*BSD News Article 26158


Return to BSD News archive

Xref: sserve comp.os.386bsd.questions:8045 comp.os.386bsd.misc:1849
Newsgroups: comp.os.386bsd.questions,comp.os.386bsd.misc
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!pipex!uknet!cf-cm!paul
From: paul@myrddin.isl.cf.ac.uk (Paul)
Subject: Re: How to bring up FreeBSD-current ?
Message-ID: <1994Jan18.152113.24605@cm.cf.ac.uk>
Sender: news@cm.cf.ac.uk (Network News System)
Organization: Intelligent Systems Lab, ELSYM, University of Wales, Cardiff
References: <CJr8BC.KLI@latcs1.lat.oz.au>
Date: Tue, 18 Jan 1994 15:21:11 +0000
Lines: 75

In article <CJr8BC.KLI@latcs1.lat.oz.au> wongm@latcs1.lat.oz.au (M.C. Wong) writes:
>Hi,
>  I just ftp'ed the FreeBSD-current src tree from freebsd.cdrom.com a few
>days ago and got it on my 1.0 R. When I unpacked the stuff, I got a new
>Makefile that includes building sharedlib compiler kits. That target
>is called bootstrapld, and is commented as :
>
># You MUST run this the first time you get the new sources to boot strap
># the shared library tools onto you system.  This target should only
># need to be run once on a system.
>
>  Since there wasn't any other docs on preparing the system to current,

Please people, do not pick up -current unless you have already spent some
time on the -hackers and -commit mailing lists for FreeBSD. The -current
release is not for general consumption, it is for those doing active
development for future FreeBSD releases. There will NEVER EVER be any
documentation on upgrading to -current.

>I tried make bootstrapld before maek world. However, that actually updated
>a few of the *.h files in /usr/include, including a.out.h, and hence the
>executable file format. As a result, some of the new executables are not
>recognized, and I once lost my as (gas). But later I got it from someone

Yep, that's what -current is all about, fixing things that don't quite work.

>who mailed it to me, and replaced some of the header files with the old
>ones, and bring most of the thing back to where they were (I think). Then
>I started to build a new kernel with the new src tree, but I got the following
>error duing loading :
>
>loading 386bsd
>wd.o: Undefined symbol _isa_biotab_wdc referenced from text segment
>wd.o: Undefined symbol _isa_biotab_wdc referenced from text segment
>*** Error code 1
>
>Stop.
>
>  This symbol is declared as external in isa_device.h and used in wd.c but
>without actual declaration elsewhere. 

This isn't even a bug, it is due to the fact that config must be rebuilt
because of changes to one of the drivers. You really, really, really must
follow the -commit and -hackers mailing lists day in, day out if you want to
use -current.

>  Note, because my FreeBSD box is not connected to the net, so I can't sup
>the src into my box, and I am very much interested in keeping pace with the
>FreeBSD development up to date, and that's why I tried out the current src.

Not having net access is a bit of a disaster for running -current. Although
we keep it in pretty good shape there are occasions when it doesn't quite
work, the exec format changes were one example where we weren't sure how much
hassle it was going to be to upgrade until we tried it ourselves.

(Incidentally, before people start asking, the exec format changes will be
compatible with the old format as well as NetBSD and BSDI)

If you can't sup the tree then you're not going to be able to get the
continuous stream of bug fixes and changes that occur. 

I can understand people wanting to keep on the bleeding edge of the
development but unfortunately you do need certain resources in order to be
able to, net access being a major one. 

We intend making FreeBSD releases more frequent in future so people won't
have to wait too long for the latest and greatest features so please be
patient and only move to a -current system if you want to help in future
development or you really know what you're doing and can go it alone. If you
do decide to start running -current you *MUST* follow the mailing lists.

-- 
  Paul Richards, 
  Intelligent Systems Laboratory, ELSYM ,University of Wales, College Cardiff
  Internet: paul@isl.cf.ac.uk,  JANET(UK): RICHARDSDP@UK.AC.CARDIFF