From: Peter Korsgaard <peter@korsgaard.com>
Date: Mon, 24 Nov 2014 21:47:45 +0100
Subject: [Buildroot] [PATCH 13/51] samba: use <pkg>_INSTALL_INIT_SYSV
	mechanism
In-Reply-To: <20141124214500.58bd4f37@free-electrons.com> (Thomas Petazzoni's
 message of "Mon, 24 Nov 2014 21:45:00 +0100")
References: <1416775163-20215-1-git-send-email-thomas.petazzoni@free-electrons.com>
 <1416775163-20215-14-git-send-email-thomas.petazzoni@free-electrons.com>
 <20141124044922.GB2241@tarshish>
 <20141124214500.58bd4f37@free-electrons.com>
Message-ID: <877fyktjf2.fsf@dell.be.48ers.dk>
List-Id: <buildroot.busybox.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: buildroot@busybox.net

>>>>> "Thomas" == Thomas Petazzoni <thomas.petazzoni@free-electrons.com> writes:

 > Dear Baruch Siach,
 > On Mon, 24 Nov 2014 06:49:22 +0200, Baruch Siach wrote:
 >> Hi Thomas,
 >> 
 >> On Sun, Nov 23, 2014 at 09:38:45PM +0100, Thomas Petazzoni wrote:
 >> > +define SAMBA_INSTALL_INIT_SYSV
 >> > +	# install start/stop script
 >> > +	@if [ ! -f $(TARGET_DIR)/etc/init.d/S91smb ]; then \
 >> > +		$(INSTALL) -m 0755 -D package/samba/S91smb $(TARGET_DIR)/etc/init.d/S91smb; \
 >> > +	fi
 >> > +endef
 >> 
 >> Why not install the init script unconditionally like you did in vsftpd?

 > Well, when I saw how many packages were still installing init script
 > (and other files) conditionally, I kind of changed my mind mid-way
 > through the patch series, and therefore for the remaining patches
 > decided to keep things as they were: keep the condition when it was
 > there, and keep the direct installation with no condition when it was
 > done this way.

 > We certainly need a decision about whether we want those conditions
 > everywhere. Their role is to potentially allow the users to provide a
 > custom version of those scripts (or configuration files) in the
 > filesystem skeleton. I personally believe it's not a good idea, for two
 > reasons:

 >  * People should use a filesystem overlay and/or post-build script
 >    instead of a custom skeleton.

 >  * Why would we install our own stuff conditionally, while the
 >    build system of all packages install things unconditionally, making
 >    it anyway impossible for a target skeleton to customize all files.

 > Peter, your call?

Yes, I agree - Lets get rid of these legacy checks. Using a custom
rootfs overlay / post-build script is a lot more flexible and robust.

-- 
Venlig hilsen,
Peter Korsgaard