Subject: Re: Changing root's shell to /bin/sh
To: dustin sallings <email@example.com>
From: Thor Lancelot Simon <firstname.lastname@example.org>
Date: 03/17/1999 11:46:26
On Tue, Mar 16, 1999 at 09:23:37PM -0800, dustin sallings wrote:
> On Tue, 16 Mar 1999 email@example.com wrote:
> // Personally, I continue to *much* prefer "BSD Authentication". We
> // can get source, and it has *significant* advantages. (e.g., a
> // non-suid program can do a setuid authentication scheme - if you want
> // to let it. You don't have to, but you *can.*) It also comes with
> // free support for the magic needed to let root have a different
> // datasize limit than other users, and some of those users get even
> // smaller limits, etc.
> The last time this came up, someone pointed out that it's very
> easy to emulate BSD auth via PAM, but not vice versa. IMO, PAM should be
> implemented with a BSD auth module. Then you can choose to do BSD auth
> via the PAM config. They both look pretty good, but many applications are
> PAM aware these days.
I don't think a whole lot of people disagree with you. But there's clearly
a point where it's senseless to discuss much more unless somebody's going
to write the code! If you want PAM so much -- write it! You may end up
having to revise it somewhat, but certainly it's not just going to get
What _do_ tend to get ignored are people saying "_X_ should be implemented."
"_Y_ should be implemented." where nobody steps up to the plate to implement
either. This is a volunteer project, after all!