From owner-freebsd-ports  Fri Nov 17 09:29:48 1995
Return-Path: owner-ports
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id JAA17059
          for ports-outgoing; Fri, 17 Nov 1995 09:29:48 -0800
Received: from silvia.HIP.Berkeley.EDU (silvia.HIP.Berkeley.EDU [136.152.64.181])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id JAA17052
          for <ports@freebsd.org>; Fri, 17 Nov 1995 09:29:39 -0800
Received: (from asami@localhost) by silvia.HIP.Berkeley.EDU (8.6.12/8.6.9) id JAA00967; Fri, 17 Nov 1995 09:29:37 -0800
Date: Fri, 17 Nov 1995 09:29:37 -0800
Message-Id: <199511171729.JAA00967@silvia.HIP.Berkeley.EDU>
To: ports@freebsd.org
Subject: Proposal 3: Non-executable file in *_DEPEND
From: asami@cs.berkeley.edu (Satoshi Asami)
Sender: owner-ports@freebsd.org
Precedence: bulk

Here goes proposal #3.  (Boy am I glad the release is out or what,
this is so much fun)

Some people have been complaining that the dependency mechanism is not
flexible enough, i.e., {FETCH,BUILD,RUN}_DEPENDS can only take an
executable as an argument.  Although this was intended to protect us
from cases when the target system has a non-standard path (e.g.,
/usr/opt instead of /usr/local) by using the "which" command that will
DTRT as long as the PATH is set correctly, it was impossible to depend
on a package that doesn't contain any executables.

Thus, I propose the following change to the *_DEPEND paradigm:

 @ For FETCH_DEPENDS, BUILD_DEPENDS and RUN_DEPENDS, if the argument
   starts with a '/', the existence of the pathname will be tested
   ("test -e").  Otherwise, they behave as before, and will be tested
   by "which".

 @ No change to LIB_DEPENDS, DEPENDS.

What do people think?

Satoshi