On Fri, Oct 12, 2001 at 04:02:00PM -0400, Greg A. Woods wrote:
> [ On Friday, October 12, 2001 at 15:12:30 (-0400), Jim Wise wrote: ]
> > Subject: Re: CVS commit: basesrc/distrib/sets/lists/base
> >
> > Why?  Does anything in netbsd use /var/named by default in the
> > non-chroot case (the default named.conf uses /etc/namedb).
> > If not, surely it's not unusual to expect that if you choose a
> > non-default path, you create that path, no?
> 
> Well I had been choosing the default path, and it had been created for me.

Greg, I don't mean to rain on your ``I want to comment on any
decision made in NetBSD contrary to my beliefs and I'll email
brow-beat anyone who will listen'' parade (but I suppose I am going
to anyway), but let's just look at the history of /var/named:

I added it to /etc/mtree/NetBSD.dist on Wed Feb 28 19:22:30 2001 UTC,
in revision 1.141, with the following log message:
	setup named chroot cage under /var/named

That change was after the creation of the netbsd-1-5 branch, and was
not pulled up to that branch.  It was part of a series of commits of
some work I was experimenting with to make it much easier to setup
named chroot-ed.

Acting as if the change from "/var/named" to "/var/chroot/named" was
something thats affected a great number of users (of our release
software) is disingenious.

If you are configuring your system away from the NetBSD defaults in a
manner that you like (and I have no problem with this), then part of
that flexibility is that you must create directories and files needed
by your non-standard configuration.


As we're considering making it easier to chroot many system
applications, it was felt (after discussion with a few people) that
a seperate hierarchy for chroot-ed applications makes sense, in
case the end-user needs configuration flexibility with items such
as separate mount point options for the chroot-ed cages. That's
why "/var/chroot/<user>" was chosen. In hindsight, "/var/named"
set a bad precedent for polluting /var with chroot cages.

Luke.