From owner-freebsd-ports  Thu Aug 29 05:50:28 1996
Return-Path: owner-ports
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id FAA06448
          for ports-outgoing; Thu, 29 Aug 1996 05:50:28 -0700 (PDT)
Received: from dfw-ix11.ix.netcom.com (dfw-ix11.ix.netcom.com [206.214.98.11])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id FAA06443
          for <freebsd-ports@freebsd.org>; Thu, 29 Aug 1996 05:50:26 -0700 (PDT)
Received: from baloon.mimi.com (sjx-ca54-17.ix.netcom.com [206.214.106.81]) by dfw-ix11.ix.netcom.com (8.6.13/8.6.12) with ESMTP id FAA18430; Thu, 29 Aug 1996 05:49:20 -0700
Received: (from asami@localhost) by baloon.mimi.com (8.7.5/8.6.12) id FAA12458; Thu, 29 Aug 1996 05:49:18 -0700 (PDT)
Date: Thu, 29 Aug 1996 05:49:18 -0700 (PDT)
Message-Id: <199608291249.FAA12458@baloon.mimi.com>
To: james@ican.net
CC: jkh@time.cdrom.com, freebsd-ports@freebsd.org
In-reply-to: <Pine.BSI.3.95.960828225450.27084B-100000@expresslane.ca> (message from James FitzGibbon on Wed, 28 Aug 1996 22:59:06 -0400 (EDT))
Subject: Re: Should this port go in ?
From: asami@freebsd.org (Satoshi Asami)
Sender: owner-ports@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

 * > Please don't do this, if it's only manpages.  I do "grep foo
 * > /usr/ports/*/*/pkg/PLIST" all the time to find the port that installed
 * > a certain file.
 * 
 * Point taken, but couldn't you just as easily do 
 * 
 * grep foo /var/db/pkg/*/+CONTENTS
 * 
 * To do the same thing ?

Not really.  There are things like XFree86 that hasn't had a release
in ages (so I have the beta installed, but x11/XFree86/pkg/PLIST will
still catch those), and I sometimes have to blow away part of
/var/db/pkg because of some dependency problems killing pkg_delete
etc.  Also, it gets too damn crowded after a while when I'm compiling
so many versions of ports.... ;)

 * That would be good.  An even better solution would be some interface in
 * the Makefiles that allows the PLIST to be built dynamically, or have some
 * kind of conditional directive, like :
 * 
 * .ifndef NOMANCOMPRESS
 * man/man1/manpage.1.gz
 * .else
 * man/man1/manpage.1
 * .endif
 * 
 * That is interpreted by pkg_create or the package registration routines
 * during the installation.

Yeah, one problem is that bsd.port.mk copies the PLIST over to
/var/db/pkg during "make install" but pkg_create puts it in the
package.  So if we're going to fix this, we need to fix it in two
places.

I think we should either add an ability in pkg_create to do the fake
/var/db/pkg installation and call that from bsd.port.mk, or hack
bsd.port.mk to our heart's content and give /var/db/pkg/*/+CONTENT to
pkg_create as an argument.

Since bsd.port.mk is the only one that has access to all the
variables, maybe the second is the way to go.  (Although it's already
overly complex, and I'm not sure if we want to do string processing in
a .mk file....)

 * > If the list of files that change is really quite large, then that's
 * > ok.  (There are already some ports that do this, most notably the
 * > pkfonts ports, they have separate PLISTs for each resolution.)
 * 
 * That's where I got the idea for the one with compressed man pages and one
 * without.  (You pointed me there as I recall)

Oh gosh.  My memory is starting to fail me, I don't remember this at
all! ;)

Satoshi