Ty Sarna writes:
> In article <22926.879103301@planix.com> you write:
> > yppasswd (yp_passwd.c) uses "passwd.byname"  to retrieve the user's account
> , 
> > which does not contain the passwd and it therefore always fails when 
> > verifying the users old password. Should yp_passwd.c be changed to use 
> > "master.passwd.byname" throughout in place of "passwd.byname"? In that case
> , 
> > what happens if the yp servers is a SunOS machine, where passwd.byname will
>  
> > have the passwords? Or, am I just confused?? 
> 
> Sounds like you are using the INSECURE=no setting. You can only use this
> if *all* the yp clients support this option too, and apparently (looking
> at the libc source) NetBSD's doesn't't yet. Also, our getgrouplist() doesn't
> know how to use the netid stuff, and will have to walk the whole group
> map :-(
> 
> I'm guessing it's kind of late for this tsuff to go into 1.3 :-(

unfortunately, yes, it's probably too late.

chuck cranor and I have been finding and fixing various bugs in ypserv 
and associated processes; these fixes will be in 1.3.

however, the libc stuff is, strictly speaking, a `feature', and
therefore it's probably too late to `pull up' (and the fact that i
have to implement it; but that'll only be a few hours work using
the original source base (freebsd, opnebsd) as a reference).

i'm putting an XXX note in the Makefile.yp saying that even tho
you can build the master.passwd maps, we don't use it ourselves yet.