Re: toolchain/45620: MAKEDEV failures remains unoticed in miniroot build

[Nicolas Joly <njoly%pasteur.fr@localhost>]

The following reply was made to PR toolchain/45620; it has been noted by GNATS.

From: Nicolas Joly <njoly%pasteur.fr@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: toolchain-manager%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost, njoly%pasteur.fr@localhost
Subject: Re: toolchain/45620: MAKEDEV failures remains unoticed in miniroot 
build
Date: Thu, 16 Feb 2012 13:49:03 +0100

 On Wed, Feb 15, 2012 at 07:25:03PM +0000, dieter roelants wrote:
 > The following reply was made to PR toolchain/45620; it has been noted by 
 > GNATS.
 > 
 > From: dieter roelants <dieter.NetBSD%pandora.be@localhost>
 > To: gnats-bugs%NetBSD.org@localhost
 > Cc: 
 > Subject: Re: toolchain/45620: MAKEDEV failures remains unoticed in miniroot
 >  build
 > Date: Wed, 15 Feb 2012 20:05:48 +0100
 > 
 >  On Tue, 14 Feb 2012 20:25:02 +0000 (UTC)
 >  Nicolas Joly <njoly%pasteur.fr@localhost> wrote:
 >  
 >  >  The following patch improves the situation. I moved the pipe to group
 >  >  sed (which is unlikely to fail) and sort. Unfortunately the error
 >  >  still gets ignored for parallel builds.
 >  
 >  I think the patch will work if you remove the leading "-" from the rm
 >  command. This is probably ok anyway, because rm -f will not fail for
 >  nonexisting files and for other errors (e.g. EPERM) failing seems to be
 >  the right thing to do.
 
 Seems the right thing to do ... I'm testing it, and will commit if
 successful.
 
 >  (As for why make decides to ignore the exit status of all commands for
 >  the target, I have no clue. I'll ask on tech-userlevel.)
 
 I can reproduce the problem with this small testcase.
 
 njoly@lanfeust [tmp/make]> cat Makefile
 foo:
         -true
         false
 njoly@lanfeust [tmp/make]> make
 true
 false
 *** Error code 1
 
 Stop.
 make: stopped in /home/njoly/tmp/make
 njoly@lanfeust [tmp/make]> make -j 1
 true
 false
 *** [foo] Error code 1 (ignored)
 
 Looks like a 'make -j' bug to me; at least, behaviour should be
 identical.
 
 Thanks.
 
 -- 
 Nicolas Joly
 
 Projects and Developments in Bioinformatics
 Institut Pasteur, Paris.