In article <9510231713.AA20452@defender.vas.viewlogic.com>,
Missing - presumed fed. <greywolf@captech.com> wrote:
> Yes, I know there's a not-completely-trivial amount of work involved in
> this, BUT it would certainly beat having to recompile libc in its entirety,
> and it would mean that we could keep the getpw* routines.
[...]
> Too much stuff out there uses the getpw* routines for us to just toss it
> by the river and let it float away; if we're going to change something
> like this, we need to make wrappers to handle the special cases.
[...]
> Or Did I Miss Something Here? [TM]

Um, yeah... I think so. Nobody is talking about ditching all the getpw*
routines. The worst that's been proposed is to have getpwent() not be
guaranteed to return all users, and to document it as deprecated. (ie,
do the same with the getpw routines what's already be done with the
gethost routines).

But it seems like we needn't even do that, if we do the Ulrix-like
passwd-# thing.