From owner-freebsd-alpha Mon Dec 28 01:52:28 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id BAA11942 for freebsd-alpha-outgoing; Mon, 28 Dec 1998 01:52:28 -0800 (PST) (envelope-from owner-freebsd-alpha@FreeBSD.ORG) Received: from vader.cs.berkeley.edu (vader.CS.Berkeley.EDU [128.32.38.234]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id BAA11936 for ; Mon, 28 Dec 1998 01:52:25 -0800 (PST) (envelope-from asami@vader.cs.berkeley.edu) Received: from silvia.hip.berkeley.edu (sji-ca32-59.ix.netcom.com [209.109.239.59]) by vader.cs.berkeley.edu (8.8.7/8.7.3) with ESMTP id BAA24003; Mon, 28 Dec 1998 01:52:03 -0800 (PST) Received: (from asami@localhost) by silvia.hip.berkeley.edu (8.8.8/8.6.9) id BAA19302; Mon, 28 Dec 1998 01:51:59 -0800 (PST) Date: Mon, 28 Dec 1998 01:51:59 -0800 (PST) Message-Id: <199812280951.BAA19302@silvia.hip.berkeley.edu> To: simokawa@sat.t.u-tokyo.ac.jp CC: sprice@hiwaay.net, alpha@FreeBSD.ORG In-reply-to: <19981228163227E.simokawa@sat.t.u-tokyo.ac.jp> (message from Hidetoshi Shimokawa on Mon, 28 Dec 1998 16:32:27 +0900) Subject: Re: Alpha ports collection? From: asami@FreeBSD.ORG (Satoshi Asami) Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org * BTW, how shall we treat shareable packages (e.g. X11 and TeX fonts). * It's waste of time and space to build packages on each architecture. * Do we need ARCH_COMMON tag and packages-common directoy in addition * to packages-i386 and packages-alpha? (Rolling eyes) I think you worry too much, Shimokawa-san. I think we can worry about those later. * Yes, it should be hard to fix many ports without commit access. Ok, I'll ask core about it. By the way, Steve, you seem to have an account on beast. If there is a problem, I can fix it. * I afraid it's difficult. It shouldn't be difficult to get cross compiler work, * but some ports may use their own bootstrap binaries (e.g. miniperl) which * cannot run on i386. Well, "some" is not too bad. If we can build most of them on i386 machines, then we can use real alphas to build the rest. The thing is that we've now got some funding to put together a "compilation farm" based on the method described in our paper. It's going to be something like 8 AMD K6-2 300's. Since alphas are still rather expensive, it's going to be much better if we can use those to build alpha packages as well as those for i386. (And maybe I can write another paper. :) * According to: * ftp://www.netbsd.org/pub/NetBSD/NetBSD-current/pkgsrc/mk/bsd.pkg.mk ftp * they use lists "ONLY_FOR_ARCHS" and "NOT_FOR_ARCHS" and use * Makefile.${MACHINE_ARCH}, patch.${MACHINE_ARCH} and so on, as well. Ok, I took a look. Yes, that looks better. I don't know when we'll go to more than two archs, but it's just about as easy to use as ONLY_FOR_ALPHA/BROKEN_ALPHA and infinitely more extendable. It doesn't look too hard to merge, but what's ${MACHINE_ARCH}? How's it different from our ${ARCH} (!= uname -m)? Is that the pc98/i386 thing? Satoshi To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message