Sebastian Dröge

Can you provide a (non-working) testcase for this, that would make it a bit clearer what you're trying to do. I suspect that it will need a little bit of...

I see, so yes that would require adding a bit of API indeed and it's not going to be too easy unfortunately. The callback goes down deep into tungstenite [here](https://docs.rs/tungstenite/0.11.1/src/tungstenite/handshake/server.rs.html#244)....

Yeah it seems useful to have, I agree :) I don't think I will have time for working on that anytime soon, but I'd be happy to review/help anybody who...

@databasedav https://github.com/sdroege/async-tungstenite/issues/70#issuecomment-748497017 has the basic idea how this could be implemented

Indeed, thanks! Would you be interested to send a PR to untangle that? There could be separate functions per TLS backend, and the functions that are not TLS-backend specific could...

Yeah I like the idea and already saw that crate a few weeks ago. Once it's stable, I think we can move over to it. Back then I didn't see...

Needs someone to investigate in any case :)

Unfortunately the memory layout for tuples is not defined. While in practice e.g. a `(u8, u8)` is the same memory layout as `[u8; 2]` this is in no way guaranteed...

This could be fixed by submitting a Rust RFC that defines that the memory layout of homogeneous tuples is equivalent to the corresponding array. I don't think there's any practical...

The support for primitive arrays from https://github.com/sdroege/byte-slice-cast/pull/18 can kind of help here. Instead of working with e.g. `(f32, f32)` you could work with `[f32; 2]`.