Maximilian Hils

@aaron-tan: I think I don't quite understand your question, could you clarify what you mean?

Ah, now I get it. Thanks. Either is totally fine IMO, whatever is more natural to implement. :)

Just from the logs it seems like GnuTLS is really adamant on us properly shutting down the TLS connection and not just sending a TCP FIN. (I haven't tested this...

There shouldn't exist an `unwrap()` method on the tls object. Is this just working because it actually errors now? 😅

> Nope it's a success, and as far as I can see the method is there. 😆 That would be very surprising - we use PyOpenSSL, and its `OpenSSL.SSL.Connection` does...

I suspect it's some OS/networking difference? Most stacks make no distinction on whether a TLS connection is properly shutdown or whether the underlying TCP socket is shutdown (or reset). GnuTLS...

I think I would be happy to merge a PR that adds this to the module index page!

Ubuntu users can very easily install recent Python versions: ```shell sudo add-apt-repository ppa:deadsnakes/ppa sudo apt-get update sudo apt-get install python3.10 ```

This looks fantastic now! Mostly nits at https://github.com/meitinger/mitmproxy/compare/quic...mhils:mitmproxy:pr-5435. The one architectural change I did was that I replaced `set/get_connection_error` with `ConnectionClosed/CloseConnection` subclasses (and also moved TCP half closes out of...

Tests look very good so far. 👍 😃 🍰