Re: Filesystem Hierarchy Standard (FHS) and NetBSD

[Jean-Yves Migeon <jym%NetBSD.org@localhost>]

On 10.05.2011 17:58, Thor Lancelot Simon wrote:
> Well, perhaps one can't blame "them" for trying *now*.  But do keep in
> mind that when this "standard" was propagated, decades of precedent
> (Unix has had a hier(7) manual page for that long!) appear to have been
> ignored once again with no solicitation of parties outside Linux for
> comment -- just the usual "Linux does it this way, let's get a standard
> that says everyone should."

And with the result we know.

IMHO, the intent is not to force everyone behind the "Linux does it this
way" stanza (I hope not, and that would assume that all Linux distros
have aligned on it -- tough challenge). Looks like more they want
parties to join in the rationalization process. For that you need to get
people involved.

Besides, we are not forced to follow it, we already have hier(7).

> I cannot really see what could be done to fix that now; there are a
> huge number of filesystem layouts in use across Unix variants and if
> it is "standardized" multiple ways, the standard is not really going
> to be very useful.  It's not reasonable to expect Linux to whack around
> its standard layout to match ours, or Solaris', or... you get the idea,
> and certainly not in the opposite direction either.  So what's to be done?
>
> One constructive thing that could be done would be to adjust the language
> of the "standard" to point out that it is informative not normative, that
> its scope is strictly limited to Linux systems, and to direct readers to
> sources like hier(7) which have documented existing Unix practice for much
> longer than the "standard" has existed.

That's not because Linux distros lack hierarchy standardization that we
should avoid cooperation. There's a tendency to amend file-system
hierarchy by fiat (/media, /run, or /cdrom for NetBSD come to mind), and
it's a mistake, IMO.

I agree that expecting FHS to be followed by all Unix variants is very
unlikely, especially for big players like Mac OS X. Just by considering
all the Linux distros, it's a very, very long run effort ahead...

-- 
Jean-Yves Migeon
jym%NetBSD.org@localhost