Re: Clarification about section 3.1 (The Root Filesystem, Purpose)

[Marcus Brinkmann]

On Sat, Oct 12, 2002 at 04:33:37AM -0700, Daniel Quinlan wrote:
> Alfred M Szmidt <ams@kemisten.nu> writes:
> 
> > This is sad news.  As the /libexec proposal has been thrown out on
> > several occasions I doubt that it will be accepted in the future.
> > Also if a new system wants to introduce a new top level directory then
> > you will need to add an specific annex for each system.
> 
> Well, it is not really a useful directory.  I think most of the
> proposals come because GNU software continually wants to use it even
> though libexec violates FHS and most Linux distributions use FHS.

When I was wondering about that question, I read up the mailing list archive
of the FHS that reached back to 1997, IIRC.  And I formed my opinion based
on what was discussed there.  I have to admit that I was shocked about how
the conclusion to drop /libexec was reached (everybody can read this up for
their own).  I think there are good technical and user-interface reasons to
have a separate directory /libexec, and the FHS position on this was not
reached by a rationale chain of arguments, nor is it explainable by such
afterwards.  (Instead, AFAICR, it was a poll that led to the decision, a
poll which was critized even by those who participated and voted in favor
of dropping for being ambiguous, unclear, and short of options).

Which means to me that a mistake was made by dropping /libexec that should
be rectified.  Just pointing at the status quo and the dominance of the FHS
among GNU/Linux distributions is not good enough.  To me this only proves
that you are in a strong enough position to introduce and uphold a mistake.

I also want to point out that SuSE had a subdirectory in /usr/sbin for
binaries that should be in /libexec in my opinion, just to get them out of
the PATH.  They eventually changed this to the FHS position.  But my point
is that they were aware that a special location was useful (although they
got the wrong one, they didn't want it to be in the PATH), and that they
probably only changed it for uniformity rather than being convinved that it
was the right thing to do (although I can not speak for them).

And it is not only GNU software, BSD is using (/usr)/libexec, too.  It's the
FHS and the GNU/Linux distributions that single out and are special here.

However, I also want to say that /libexec is a rather small thing.  We think
that almost everything else in the FHS is fine work and right to the spot,
and we think that we can document the small differences we have in a
GNU/Hurd-specific annex.  Our goal is not to make everyone else to use
/libexec if they don't want to, we just want to use it for ourselves.
And, to conclude this, we don't want it just because GNU packages still
install there, but because we are convinced that it is useful and the right
thing to do.

> > Would it be possible to add an GNU/Hurd specific annex?  That has
> > /libexec, /hurd, /servers and similar directories that we might want?
> 
> I would welcome it.  A GNU/Hurd annex is quite possible.  We'd need to
> rationalize "libexec" and "com", but the others should be relatively
> easy.

That sounds good, thanks a lot.

> > I think that it would be easier to just change the rational to say
> > that no third party software may introduce new directories in the root
> > directory, and that distributions are free to do this.
> 
> It might be easier to have no standard at all, but part (most?) of the
> purpose of the standard is to achieve similarity between distributions.
> 
> If there is more than one GNU/Hurd distribution (always possible with
> free software), then you might want to reconsider.

We definitely want more than one GNU/Hurd distribution, and we would like to
have them all the same filesystem standard.  And we definitely want to build
on the good work you did with the FHS and use that for GNU/Hurd systems,
rather than writing another file system standard.   Not only to reach
similarity among GNU/Hurd systems, but also between GNU/Hurd and GNU/Linux
systems (even if there will be small differences, people will be able to
read up the differences in the FHS).

> > Side note, I added help-hurd@gnu.org to the CC list as I don't know how
> > many Hurd developers read the FHS mailing list.
> 
> Okay.  We have had involvement from one GNU/Hurd developer in the past,
> but it's been a while (5 or 6 years?) since he was here.

That was Thomas Bushnell, BSG.  He is still around here.

We are looking forward to good cooperation.

Thanks,
Marcus

-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    marcus@gnu.org
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
Marcus.Brinkmann@ruhr-uni-bochum.de
http://www.marcus-brinkmann.de/