From owner-freebsd-current Thu Jul 31 00:58:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA14522 for current-outgoing; Thu, 31 Jul 1997 00:58:46 -0700 (PDT) Received: from internet1.mel.cybec.com.au (internet1.mel.cybec.com.au [203.103.154.130]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA14514 for ; Thu, 31 Jul 1997 00:58:36 -0700 (PDT) Received: from tech34 (tech34.mel.cybec.com.au [203.103.154.37]) by internet1.mel.cybec.com.au (post.office MTA v2.0 0813 ID# 0-14031) with ESMTP id AAA671; Thu, 31 Jul 1997 17:59:46 +1000 Message-ID: <33E04610.81C2C2EC@cybec.com.au> Date: Thu, 31 Jul 1997 18:00:16 +1000 From: TLiddelow@cybec.com.au (Tim Liddelow) X-Mailer: Mozilla 4.01 [en] (WinNT; I) MIME-Version: 1.0 To: "Jordan K. Hubbard" CC: Chuck Robey , FreeBSD current Subject: Re: core group topics X-Priority: 3 (Normal) References: <19990.870331913@time.cdrom.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Jordan K. Hubbard wrote: > > Erm, "none excepting inertia and fear of chaos, confusion, shock and > increased entropy all around." :) Inflicting another ABI on our user > base, one which will result in a new class of packages which all older > FreeBSD releases cannot run, is NOT something to be considered > lightly. > > Jordan It doesn't have to be _that_ hard. The Linux camp did the dual ABI thing for quite a while. Sure - the transition wasn't all beer and skittles but they did it. And they're reaping the rewards (see previous postings) for it. Another thing - so what if OLDER FreeBSD releases can't run the ELF packages ? You have symlinks for packages for a certain release, right ? There's no way if I am running FreeBSD 4.0 (ELF) that I would try to add an a.out package. And vise versa. OLDER FreeBSD releases will continue to use the a.out packages. New packages would have to be made for both 2.1.7.x release(s) and the newer ELF release. Noone is suggesting there aren't going to be hurdles. What has to be considered is the cost in remaining a.out, and the advantages of ELF in the longer term. We rely on third party (GNU) software that supports ELF "out of the box". If we continue down the a.out path for too long, we risk obscurity in terms of language support, etc. When gcc 2.8 comes out (fingers crossed) will it support a.out correctly ? C++ support on FreeBSD is something that really interests/concerns me. FreeBSD plugs itself as a great networking/internet server. What about a development box ? ELF is great for development - e.g. Java, C++, etc. Just my $0.02. Tim.