Bittorrent-like swarming be possible in WebP2P?


Thats something clients will have to implement. Thats is, the JavaScript needs to split files into chunks and download from peers.


does that mean you the http server is like a tracker / UDP hole puncher?


Strictly speaking, PPSP is a swarm transport protocol ("give me that data over IP, UDP or what have you, use any source available"). I mean, "sharing" is a concept unrelated to the TCP/IP stack thus PPSP is not a sharing standard.

For example, it is easy to imagine HTTP over PPSP (instead of TCP) -- which is essentially a serverless Web.


Serverless web... "use any source available".. Wow, feel free to share the code of that with us. Would be quite powerful and draw HN interest IMHO.


By "any source available" I meant Sec 3.10. "Peer Address Exchange". HTTP over PPSP is a weekend project as long as we mean good old static HTML Web. By the way, what is the current public repo for TUD PPSP?


What's "neighbor invite" NAT traversal? This term doesn't resolve to anything on Google.





Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: