My biggest gripe with the Tor project is that it is so slow.
I don't think merely moving to Rust makes Tor faster either. And I am also not entirely convinced that Rust is really better than C.
My biggest gripe with the Tor project is that it is so slow.
I don't think merely moving to Rust makes Tor faster either. And I am also not entirely convinced that Rust is really better than C.
Client: "Hi, some.web.site.example please, I want to talk HTTP and I assume you know how AES works and I've randomly picked these numbers to agree the AES key"
Server: "Hi, I do know AES and I've picked these other numbers so now we're good."
Included in the very same packet as that response from the server is the (now AES encrypted) first things the TLS server wants to say e.g. to prove who it is, and agree that it knows HTTP as well.
0RT is a (very dangerous, do not use unless you understand exactly what you're doing) extension for some niche applications where we can safely skip even this roundtrip, also included in TLS 1.3
This TLS handshake can only happen after the TCP handshake, right? So 1 rtt for TCP, + 1 rtt for TLS. 2 rtt total. (2.5 rtt for the server to start receiving actual data. 3 rtt for the client to receive the actual response.)
QUIC really wants to do discovery to figure out a better way to move the data and of course Tor doesn't want discovery that's the whole point, so these features are in tension, but that's not hard to resolve in Tor's favour from what I can see.