From owner-cvs-all Wed Oct 4 13:37:18 2000 Delivered-To: cvs-all@freebsd.org Received: from guru.mired.org (okc-27-149-77.mmcable.com [24.27.149.77]) by hub.freebsd.org (Postfix) with SMTP id 859CF37B503 for ; Wed, 4 Oct 2000 13:37:12 -0700 (PDT) Received: (qmail 17275 invoked by uid 100); 4 Oct 2000 17:50:31 -0000 From: Mike Meyer MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <14811.28135.356246.136840@guru.mired.org> Date: Wed, 4 Oct 2000 12:50:31 -0500 (CDT) To: Jordan Hubbard Cc: cvs-all@freebsd.org Subject: Re: How long for -stable [ Re: cvs commit: src/usr.bin/finger finger.c ] In-Reply-To: <72919698@toto.iv> X-Mailer: VM 6.72 under 21.1 (patch 10) "Capitol Reef" XEmacs Lucid X-face: "5Mnwy%?j>IIV\)A=):rjWL~NB2aH[}Yq8Z=u~vJ`"(,&SiLvbbz2W`;h9L,Yg`+vb1>RG% *h+%X^n0EZd>TM8_IB;a8F?(Fb"lw'IgCoyM.[Lg#r\ Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Jordan Hubbard writes: > > What about ports? How do you propose that they be tested, as opposed > > to "it-compiles-so-ship-it" on these 3.x boxes if, say, the developer > > in question only runs 4.x boxes, with a single not-yet-built 5.x box > > for when 5.x settles down? > I think we need to go back to providing dynamically created > "sandboxes" again, where the would-be tester can quickly create a > minimal (e.g. nothing more than strictly required) chroot tree from > scratch, chroot into it and build the port in question so that it and > all its deps get properly built and tested. We used to do that back > in the "old days" and then stopped, probably because people got > worried about root access for chroot and killed sandboxes rather than > simply firewalling the heck out of a sacrificial box and moving them > there. Anyway, these sandboxes should furthermore live on a 3.x > reference box which the project provides (so the developer's not on > the hook for it) and is specially selected for having cojones muy > grande in the disk and CPU department. I can arrange the hardware, > I'm fairly confident of that part. If you're going to start thwacking on the ports testing, I for one would *love* to see LOCALBASE/PREFIX != "/usr/local" become part of the standard testing. I'm not sure that LOCALBASE != PREFIX can be expected to work in all cases (nuts, LOCALBASE(now) != LOCALBASE(dep-install) will cause problems), but having them both pointed somewhere else ought to work reliably. Thanx,