Discussion of krb5_get_init_creds_password() behavior wasRe:problem with the kinit_prompter in kfw 2.5

[John Hascall]

> >Do you have time to work on putting together a patch in the near future?

> "Maybe".  I've been looking at it a bit; changing the prototypes and
> the upper-level code is easy enough.  What to do at the lower-level seems
> to be a bit trickier.

> What do people think about changing krb5int_sendto() to return the address
> of the host it finally got a reply from?  Then it would be easy to match
> that address against one of the master addresses.  If that's okay,
> then I could probably cook that up before I head off to Seoul.

> I was thinking of something like:

> krb5int_sendto(krb5_context ctx, const krb5_data *message,
> 	       const struct addrlist *addrs, krb5_data *reply,
> 	       struct sockaddr *localaddr, socklen_t *localaddrlen,
> 	       struct sockaddr *remoteaddr, socklen_t *remoteaddrlen)
> (The last two arguments are the new ones).

alternatively it could just return 'winning_conn' (the index of
which one of 'const struct addrlist *addrs' worked).

  krb5int_sendto(krb5_context ctx, const krb5_data *message,
  	       const struct addrlist *addrs, krb5_data *reply,
  	       struct sockaddr *localaddr, socklen_t *localaddrlen,
               int * addr_used)

John