database64128
database64128
Made some changes: - Removed the EIH section. - Added EIH spec in a separate document.
## ⚠️ Major breaking change To have better defense against active probes, we made a breaking protocol change. Part of the header, specifically type and timestamp, now sits in the...
@yzou The response stream uses a new salt.
@FranzKafkaYu It's not a new URL scheme. The plain `method:password` format without the extra base64url encoding has existed for a long time, and is well supported by many clients, including...
> why is it necessary It's not necessary, but it's easier and much more efficient this way. With one UDP socket per session, your downlink (server -> client) relay loop...
> all client session source ports change, and return packets will fail until a new packet is received from the client. How can UDP sessions survive client network changes? You...
> No, in the spec, it'll open a new relay session. That sentence you referenced is describing how clients handle incoming UDP packets from local clients, not servers. Server behaviors...
> I know, but it is only in the TCP section, and not mentioned in the UDP section. Appreciate the feedback. Made some changes: https://github.com/Shadowsocks-NET/shadowsocks-specs/commit/9c29d3ea529980cde6990d78d9ad8b03c5784b50 > it's repeated nonce, not...
> In TCP Detection Prevention section, what does the "receive buffer" stand for, the kernel or the application buffer? It's the socket's (in-kernel) receive buffer. > If the server closes...
> Is current spec stable? Yes, it was declared stable on June 12 2022 when I merged Shadowsocks-NET/shadowsocks-specs#2, and has been available in many popular proxy tools for about a...