Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 18 Oct 2003 09:25:24 -0700
From:      Tim Kientzle <kientzle@acm.org>
To:        Poul-Henning Kamp <phk@phk.freebsd.dk>
Cc:        obrien@FreeBSD.org
Subject:   Re: cvs commit: src/share/man/man4 umass.4
Message-ID:  <3F916974.5030502@acm.org>
In-Reply-To: <7842.1066492471@critter.freebsd.dk>
References:  <7842.1066492471@critter.freebsd.dk>

next in thread | previous in thread | raw e-mail | index | archive | help
Poul-Henning Kamp wrote:
> In message <3F9161AB.5080608@acm.org>, Tim Kientzle writes:
>>Poul-Henning Kamp wrote:
>>>In message <20031017220732.GA61428@dragon.nuxi.com>, "David O'Brien" writes:
>>>
>>>>>I'm starting the s/disklabel/bsdlabel/ update in our manual pages.
>>>>>Any comment, opinion, etc...?
>>>>
>>>>Why?!?!?  bsdlabel is MD and is most likely the wrong thing for Sun's.
>>>>disklable is our MI name.
>>>
>>>You are just about as wrong as you can be there David.
>>>
>>>$ which sunlabel
>>>/sbin/sunlabel
>>
>><scratching head>  Okay, Poul-Henning, I give up.  You've shown
>>that /sbin/sunlabel exists, and use that as proof that "bsdlabel"
>>is the right way to label a Sun disk?
> 
> No, to show that disklabel isn't the right one on any platform.

Okay, I clearly missed something, then.  I thought that
the intent of the 'disklabel' name (which is linked to 'bsdlabel'
on i386, pc98, and amd64) was to provide an easy handle for
the default label program for the current platform.  Apparently,
David thinks so, too.

Is the "disklabel" name merely a transition-aid?  Is it
scheduled to go away?

I ask because what started this thread was a comment about
documentation.  We either have:

   * A 'disklabel' name that is "the default label program
     for this platform."  In that case, the man pages
     need to be checked to ensure that they use 'disklabel'
     in the generic case, and we need a 'disklabel.1' man
     page that clearly states that 'disklabel' is a synonym
     for different programs on different platforms.

   * No such default name.  In that case, the docs could get
     pretty ugly if we have to say things like "... after
     using fdisk, you need to label the disk with bsdlabel,
     sunlabel, foolabel, or xyzlabel, depending on your platform
     and disk-labelling preferences."  I'm not sure I want
     to be the one who has to explain that one.

In any case, it seems that David was right that a global
s/disklabel/bsdlabel/ is a bad idea.

Hmmmm....  An alternative might be to have a "disklabel" wrapper
along the lines of "mount" or "fsck" that accepts an
argument for which disklabel sub-program to run, with the
default varying by platform.  That would give a single interface
(easier to document) that allows savvy users to select the labelling
format.  It also opens the door for adding format-detection
intelligence (i.e., look at the current disk label and
use that format).

Of course, I suppose one could argue that all disk label
management should be pushed down into the kernel somehow.
Wish I had time to figure out how that might work.

Tim Kientzle



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3F916974.5030502>