On Jan 25, 2023, at 7:55 AM, Eliot Lear <[email protected]> wrote:
> If you want to add some implementation guidance I think that's fine, but this 
> isn't really a protocol consideration.  I could envision the following 
> circumstance, by the way: a username triggers the server to initiate a new 
> inner method, and the password is not sent, knowing this.  Or the password is 
> ignored.

  The auth flow is:

server requests password auth via Basic-Password-Auth-Req TLV, which contains 
nothing other than a prompt

supplicant sends a password auth via Basic-Password-Auth-Resp TLV, which 
contains a username && password.

  The server can choose to reject (or ignore) the password, of course.

  What I'm not clear on is how they would negotiate a special username which 
triggers a new auth, but without doing a password check.  That seems odd to me.

  If the supplicant doesn't want to do password auth, it can just reject the 
Basic-Password-Auth-Req TLV, and return a NAK TLV.  It shouldn't send a 
username && empty password.

  Alan DeKok.

_______________________________________________
Emu mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/emu

Reply via email to