The following reply was made to PR bin/32589; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg@britannica.bec.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: bin/32589
Date: Sat, 23 Sep 2006 15:22:07 +0200

 On Sat, Sep 23, 2006 at 01:15:06PM +0000, Daniel de Kok wrote:
 > The following reply was made to PR bin/32589; it has been noted by GNATS.
 > 
 > From: Daniel de Kok <danieldk@pobox.com>
 > To: gnats-bugs@netbsd.org
 > Cc: 
 > Subject: Re: bin/32589
 > Date: Sat, 23 Sep 2006 15:12:21 +0200
 > 
 >  Hi Joerg,
 >  
 >  On Fri, 22 Sep 2006 22:31:05 +0200, Joerg Sonnenberger wrote:
 >  > But this is a hard error, since the expectation are not matched.
 >  
 >  Maybe it is best to use warn() first, and leave this PR open. If
 >  the change doesn't cause havoc, we could switch to err(). What
 >  do you think?
 
 It *did* cause havoc. The problem with warn(3) is, that it will most
 likely just ge tlost e.g. in build outputs, but a failure will be a lot
 more visible. For users not checking the return value of install, it
 doesn't make a different at all.
 
 Joerg