Support keyboard-interactive authentication #116
Conversation
wez
force-pushed
the
kbdauth
branch
5 times, most recently
from
6571048 to
6d1ca83
Compare
| @@ -248,6 +248,30 @@ pub type LIBSSH2_PASSWD_CHANGEREQ_FUNC = extern "C" fn( | |||
| abstrakt: *mut *mut c_void, | |||
| ); | |||
|
|
|||
| pub type LIBSSH2_USERAUTH_KBDINT_RESPONSE_FUNC = extern "C" fn( | |||
There was a problem hiding this comment.
@alexcrichton here's another libssh2-sys change; this one exposes keyboard interactive authentication types
| .skip_fn(|f| { | ||
| f == "libssh2_userauth_password_ex" | ||
| || f == "libssh2_session_init_ex" | ||
| || f == "libssh2_userauth_keyboard_interactive_ex" |
There was a problem hiding this comment.
@alexcrichton systest seems to not like the callback types, so I added this to the skip list here
There was a problem hiding this comment.
Ok ,that seems fine so long as it's been manually verified
Increase the level of diagnostics in the hope that we learn more about what's happening with the failure on travis: https://travis-ci.com/alexcrichton/ssh2-rs/jobs/221391881
|
@alexcrichton I'm going to merge even though you haven't explicitly approved it because I don't think it is contentious to do so; hope you don't mind! (The CI is actually green; it shows as red because rust-lang/rust#63115 has broken nightly rust, but the stable builds are good) We'll need you to make a release of |
There's magic happening in the systest stuff that I couldn't trace; I couldn't find information on how it generates the C code that it compiles, and it seems to trip over the other existing function types, so I arranged for those that I added here to be skipped too.
For the prompt callback; I chose to implement this using a trait object. This is a combination of flexible and more easily documentable. My use case is embedding this into a terminal emulator and I'd like to be able to associate the callback and prompt information with the relevant portion of the UI, so having that context available is highly desirable.
I've included an integration test that shows that things basically work, although we can't have the test end-to-end successfully use interactive auth because we don't know the password of the user running the test(!)
#65