Re: When can we expect a version 1.0 of the GNU Operating System?

[Bruno Félix Rezende Ribeiro]

Em Mon, 24 Nov 2014 09:44:23 +0000
Brandon Invergo <address@hidden> escreveu:

> Bruno Félix Rezende Ribeiro <address@hidden> writes:
> 
> > Incompatible?  So how are we supposed to release "the GNU system" if
> > no distribution could be called that way?  Are you implying that
> > only Hurd-based GNU systems deserve the label?
> 
> This goes back to my point that the problem with calling something
> "The GNU System" is that it implies that there is a single, specific
> set of software that defines the system.

It would only imply that if we intended it to mean so.  However, what
we are proposing is that using the term "the GNU system" rather than
"a GNU-variant system" would imply that the distribution at hand is a
GNU project's distribution of the GNU system, rather than of any other
organization, being the latter term used to mean this.  Even so, it
wouldn't imply that every GNU-variant system should have the same set
of components.  Summing up, the term "the GNU system" is expressing
mainly a matter of affiliation to the GNU project, rather than any
technical definition of what a GNU(-variant) system ought to be.


> The kernel issue then comes crashing to the forefront: if the GNU
> project has two kernels, and obviously only one can be in use at any
> time, which kernel does "The GNU System" use?

By our definition, given above, any GNU kernel that the GNU project
has a GNU system distribution for.  I would insist, though, that it is
best if the GNU project doesn't have two distinct distributions for
the same kernel.


> This is why I argued that we should instead talk about having a
> "reference GNU/Linux distribution" or something to that extent.

Making the GNU project's distributions of the GNU system reference
distributions could well have the effect you are arguing against:
promoting a strict set of software as mandatory to define any
distribution as a GNU system according to GNU project's view.  Rather
we are only talking about naming the GNU project's distributions of
the GNU system as "the GNU system", whatever the kernel, just to
convey the sense that they are distributions of the GNU project.  It
wouldn't imply nothing more than that.


> "Official GNU System" might have been imaginable decades ago but
> things have turned out differently: we have many thousands of good
> free software packages that can interchangeably comprise a GNU-like
> system (all for the better, in my opinion).  To declare some specific
> subset of them to be the "official" combination is not productive and
> dismisses a lot of perfectly fine free software.

I will reiterate that it's not a technical issue but rather a
strategical one.  We are not proposing to declare a particular set of
software official --- that's not relevant.  What we are arguing for is
to use the name "the GNU system", or simply "GNU", for every GNU
project's distribution of the GNU system.


> A "reference" distribution, on the other hand, is less constrictive
> and more like a recommendation.

The GNU project's distributions of the GNU system are not necessarily
any kind of recommendation at all.  In principle, they are just the
GNU system distributions of the GNU project, a.k.a, the GNU system.


> It allows room for other GNU/Linux distros to experiment with what
> they think a GNU System should be like, while giving them some
> recommendation about how we think it should be (all the while, of
> course, maintaining strict *requirements* with regard to software
> freedom).

Other GNU+Linux-libre distros can always experiment the way they
please, because GNU is free software.  It's not required that we
recommend anything about how a GNU system should look like.  Naming
our distributions as "the GNU system" only imply they are distribution
of ours.


> > If GNU is a system of multiple kernels, *every* GNU project's
> > distribution of the GNU system, be it Linux-libre, Hurd or a third
> > GNU kernel based, deserves to be called "the GNU system".
> > Technicalities that differ them are for tech-savvy people, not the
> > common public.
> 
> Ah, now you've finally seen my point :)  With your above words in
> mind, please go back and re-read my original ruminations on
> terminology.

I'm sorry, I really don't follow.


> The "GNU System" is something that arises out of a combination of
> software.

As you said, you are trying to make analogies between computer systems
and biological organisms.  I'd love to discuss that over a beer,
however those philosophical questions are out of scope for our
strategical concerns.  Philosophical questions aside, politically the
GNU system is, actually, what RMS claims it to be.  We believe he
should claim it's any of the GNU project's distribution of the GNU
system.


> The specific software building blocks can be interchanged without
> affecting the overall status as a GNU system.

In fact, according to our definition, any distribution made by other
organization than the GNU project is not a GNU system, even if
technically nearly identical to GNU project's distributions, but are
rather a GNU-variant system.  As you can see, that's a political
difference, not a technical one.


-- 
 ,= ,-_-. =.  Bruno Félix Rezende Ribeiro (oitofelix) [0x28D618AF]
((_/)o o(\_)) There is no system but GNU;
 `-'(. .)`-'  GNU Linux-libre is one of its official kernels;
     \_/      All software must be free as in freedom;

signature.asc
Description: PGP signature