From owner-freebsd-ports  Mon Dec 16 02:17:05 1996
Return-Path: <owner-ports>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id CAA16604
          for ports-outgoing; Mon, 16 Dec 1996 02:17:05 -0800 (PST)
Received: from dfw-ix6.ix.netcom.com (dfw-ix6.ix.netcom.com [206.214.98.6])
          by freefall.freebsd.org (8.8.4/8.8.4) with SMTP id CAA16598
          for <ports@freebsd.org>; Mon, 16 Dec 1996 02:17:02 -0800 (PST)
Received: from baloon.mimi.com (sjx-ca37-04.ix.netcom.com [204.31.236.196]) by dfw-ix6.ix.netcom.com (8.6.13/8.6.12) with ESMTP id CAA00961; Mon, 16 Dec 1996 02:16:29 -0800
Received: (from asami@localhost) by baloon.mimi.com (8.8.4/8.6.12) id CAA01299; Mon, 16 Dec 1996 02:16:26 -0800 (PST)
Date: Mon, 16 Dec 1996 02:16:26 -0800 (PST)
Message-Id: <199612161016.CAA01299@baloon.mimi.com>
To: msmith@atrad.adelaide.edu.au
CC: ports@freebsd.org
In-reply-to: <199612160215.MAA19272@genesis.atrad.adelaide.edu.au> (message from Michael Smith on Mon, 16 Dec 1996 12:45:08 +1030 (CST))
Subject: Re: Ports tidy-up sweep
From: asami@freebsd.org (Satoshi Asami)
Sender: owner-ports@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

 * If I read this correctly, a correctly-formatted md5 entry for an 
 * 'ignored' file will be :
 * 
 * MD5 (foo.bar.gz) = IGNORE
 * 
 * Is that right?

Yes.  Of course, the "MD5" in the front and parens are not really
necassary but it's easier to just copy the existing format (and
minimize bsd.port.mk changes.... :)

 * Sounds reasonable, especially if you're in a position to enforce it.
 * From my point of view, I have to trust that the contents of the md5
 * file is a definitive list of all the files that the port uses.  I tried
 * various hacks using make to emit the list of files used by the port,
 * but that blows the tree scan time out by an order of magnitude, which
 * really isn't an option.

I understand.  The only real "correct" way to do it is go through make
but that's really going to be dog-slow (I know 'cause I build the
INDEX file regularly!).

I think the best we can do is to make the framework as powerful as
possible so the porters don't need to set MD5_FILE.  (For instance,
the recent addition of ${DIST_SUBDIR} in front of the md5 pathname got
rid of www/netscape3/files/md5.gold.)

 * As it is just now, I only have one port left to complain about;
 * postgres-95.
 * 
 * Why does it have three md5 files, where two of them are for the
 * flex-2.5.2 binary package?  With flex in the tree now, this seems
 * silly.

And usr/src flex just went up to 2.5.4. ;)  It will be great if this
port can be updated to use the one we already have....

 * Thanks immensely for your support with my annoying feature requests!

No problem -- your "feature requests" are very helpful for me to
review the ports framework, and have led to a number of important
improvements.  (Or am I giving myself too much credit here. :)

Satoshi