Subject: Re: HEADS UP: migration to fully dynamic linked "base" system
To: Johnny Billquist <bqt@update.uu.se>
From: Robert Elz <kre@munnari.OZ.AU>
List: current-users
Date: 08/27/2002 21:01:40
    Date:        Tue, 27 Aug 2002 15:21:55 +0200 (CEST)
    From:        Johnny Billquist <bqt@update.uu.se>
    Message-ID:  <Pine.LNX.4.21.0208271512380.4775-100000@Tempo.Update.UU.SE>

  | I don't think /rescue is such a brilliant solution as you obviously have
  | decided it is.

It doesn't need to be brilliant, just functional enough to be able
to fix many of the problems that can occur (obviously it can't handle
everything, a "newfs /dev/wd0a" (/) isn't likely to get undone that way.

In what way do you see it as being unable to handle what it needs to do?

  | I think that it is a good thing that programs in /bin and /sbin be
  | runnable with no dependencies on other stuff.
  | That's my reason, and I've said it a number of times.

That isn't a reason, it is an opinion.   The reason would be why that
is good thing.   How do you benefit from that?   Or what do you lose
if it isn't true?

  | Eh? Not having a /lib is hardly recent. Or do you consider 20 years ago
  | recent?

20 years ago was about 4.2bsd -- /lib still existed.
/lib still existed in 4.3bsd, and in the myriad 4.3xyz releases.
I didn't ever run a genuine pure 4.4, so I'm not sure whether it
had gone from there or not, but that's much more recent than 20 years.

But in any case, how recent it was doesn't matter, the fact is that
/lib used to be part of unix, before /usr/lib existed, and it can't
possibly be wrong to have it come back again.

There are arguments in favour if /usr/lib (which is why everything
ended up moving there) but they apply to almost no NetBSD users,
which is most likely why you're not making them.

  | (And I assume you mean "dynamic linking", since static is what we've
  | always had.)

Yes... ooops - slip of the brain!

  | This all started with a request that those who don't want this change be
  | gived a way of not having it.

What I have read has suggested that something like that might happen
(though you'd have to compile for yourself to get it, what comes in the
distributions won't be like that).

But before any effort gets put into supporting requests like this, it would
help to know if there's some kind of rational reason behind it, rather than
just "it isn't what I am used to", or "someone I once knew had heard about
some other system where the rumors were there was a problem because of
something that sounded a bit like ..."

There are obvious advantages to be had here, supporting the status quo
as an alternative, just because it has been (for a while) the status quo,
isn't sensible.

kre