From owner-freebsd-hackers Thu May 23 07:24:21 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id HAA15450 for hackers-outgoing; Thu, 23 May 1996 07:24:21 -0700 (PDT) Received: from hamby1.lightside.net (hamby1.lightside.net [198.81.209.17]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id HAA15379; Thu, 23 May 1996 07:24:15 -0700 (PDT) Received: from localhost (jehamby@localhost) by hamby1.lightside.net (8.7.5/8.7.3) with SMTP id HAA00557; Thu, 23 May 1996 07:26:59 -0700 (PDT) X-Authentication-Warning: hamby1.lightside.net: jehamby owned process doing -bs Date: Thu, 23 May 1996 07:26:58 -0700 (PDT) From: Jake Hamby X-Sender: jehamby@hamby1 To: "Jordan K. Hubbard" cc: Michael Smith , sos@freebsd.org, gpalmer@freebsd.org, hackers@freebsd.org, peter@freebsd.org Subject: Re: src/gnu In-Reply-To: <3393.832850820@time.cdrom.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Thu, 23 May 1996, Jordan K. Hubbard wrote: > > If it's kept as a 'port' rather than mutilated and stuffed into the source > > tree, it should _theoretically_ be easier to get the FSF people to accept > > patches to it. I think the idea has a lot of merit. > > One plus is that with the ports collection model (and it's funny that > I never thought of anything outside of /usr/ports including > bsd.port.mk before, but thinking about it now it makes perfect sense :-), > all the patches will always be broken out in ONE location and > it becomes a very simple exercise to work with the FSF until your > patches directory goes away - what other metric could be simpler? > > I'm not sure that having gcc bmake'd has ever bought us much anyway. > Same goes for groff, for that matter. > > Jordan You know, this sounds like an excellent idea! I like the way that patches aren't jumbled up with the FSF source tree. I also like the idea that this makes it easier for end-users to upgrade, for example, GCC, on an experimental basis and have it install in the correct location (/usr/libexec) without much work, then if there are no problems, it is much easier to suggest that FreeBSD import the latest GCC. It also has the added bonus of making it easier to back up to old versions if something goes wrong. Also, in theory, this makes our CVS tree much smaller because we only need to include patches and a link to the FSF .tar.gz file. Then, a simple "make extract" before the make world, will extract all the .tar.gz files into the correct place (work subdirectory?) and everything proceeds normally. I like this idea! This won't work with programs like gas and ld, for which our version has diverged too much from the latest GNU binutils. Of course if we switch over to ELF, this becomes viable again. In the meantime, I propose the "ports" idea be used for GCC, CVS, RCS, groff, gzip, tar, and anything else in the tree from FSF. Even non-GPL stuff like ncurses would benefit from this! Comments? ------------------------------------------------------------------------------ |Jake Hamby| Ask me about Unix, FreeBSD, Solaris, The Tick, Motif, or NT, eh?| ------------------------------------------------------------------------------ "Hi, can I interest you in buying some meat over the phone?" -Lotus commercial