that old GNU/Linux argument

[Marko Vojinovic]

On Friday 25 July 2008 07:56, Gordon Messmer wrote:
> I think that some knowledge of history would probably change your
> perspective, and is certainly relevant to the conversation.

Probably, but it seems that the argument I am trying to communicate is based 
on discussing the *purpose* of Linux and GNU, not on who started first or who 
did more work. Of course, knowledge of history can't hurt, but atm I have no 
time to read about it (I barely keep up with the thread itself...).

> An analogy is a fine way to clarify subject matter that someone doesn't
> understand, but it's an illustration, not an argument.  You're free to
> *use* an analogy as an argument, but it's very weak grounds.

Of course it's an illustration (of the argument itself). I did not mean it to 
be *the* argument. The argument itself goes as follows (in short): given the 
whole collection of executable code that makes up a distro, the distro itself 
is named for the group that packaged it (for example, Fedora) and the piece 
of code whose purpose is most essential --- the kernel. So one ends up with 
the "Fedora Linux" name. The main point is about evaluating my statement that 
the kernel software is more essential than GNU software. This is based (imho) 
on the point of view that there is a qualitative difference (regarding 
purpose) between the kernel and any other piece of code in the distro, 
including GNU, X, Gnome, RPM, whatever.

The analogy with the car engine is just (supposed to be) an easy way to 
illustrate what I mean by "qualitative difference" without going too much 
into philosophy. I agree that such an analogy is not an argument itself, but 
it (hopefully) sheds light on what I am trying to say.

> You argued that the automotive industry names their cars for their
> engine, but I just don't see it.  GM puts the same engine in Chevy,
> Pontiac, and Saturn cars, but do you know which ones?  I suspect not.

I gave the example of Formula 1 teams, where this naming is practically always 
present, and this is precisely because the engine of a Formula 1 car is so 
vital component that it's producer deserves to be mentioned in the name. Of 
course, this need not be a general rule in automotive industry. But it seems 
that we are going OT here. :-)

> Similarly, Windows is an operating system that people do not call
> kernel32.dll.  Mac OS X is an operating system that people don't call
> XNU.  GNU is an operating system that people inaccurately call Linux.
> The kernel is Linux.  The operating system is GNU.

Windows the OS runs on the kernel that doesn't have a distinguished name, 
because Microsoft didn't bother to give it one. The file name itself is 
irrelevant, even the Linux kernel file is called vmlinuz-(version), not 
"Linux" (last time I checked, at least).

As for GNU, I could simply say that an operating system doesn't exist without 
a kernel, but this is a matter of definition of an "operating system".

> > RMS? (I hope to know at least that much history.) So what about it? I see
> > no point here?
>
> The point is that RMS and GNU began creating an operating system, piece
> by piece, long before Linus began work on his kernel.  It was
> significantly easier to start at the top and work down than to work the
> other way around.  Thus, user applications were written before the kernel.

Sure, RMS and GNU did begin to create an operating system, but failed to 
finish it. Furthermore, they failed to complete the most vital component 
without which there is and cannot be an operating system. And now they ask 
for credit? For what? For cloning&enhancing Unix OS infrastructure and 
GPL-ing it? (ok, I am being a bit over the line here, I know, sorry... ;-) )

> > Well, from my (freshman's) point of view, it was big and professional,
> > but incompetent to produce the equivalent of 4 months work of a single
> > computer science student in Helsinki.
>
> Engineering is often like that.  Quick and dirty ends up beating out
> better engineered designs.

Yes, I have lived long enough to witness the PC platform beating out the Amiga 
platform, which was (afaik) better designed in those times. But that is life, 
a game of survival --- not the best-designed project survives, but the one 
that makes more money. :-)

> > In other words, kernel is qualitatively more important than all other
> > utilities that are used in conjuction. GNU did make all those utilities,
> > but failed to make the most important part.
>
> Whoa, there.  Is that what you think GNU is, a bunch of utilities?
> Hardly.  The GNU system is developed as separate projects, but the
> operating system isn't a mere aggregation of them.  The operating system
> is a whole that provides the services and interfaces that applications
> require to run.  That's vastly more than a kernel.

Ok, ok, my choice of word "utility" was off. Please substitute "utilities" 
with "software" (as a most neutral description). But the point remains. And 
also the fact that an operating system without a kernel is simply not an 
operating system. Because it cannot operate, to put it that way.

Best, :-)
Marko