Joe Birr-Pixton

Yes, this was really fixed. That issue is different -- it is a client-side interop bug. As far as we know, it affected Apple and OpenSSL clients. Apple fixed theirs...

Worth noting that the webpki crate design explicitly vouches that doing this is fast and sensible: > Although it would be less error-prone to combine all these steps into a...

I think as time goes on, my preference to address this is to "bite the bullet" and support async as a first class feature in rustls. That would mean having...

> it allows rustls to be used in e.g. embedded systems without `alloc`. At the moment, operation without `alloc` isn't in scope for this crate. That might change in the...

> maybe the "tls12" feature is good enough. Really, cargo features as a means for detailed conditional compilation doesn't work well. This is quite a mess: ``` $ ack 'feature...

I think we fixed this -- `internals` is now much smaller, and public enums don't appear there.

I have been thinking a bit about how we could support this with the absolute minimum impact on other users. Here's some ideas, starting with three observations: First observation: a...

Apologies, this was more a time-out than a "never". But soon we plan to add lower-level APIs that don't keep an internal per-connection buffer, which should give a more powerful...

> persist::Tls12ClientSessionValue > persist::Tls13ClientSessionValue These are exported as `rustls::client::Tls12ClientSessionValue` and `rustls::client::Tls13ClientSessionValue`: https://github.com/rustls/rustls/blob/main/rustls/src/lib.rs#L424-L426

yeah afaict the only reason `ClientSessionCommon` is pub is so bogo_shim can call `rewind_epoch` on it, and the Deref doesn't work there because it needs mutation. how about instead: https://github.com/rustls/rustls/pull/1232