danny at cs.huji.ac.il
Tue May 26 15:26:53 GMT 1998
i keep typyng R, and it get's only to you, and i sometimes forget to
add the cc. will try and be more careful, sorry.
In message <Pine.LNX.3.96.980526150309.800m-100000 at regent.cb1.com>you write:
}On Tue, 26 May 1998, Danny Braniss wrote:
}> }> lib/rpc/server/srv_netlog.c:
}> }> net_login_interactive does is localy
}> }i don't like this. i originally created a separate function which
}> }appears to have disappeared.
}> yes, i remmember: arc4
}> }> net_login_network calls smb_password_check
}> }> password.c:
}> }> smb_password_ok calls smb_password_check
}> as to the API:
}> great stuff, i managed to put all the code in one file, some
}> trivial fixes to passwdb.c and i almost had it, then another fix to
from the top of my head:
dummy_function should be static. (if at all, i think some
compilers don't like empty files ...)
in includes.h there is a line #if !defined(USE_NIS) && (...) then
#defines USE_SMB ...
there should be a cleaner way.
in passwd.c the initialize routines is called, and i assume you
can have only one (NIS or LDAP or SMB or HUJI or ...) then it should be
if then elsed, or if more that one the some changes should be made.
}> then the hunting for the real authetication started. so if
}> the latter could also be api'ed it would be a easy sailing.
}ok, if it's only smb_password_ok() and the other one (not yet created :-)
}i think it would be best if you did some #ifdef work, we get it in to the
}main cvs tree and _then_ go for the password api, following _your_
}experiences on exactly what functions were needed to turn into an api.
}so you are like a prototype developer: then we design the api, then i code
}the api, then you or i convert your system to the new api.
}it should be really simple: two functions only.
ok, as soon as i get the kids to bed (it the wifes day off :-) i'll do
it, test it at work tomorrow (or if i get realy ambitious try it at
home). so you should have a copy sometime 2morrow.
More information about the samba-technical