[561] in Kerberos_V5_Development
Of krb5_rd_req and men....
tytso@ATHENA.MIT.EDU (tytso@ATHENA.MIT.EDU)
Mon Jan 7 18:46:37 1991
The API doc has the following to say about the argument sender_addr in a
call to krb5_rd_req:
krb5_address *sender_addr
sender_addr specifies the address(es) expected to be present in
the ticket.
It shouldn't be "address(es)", since you can only pass in one address.
This of course brings up an interesting question, since the routine
krb5_os_localaddr() fills in an array of krb5_addresses: namely, what
should the applications programmer pass in to krb5_rd_req?
Should the programmer be required to try each address, one at a time,
and stop when he doesn't get an error return from krb5_rd_req?
Should the programmer blithely assume that the first address will work?
Or should the programmer wimp out (following the lead of the example
we've provided in telnetd) and just pass 0 to sender_addr, which skips
the address checking step altogether?
