From owner-freebsd-questions@FreeBSD.ORG Sun Nov 23 08:05:18 2003 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 1B05316A4CE for ; Sun, 23 Nov 2003 08:05:18 -0800 (PST) Received: from smtp.hispeed.ch (mxout.hispeed.ch [62.2.95.247]) by mx1.FreeBSD.org (Postfix) with ESMTP id EE70A43FF9 for ; Sun, 23 Nov 2003 08:05:13 -0800 (PST) (envelope-from murphyf+fhs@f-m.fm) Received: from dclient80-218-155-73.hispeed.ch (dclient80-218-155-73.hispeed.ch [80.218.155.73])hANG5BCR008545; Sun, 23 Nov 2003 17:05:12 +0100 From: Frank Murphy To: Charles Swiger Date: Sun, 23 Nov 2003 17:05:43 +0100 User-Agent: KMail/1.5.4 References: <20031121144116.A712D7E40E@server2.messagingengine.com> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200311231705.46849.murphyf+fhs@f-m.fm> cc: freebsd-questions@freebsd.org Subject: Re: FreeBSD, FHS, and /mnt/cdrom X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 23 Nov 2003 16:05:18 -0000 On Saturday 22 November 2003 8:18, Charles Swiger wrote: > /mnt should be reserved as a default temporary mount point-- it's silly > to risk breaking existing tools or procedures. Anyway, I suggest you > solicit feedback from Solaris users and possibly MacOS X people as > well. Solaris features vold (implied by wanting to use /vol), and the > latter OS places temporary removable mountpoints under /Volumes. The point that /mnt should be left alone is pretty clear, but I'm glad to be able to say "Folks on the FreeBSD questions list agree." As for Mac OS X, they have no intention of being FHS-compliant, so while we may learn some lessons from them, they won't worrz about what we have to say. > I happen to think that OS X handles things well from a user interface > standpoint-- the Finder in Panther with Miller column display and an > eject symbol next to the volume name, but I'm not sure how relevant > that is. Frank, is your group's standard concerned about physical > volume names, logical volume names intended for human > identification/access, or both? > > Physical device names ought to have unit numbers or even be part of a > tree-like device hierarchy-- for instance, what does /cdrom refer to in > a machine with two CD-ROM devices? In the current version of the standard (2.2), nothing. But in the next revision, /foo/cdrom will be a symlink to /foo/cdrom0, and /foo/cdrom1 won't have a link. Managing these links is not the scope of the standard (yet). The priority for the next revision is to define what "/foo" should be. > Human-readable names also run the risk of two removable devices having > the same name; people are happy seeing a list containing duplicate > names (eg, particularly if one name has a CDROM icon next to it, and > the other has a floppy or USB pen icon :-), but that doesn't tell you > what to do with your filesystem hierarchy layout. The actual names of the directories will be undefined, but there will be some suggestions (cdrom, floppy, etc.) > Obviously, a standard that says "place mount points anywhere you want" > isn't very useful. But if you did come up with a standard, who should > follow it and what would they gain? As for who should follow it, Linux distributions (Debian, Red Hat, SuSE) as well as the *BSDs (though I'm not sure exaclty what that means in the BSD world). What would be gained is more for application support. Basically, xmms and xcdroast could configure a /foo/cdrom as a default location, and it will be correct for all FHS-compliant systems. Frank