Subject: Re: HEADS UP: migration to fully dynamic linked "base" system
To: Greywolf <>
From: Bill Studenmund <>
List: current-users
Date: 08/26/2002 13:04:12
On Mon, 26 Aug 2002, Greywolf wrote:

> On Mon, 26 Aug 2002, Bill Studenmund wrote:
> # Please read the thread on tech-userlevel. There are a lot of good reasons
> # for doing this, all of which involve now being able to dlopen() libraries.
> Begging pardon, folks, but I've noted that the pattern of implementation
> for things like this seems to be:
> - someone thinks it's a good idea.
> - A discussion gets underway and beaten to death, with no clear decision
>   made.
> - The discussion dies off since nobody can come up with anything new to
>   contribute to it.
> - The ensuing length of silence is mistaken for consent.

I don't think the silence was mistaken for consent. I think the decision
was reached that we should do it, but after 1.6 was out.

> - The good idea gets implemented, and when objections are re-raised,
>   the objectors are instructed to see the archives.

From what I recall, objections come (came) in two flavors: 1) I see this
problem, and 2) I don't like this. For 2), all we can do is decide and go
from there.

For 1), though, I think objections (well issues raised) have been getting
addressed. For instance, we have /rescue/*, so that you can live with a
dynamic init. We are discussing Make variables so you can get back the
/bin and /sbin static flavor.

> This is kind of how /etc/rc.d came into being, and although I don't disagree
> with its objective, I'm not terribly fond of the method by which it was
> "agreed upon".
> I still think that only /sbin/init is going to really require this change,
> and there needs to be a keyword ("traditional"?  "static-root"?) such that
> we can continue to build root static and /usr dynamic if we want to.

What requires this change depends on what you see as the gain. If all it
is is to get password modules into init, then yes, only init needs it. But
I can think of other /{,s}bin programs that may need password/alternate
user info modules.

However from that thread, one of the other advantages was that we could
better support different locales. i.e. all of the international character
set support. I can't think of a program in /{,s}bin that would NOT use
that, so there is a definite win for making it all dynamic.

Take care,