Tim Robinson

Yes, this is due to the `ssl-native-tls` and `ssl-openssl` features being an either/or selection. Maybe `default-features = false` should deactivate SSL completely.

I think the websocket-lite benchmark should be ok. You should expect zero memory allocations from the profiler loop (that is, after the initial client connect). When developing websocket clients I...

This is an interesting result, thanks for putting together the benchmark. The initial focus of websocket-lite was to have zero memory allocations after initialisation. Allocations were inherent in the tungstenite...

I think the timings are close because they're dominated by the network itself, even on localhost; that is, the `recv()` call shows up in the profiler as the hottest function....

Thanks! Could you add a test? I think you can make one based on the `reserves_buffer` test: use the client codec to encode a message, then try to decode it...

I think this test triggers the size limit logic correctly: ```rust #[test] fn applies_message_size_limit() { let message = Message::binary( std::iter::repeat(b"abcdefghijklmnopqrstuvwxyz") .take(100) .flat_map(|b| b.iter().copied()) .collect::(), ); let mut data = BytesMut::new();...

Take a look at #205. I think it does the job, it's not merged yet as it was missing tests.

Sounds good to me, thanks

Thanks for reporting this, I'll take a look and sort it out

In the `foo` repro above I don't see why cargo is choosing tokio-util 0.6 for websocket-codec and 0.7 for hyper-websocket-lite, but from reading https://doc.rust-lang.org/cargo/reference/resolver.html it looks like it's allowed to...