From owner-freebsd-ports  Fri Aug  2 13:27:19 1996
Return-Path: owner-ports
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id NAA28140
          for ports-outgoing; Fri, 2 Aug 1996 13:27:19 -0700 (PDT)
Received: from silvia.HIP.Berkeley.EDU (silvia.HIP.Berkeley.EDU [136.152.64.181])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id NAA28135
          for <ports@freebsd.org>; Fri, 2 Aug 1996 13:27:13 -0700 (PDT)
Received: (from asami@localhost) by silvia.HIP.Berkeley.EDU (8.7.5/8.6.9) id NAA07816; Fri, 2 Aug 1996 13:24:02 -0700 (PDT)
Date: Fri, 2 Aug 1996 13:24:02 -0700 (PDT)
Message-Id: <199608022024.NAA07816@silvia.HIP.Berkeley.EDU>
To: csdayton@midway.uchicago.edu
CC: imp@village.org, ports@freebsd.org, mark@grondar.za
In-reply-to: <199608021709.MAA04372@woodlawn.uchicago.edu> (message from Soren Dayton on Fri, 02 Aug 1996 12:09:47 CDT)
Subject: Re: Automatic ports Makefile generator?
From: asami@cs.berkeley.edu (Satoshi Asami)
Sender: owner-ports@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

 *  This is what I am concerned about.  Patches seem to be applied in
 *  blocks that do not necessarily map nicely onto the files that they change.

Yes, everything will be lumped into one big "patch-aa" or it will be
one file per patchfile.  Either way it loses the nice logical
separation that many ports have ("patch-aa for Makefile tweaks,
patch-ab for malloc.h changes, etc.").

Also, I'd hate to have people send in ports with mega-patches that
change Makefile when the port uses Imakefile, configure when it can be 
done by a 2-line patch and autoconf, etc.  The "generator" should be
real smart to be able to do this.

 * will it generate the list of files?  That is the most frustrating thing
 * for me for big packages.  something that reads the make output and
 * figures out where the files went would be marvelous.

I've heard someone wrote a tool to do this (on HP-UX I believe) a
"find /usr/local | touch; install; find -newer" kind of thing to at
least get the list correctly.  With our implementation of symbolic
links, I don't know how well it will work for us, though.

Hmm, maybe a diff of locate database?

Satoshi