that old GNU/Linux argument

Alexandre Oliva aoliva at
Thu Jul 17 11:09:29 UTC 2008

On Jul 17, 2008, DJ Delorie <dj at> wrote:

> Alexandre Oliva <aoliva at> writes:
>> You can't go 'hey, ma, look this book I wrote' when what you wrote
>> was only the preface.

> Likewise, you can't go "hey, world, look at this operating system I
> wrote" when all what wrote was only the C library.

Agreed.  I hope you're not trying to imply that the GNU operating
system is just a C library.  That would be silly.

> Maybe we should call it glibClinux, like we do for uClinux?

uClinux is actually a variant of the linux kernel targeted at
micro-controllers.  You know that.

It's most often used with uClibc, a C library targeted at
micro-controllers.  You also know that.

On top of that, there's more userland, a big part of which is still
GNU userland.

I haven't measured typical numbers here, but I doubt it would still be
appropriate to take GNU out of the name of the combination, if not for
strict copyright ownership, at least for heritage.

> You're been using Windows as an example of "an operating system" but
> Windows isn't Windows without the graphics.  By your examples, we
> should be calling the OS X/Linux.  Or XglibClinux.  Or kdeXgliClinux.

As written before and covered in the FAQ, this is not about who else
you cite.  It's about not citing the main contributor *while* citing a
minor one.

I.e., it's morally ok if you want to call it X/GNU or Gnome/GNU or
KDE/GNU.  But KDE/Linux, X/Linux, isn't.

> Oh wait, you included DOS in your example OS list.  DOS comes with
> ZERO libraries and applications[*] (does even count? even
> if it did, there were a lot of substitutes for it).

There are a number of utilities that added up to an operating system,
as limited as it was.  Limiting your analysis to just the kernel won't
prove anything about what the operating system amounts to.  If you
replace enough of the original operating system with something else,
you may indeed end up with something else.  But if you don't, it's not
right to rename it just because you want to.

Consider this: you take DOS, replace all of its userland with stuff
you developed yourself.  You still use the DOS kernel.  How would call
that?  Do you think you still get to name it all DOS?

Wouldn't it be more appropriate to name it after your own project's
name, or under a different name that didn't highlight only DOS in
detriment of your own project?

> The FSF is defining "operating system" as "the stuff the FSF did,
> plus Linux",

The FSF is using a definition of Operating System that is widely
accepted, even by Linus himself, as shown in his first announcement of

Do you disagree that GNU is a project to develop a complete Free
Unix-like Operating System?

Do you disagree that say linux-2.6.26.tar.bz2 is a compressed tarball
containing a kernel that works with that operating system?

> in order to convince the world
> that they deserve more attribution than everyone else. That the other
> contributors are NOT making similar demands on the world speaks

that they're not entitled to it, unlike the GNU project?

that they don't care about the freedom goals as much as the GNU
project does?

> That other people define "operating system" differently seems to be
> an ignored point by the FSF proponents.

In the end, it doesn't matter much how you choose to define operating
system.  RMS announced GNU as an operating system.  Linus announced
Linux as a kernel that works with the GNU operating system.  Both grew
over time, but GNU is still much bigger than Linux.  However you
define things or set boundaries (even against the opinion of the
original authors of both pieces) doesn't change the provable and
measurable fact that there's far more GNU than Linux in any GNU+Linux

Say, we write a book together, you write a 10-page chapter, I wrote 10
such chapters, and then you demand to have my name removed from the
cover, such that only yours remain there?  Why are we even having this
discussion?  How can you even try to make it sound like the FSF is the
bad guy, when it's just asking for the error *forced* by the Linux
promoters to be rectified.  If you claim the FSF is wrong in doing
that, why don't you *also* criticize the Linux promoters who first did
this to GNU?  Please try to at least be consistent, to yourself and to

Alexandre Oliva
Free Software Evangelist  oliva@{,}
FSFLA Board Member       ¡Sé Libre! =>
Red Hat Compiler Engineer   aoliva@{,}

More information about the fedora-list mailing list