question: async_tungstenite

[0xJepsen]

tlsn/notary/server/src/service/axum_websocket.rs

Lines 1 to 5 in 3554db8

	//! The following code is adapted from https://github.com/tokio-rs/axum/blob/axum-v0.7.3/axum/src/extract/ws.rs
	//! where we swapped out tokio_tungstenite (https://docs.rs/tokio-tungstenite/latest/tokio_tungstenite/)
	//! with async_tungstenite (https://docs.rs/async-tungstenite/latest/async_tungstenite/) so that we can use
	//! ws_stream_tungstenite (https://docs.rs/ws_stream_tungstenite/latest/ws_stream_tungstenite/index.html)
	//! to get AsyncRead and AsyncWrite implemented for the WebSocket. Any other modification is commented with the prefix "NOTARY_MODIFICATION:"

I am curious if would be better to use this fork of ws_stream_tungstenite and use tokio_tungstenite all the way up. The reason i am curious in this is that it seams to be the case to me that tokio_tungstenite has matured and become more of the standard and now has the support for the async properties that async_tungstenite initially added. These now seems to make a fully stale toolchain which ws_stream_tungstenite is a part of.

[sinui0]

I'm for switching to tokio_tungstenite 👍 We would need your fork to be on crates.io before switching, happy to accept PR afterwards. I do wonder if there is already a good alternative to ws_stream_tungstenite built on tokio_tungstenite? I feel using it as a duplex bytestream must be common enough.