On Thu, Jun 27, 2013 at 12:13:51PM -0700, Garrett Cooper wrote: > On Jun 22, 2013, at 3:51 PM, Glen Barber wrote: > > > On Sat, Jun 22, 2013 at 03:39:41PM -0700, Garrett Cooper wrote: > >> On Jun 22, 2013, at 3:12 PM, Glen Barber wrote: > >> > >>> Author: gjb > >>> Date: Sat Jun 22 22:12:24 2013 > >>> New Revision: 252101 > >>> URL: http://svnweb.freebsd.org/changeset/base/252101 > >>> > >>> Log: > >>> If MAKE_CONF and/or SRC_CONF are set and not character devices, > >>> copy to chroot to use with release build. > >>> > >>> Submitted by: Garrett Cooper (original version) > >>> Approved by: kib (mentor) > >>> MFC after: 1 week > >>> > >>> Modified: > >>> head/release/release.sh > >> > >> Technically copying /dev/null to a file creates a 0 byte file. > >> That's why I used cp in the patch I submitted over email. > >> Thanks! > > > > Right. But did you read the script? > > > > Both __MAKE_CONF and SRCCONF are set to the path of the local files. > > Oh, interesting… I see what you did there instead of what > I proposed; I copied __MAKE_CONF to /etc/make.conf and SRCCONF to > /etc/src.conf for symmetry with the base system, but your method > works more than mine did (I forgot that you were passing those vars > in the chroot environment). Should the files be installed to /etc/ > in the system default locations ?
My initial reaction is "no." Those who are using __MAKE_CONF and SRCCONF in such an environment know why they are doing so. So, in my opinion, it is up to them to modify the end-release as needed. Since __MAKE_CONF and SRCCONF are pointing to /dev/null in release builds, I do not see this as necessary. Moreso, I do not want to introduce a level of confusion where /etc/make.conf and/or /etc/src.conf are modified within the release, or worse, in an altogether different location. There is too much complexity here. Glen
pgpS7GMkeAMXv.pgp
Description: PGP signature
