TCP/TLS implementation

[leninalive]

No description provided.

[w0rng]

Здравствуйте! Зачем усложнять исходный протокол подобной логикой? Кажется, после применения этих изменений поддерживать актуальность этого форка станет сложнее, tcp сделает изначальную реализацию более медленной.

Мне кажется, что реализацию AWG стоит сохранять близкой к оригиналу, а если есть потребность в маскировке трафика, стоит просто выбрать другой протокол.

[pokamest]

@w0rng
It's not replacement of awg, it's attempt to implement another VPN protocol, resistant to blockages in countries with super high censorship level

[w0rng]

@w0rng It's not replacement of awg, it's attempt to implement another VPN protocol, resistant to blockages in countries with super high censorship level

I understand that this is not a replacement for AWG. I don't understand why making the good protocol more complicated. If you need a tls disguise, why not use xray?
For me, the main advantages of WG is its simplicity and speed. This changes, in my humble opinion, can make the awg worse.

I'm not insisting on rejecting this PR, I'm trying to figure out what it's for.

[wwwiretap]

Please consider keeping this feature, there are no good implementations of Wireguard in TCP and UDP is easily blocked. This patch will help many downstream.

A very major Mobile operator in Iran, with over 70 million subscribers is adding very high latency to UDP and effectively breaking the connection, stopping the handshake to make it. They are dropping 50% of the packets. In absence of TCP, which would add some re-transmission logic, Wireguard and AWG are extremely easy to block at mass.

I hope this problem supports accepting this patch upstream. Or one would have to start messing with the core encryption logic of AWG directly which is not preferred. For example, sending multiple handshake packets all at once.