Re: toolchain/47188: MKTOOLS=no broken

["Simon J. Gerraty" <sjg%crufty.net@localhost>]

The following reply was made to PR toolchain/47188; it has been noted by GNATS.

From: "Simon J. Gerraty" <sjg%crufty.net@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: apb%NetBSD.org@localhost, gnats-admin%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost
Subject: Re: toolchain/47188: MKTOOLS=no broken
Date: Sat, 17 Nov 2012 15:43:12 -0800

 On Thu, 15 Nov 2012 09:20:05 +0000 (UTC), Alan Barrett writes:
 > >TOOLDIR=/var/obj/NetBSD/current/tools/NetBSD-5.99.55-x86_64 \
 > >sh -x ./build.sh -m amd64 -M /var/obj/NetBSD/current/amd64/obj tools
 > >
 > >and then running:
 > >
 > >TOOLDIR=/var/obj/NetBSD/current/tools/NetBSD-5.99.55-x86_64 MKTOOLS=no \
 > >sh -x ./build.sh -u -U -m amd64 -M /var/obj/NetBSD/current/amd64/obj \
 > >-D /var/obj/NetBSD/current/amd64/destdir \
 > >-R /var/obj/NetBSD/current/amd64/release release
 > >
 > >fails [...]
 > 
 > I think there's a mistake in the above instructions.  I think that the
 > problem manifests only when the two build.sh invocations use different
 > obj directories, but the above instructions use the same -M option both
 > times.
 
 The above were cut/pasted from the logs of the tools vs release builds
 For the release build it also does:
 
 rm -rf /var/obj/NetBSD/current/amd64/destdir
 rm -rf /var/obj/NetBSD/current/amd64/release
 mkdir -p /var/obj/NetBSD/current/amd64/obj \
 /var/obj/NetBSD/current/amd64/destdir \
 /var/obj/NetBSD/current/amd64/release
 
 
 before running the above.
 
 It is quite possible that I did an rm -rf /var/obj/NetBSD/current/amd64
 at some point after the initial failure.
 
 That would then be equivalent to two different obj trees.
 
 Thanks
 --sjg