Add muxer compliance test suite and refactor tests to not depend on libp2p-swarm

[thomaseizinger]

Description

The first part is likely to not be controversial.
Migrating away from libp2p-swarm avoids circular dependencies across our workspace. libp2p-quic is a transport and I'd argue that if the implementation only depends on libp2p-core, we should also test it outside of a Swarm context.

For example, this brings up the question on what to do with the wrong_peer_id test. Previously, this was testing functionality that is in libp2p-swarm.

Links to any relevant issues

Open Questions

Change checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• A changelog entry has been made in the appropriate crates

[mxinden]

This looks good to me. Thanks @thomaseizinger.

Great to see the muxer tests used here.

I suggest waiting for @elenaf9's approval.

[kpp]

For example, this brings up the question on what to do with the wrong_peer_id test. Previously, this was testing functionality that is in libp2p-swarm.

There was an intention to test TLS handshakes: you can't setup a connection if you dial with one PeerID and you get another in response.

[thomaseizinger]

For example, this brings up the question on what to do with the wrong_peer_id test. Previously, this was testing functionality that is in libp2p-swarm.

There was an intention to test TLS handshakes: you can't setup a connection if you dial with one PeerID and you get another in response.

Fair enough. But to actually test this on the TLS level, we need to pass in the expected peer ID as part of creating the config:

https://github.com/libp2p/rust-libp2p/blob/f9b4af3d9d5b12b20756e496349b0866baa862da/transports/tls/src/lib.rs#L44

But we currently pass None here:

rust-libp2p/transports/quic/src/endpoint.rs

Line 82 in 58c3f67

let client_tls_config = Arc::new(libp2p_tls::make_client_config(keypair, None).unwrap());

The code will have to undergo some refactoring where we don't reuse the same config for all connections so we can pass the expected PeerId in here.

I'd suggest we do this as a follow-up. The current test only validates logic that is present in libp2p-swarm so we are not losing anything but not running this test atm.

[elenaf9]

Sorry for the delay. Thanks @thomaseizinger!

The concurrent_connections_and_streams_async_std test runs forever when I run it locally. Any idea why that could be the case @thomaseizinger?

[thomaseizinger]

I don't know why GitHub is stupid and doesn't let me re-request a review from both of you @elenaf9 @kpp. In any case, I'd appreciate another review :)

[elenaf9]

Thanks @thomaseizinger!