From owner-freebsd-current  Wed Dec  4 09:48:01 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id JAA09465
          for current-outgoing; Wed, 4 Dec 1996 09:48:01 -0800 (PST)
Received: from diablo.ppp.de (diablo.ppp.de [193.141.101.34])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id JAA09427
          for <FreeBSD-current@FreeBSD.ORG>; Wed, 4 Dec 1996 09:47:56 -0800 (PST)
From: Greg Lehey <grog@lemis.de>
Received: from freebie.lemis.de by diablo.ppp.de with smtp
	(Smail3.1.28.1 #1) id m0vVLQB-000QrKC; Wed, 4 Dec 96 18:47 MET
Received: (grog@localhost) by freebie.lemis.de (8.8.3/8.6.12) 
       id SAA19269 for FreeBSD-current@FreeBSD.ORG; Wed, 4 Dec 1996 18:47:45 +0100 (MET)
Organisation: LEMIS, Schellnhausen 2, 36325 Feldatal, Germany
Phone: +49-6637-919123
Fax:   +49-6637-919122
Message-Id: <199612041747.SAA19269@freebie.lemis.de>
Subject: Installation: still not perfect
To: FreeBSD-current@FreeBSD.ORG (FreeBSD current users)
Date: Wed, 4 Dec 1996 18:47:45 +0100 (MET)
X-Mailer: ELM [version 2.4ME+ PL28 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

In the last couple of days I have reverted to the position of beginner
fighting the FreeBSD installation process.  It's been a frustrating
but instructive time, and it's brought out a number of points I'd like
to discuss.

As the subject says, I think there's a lot that could be improved.  I
also suspect that there are a number of things we say about the
installation process that are just plain wrong.  Here's a brief
overview:

1.  I am doing the installation on a notebook (AcerNote Light, if
    anybody's interested; 75 MHz Pentium, 24 MB main memory in my
    config, 528 MB hard disk, 2 PCMCIA slots, in one of which I have a
    3Com 3C585C).  It also sported a sticker proclaiming "Designed for
    Microsoft Windows 95%", which I found convenient to put on my
    waste paper basket.

2.  This machine doesn't have a CD-ROM, so I had intended to install
    via the Ethernet board.  The boot kernel recognizes the Ethernet
    board, thinks it initializes it, but nothing comes out.  This
    appears to be due to the fact that it doesn't have PCMCIA
    support.  If this is the case, it would be nice (a) for the driver
    to notice the fact and not pretend that everything's working, and
    (b) to have a boot diskette which does support notebooks.

3.  It may be that the Ethernet board is defective.  It's difficult to
    be sure: 3Com have designed what must be a superlative in their
    line of decreasingly useful diagnostic software.  I returned one
    Ethernet board after the diagnostics said it was bad, but the
    other one behaves just the same: if I write the config info first,
    the diag software passes, but if I then stop the program and
    re-enter, it fails again (can't find the registers).  Has anybody
    else seen this?

4.  So, I thought, let's try using lp0 as the interface.  Doesn't
    work: it looks as if the setup scripts aren't expecting a
    point-to-point connection, and they don't specify the address of
    the other end of the link, so ifconfig fails.  Is there a trick
    here?

5.  OK, let's do a floppy install of the minimum necessary.  That
    doesn't work either, although I stuck to the letter of what the
    installation scripts say.  Looking at the debug output under F2,
    it shows that it succeeded in mounting the disks (I tried both
    MS-DOG and ufs formats) on /dist, and then went looking for
    /dist/bin/bin.tgz and /dist/bin/bin.inf, which it didn't find.
    There's nothing in any documentation I know about which even
    mentions these files.

6.  Finally, I made my own tar floppies and copied them across (after
    first finding that there's no stand/tar, so I had to copy that
    across with cpio).  That's a multivolume tar archive, so it has to
    be uncompressed, and I'm currently on diskette 25.  There's a
    *lot* of stuff in there which I wouldn't consider essential to a
    minimal system, including the C++ compiler, groff fonts, and
    perl.  Wouldn't it be possible to trim this down a bit?

Of course, it could be that I have tripped over something non-obvious
(at least to me).  If that's the case, it will happen to others as
well.  There are also a couple of assumptions I have made, some of
which are probably wrong.  Please let me know if I've screwed up, and
I'll document it for the next edition of The Book.

Greg