Hugo Landau

Alright, let's just make it do an actual authorization. Since there's no need for people to manually renew authorizations, I'd like to keep it named challenge-test or authorization-test or similar.

Deferring this until I figure out a better way to get the FQDN than executing `hostname -f`.

Still good.

So long as this simpler solution can work with the server using our existing callback options to select a certificate dynamically, with the server not necessarily knowing the set of...

Updated.

Alternatives to the current API: 1. Add an `int *err` argument to the function 2. Use `errno` 3. Use `get/set_last_sys_error` (which is either `errno` or `Get/SetLastError`) 4. Use `get_last_socket_error` (which...

I've updated this to have a new `int *err` parameter. I think this is looking a lot better now. The semantics of the return value are also a lot clearer...

So basically we have a situation where: - We have accidentally exposed side effects of internal implementation details (errno); - Applications have ended up depending on this abstraction leak, in...

(But leaving all of the above aside, since `int *err` provides a system error code, if you want to have `libssl`'s calls to `BIO_readmmsg` work the same way, you would...

That's some interesting history... So basically the question for now is whether `errno = err;` is in `BIO_sendmmsg` or `libssl`. Since it's just one line, it doesn't make a massive...