Re: horse carcas flogging (was: traceroute in /usr/bin, not /usr/sbin)
Hello,
On Fri, Jun 15, 2001 at 09:39:23PM +1000, Hamish Moffatt wrote:
> Hmm. You could be really tricky and use dpkg-divert to move traceroute
> from /usr/sbin to /usr/bin! Even better we could stick the diversion
> in the base-files package and preempt Herbert.
I don't really want to get into a pissing contest here. I think
traceroute (and possibly others) should be in /usr/bin, Herbert disagrees,
as is his prerogative. (What should perhaps be explored is if it really
violates the FHS to put traceroute in /usr/sbin, and if so if Debian Policy
allows/requires action to fix a broken package.) I think making a symlink
is an acceptable alternative, but the maintainer of the package in question
disagrees, so I'm proposing one way to "fix" such problems.
It is not a maintainer's prerogative to make choices that are contrary
to Policy (and thus, maintainers may not violate the FHS unless some other
part of Policy requires them to[1]), so to me that is the primary question
that we must answer. I'm suggesting that a "traceroute-fhs" or
"fhs-compatibility" package may be helpful is so that it can serve as an
overarching family of symlinks to fulfil our promise of FHS compatibility
for people that don't mind the mess of symlinks. I see this as similar to
(but much more kludgey than) the POSIXLY_CORRECT environment variable.
We may have to modify policy (ugh, I *really* don't like that idea) to
say that we are FHS compatible only with the "fhs-compatibility" package
installed. Ugly, isn't it? But the only way that I can see out of this
mess if individual maintainers won't step in line (if it is determined that
we are violating the FHS).
> Since our whole system is supposed to be FHS compliant it shouldn't
> be necessary. Let's just fix the broken packages instead.
The problem is that not everyone agrees that traceroute (for example) is
violating the FHS. Perhaps it is telling that no one seems to have argued
that traceroute is in the FHS mandated place despite my explicit request for
such arguments, but the traceroute apologists have not admitted that it is
violating the FHS either. The closest that has been come is arguing that
the FHS is wrong for requiring that traceroute be moved[2], which could be
read as an implicit affirmation that the FHS does mandate that it should be
moved.
So how about it? Is someone willing to claim that the FHS does not
require traceroute be moved? How about just the argument that the FHS
is imprecise on that question? Alternately, someone has to argue that
Debian Policy does not, in fact, promise compatibility with the FHS, or that
moving traceroute elsewhere would cause a conflict with the rest of Debian
Policy (even with a symlink left behind). Finally, someone could argue that
having symlinks from /usr/sbin to /usr/bin for traceroute (or even the other
way around) is unbearably messy and worth ignoring the rest of Debian
Policy.
These are the only arguments that come to mind, because quite frankly,
"The fact that nobody is following that rule suggests to me that this rule
may be removed/revised in a future revision of the FHS. Thus blindly
following it at the moment is probably a foolish thing"[3] is not a
compelling argument for me, especially given the option of making a symlink.
Rene, who still thinks this is a bad idea, but is not willing to lie to our
users.
References:
[1] <http://www.debian.org/doc/debian-policy/ch-opersys.html>, and
<http://bugs.debian.org/98291>
[2] <http://lists.debian.org/debian-devel-0106/msg00865.html> and
<http://lists.debian.org/debian-devel-0106/msg00879.html>
[3] <http://lists.debian.org/debian-devel-0106/msg00879.html>
--
+--- (Rene Weber is <rene_autoreply@elvenlord.com>) ---+
| "She dared not look too long at the [illustrated editions of children's |
| classics] for fear of seeing the name of Walt Disney, whom she regarded |
| as the patron saint of plagerism. _Walt Disney's Alice In Wonderland_, |
| indeed! It made her want to scream." -- Timothy Findley, "Headhunter" |
+--- E-Mail Policy & web page: <http://satori.home.dhs.org/~rweber/> ---+
Reply to: