Jacob Hoffman-Andrews

The mix-and-match nature of the CryptoProvider API leads to this potentially surprising bit of code, if both `ring` and `aws_lc_rs` are configured on: ```rust use rustls::crypto::aws_lc_rs::cipher_suite::TLS13_CHACHA20_POLY1305_SHA256; use rustls::crypto::aws_lc_rs::kx_group::X25519; use rustls::crypto::ring::RING;...

Mostly yes. I understand the mix-and-match nature, but I think it would still be useful to have a list of what needs implementing for the use case "I want to...

I agree we should keep the name matching the spec, and also renaming it would be a semver-breaking change, which it would be best to avoid. However, I think we...

There's discussion at https://github.com/rustls/rustls/issues/469#issuecomment-812601706 about what we would want from a dependency that implements an LRU cache, if we find that performance merits one.

Nice work, thanks for doing this! A couple of pieces of feedback: - For a realistic workload, matching the number of threads to the number of cores is important. You...

@aochagavia Thanks for all your work on this! I agree it's useful to keep the benchmarking code around for future measurements but it doesn't need to be upstreamed into the...

This is not currently possible, to my knowledge. To do this I think we'd use the [read_vectored](https://doc.rust-lang.org/std/io/trait.Read.html#method.read_vectored) method of `io::Read`. There's already some support for [write_vectored](https://doc.rust-lang.org/std/io/trait.Write.html#method.write_vectored).

The current status quo: rustls offers the [ClientCertVerifier](https://docs.rs/rustls/0.20.8/rustls/server/trait.ClientCertVerifier.html) trait, which takes as input [`rustls::Certificate`](https://docs.rs/rustls/0.20.8/rustls/struct.Certificate.html), which is a newtype around `Vec` - an unparsed, raw certificate. It is up to implementors...

#1145 implemented the new client session behavior by default, and added a new client persistence interface. After some discussion on #1228 we decided that the Tls1[23]ClientSessionValue types should not be...

That's correct.