Re: tlug: Last night's RMS discussion

["Stephen J. Turnbull" <turnbull@example.com>]

>>>>> "Simon" == Simon Cozens <simon@example.com> writes:

    Simon> Manuel M. T. Chakravarty (lists.tlug):

    >> Technically speaking, the OS that most of us use - according to
    >> rms - should be called GNU/Linux (or Linux/GNU if you prefer),

GNU over Linux makes more semantic sense.

    >> because the GNU tools and the Linux kernel are the two most
    >> important components in this system.[1] Which one is more
    >> important is a philosophical issue and of no real relevance in
    >> this discussion, I think.  (Don't underestimate the importance
    >> of the GNU tools, I am not talking `cp' here...without `gcc',
    >> the Linux kernel would be a pile of worthless bits.)

Oh, if there had been no gcc, Linus would have found another way to
compile the kernel.

Another important issue is GNU libc.  (And it's always been GNU libc,
even though HJ Liu's version was heavily hacked.)  It would be very
difficult to substitute any of the BSD libcs, let alone newlib, for
glibc, due to lack of POSIX (for newlib) and Unix9x support.

    Simon> You could make a case for GNU/FreeBSD on the same grounds,
    Simon> but RMS does not. I wonder why not.

BSD supplies its own libc, as well as many of the relevant tools (make 
and yacc for sure, and I would suspect there used to be a C compiler
even if they are using gcc now).  People used to the GNU utilities
(fileutils, shellutils, textutils, bash, etc) are generally upset by
the BSD versions.  :-)

BSD borrows about as much GNU stuff (except if they borrow GCC that
would be a big hunk) as GNU/Linux does BSD stuff.

    Simon> If you accept those terms. If we accept `GNU' as being `one
    Simon> or two bits of software provided by the FSF', would it not

One or two bits?  The C toolchain is not "one or two bits".  Nor is
libc.

    >> (1) He actually had the plan to make a complete system very
    >> long ago and the Linux kernel was only the last piece in the
    >> puzzle.

    Simon> But the most important.

Bullshit.  I have seen a GNU/Solaris system in operation.  In fact,
rms won't acknowledge it, but there is a pretty close approximation to
a GNU/DOS system available, of course most of the utilities like
`chown' were noops on it. :-)  Nobody in their right mind would use
such a thing, but kernels are a dime a dozen.

    Simon> His idea was never to sit on top of Linux, but to build a
    Simon> kernel and use that. Which makes me suspect even more
    Simon> highly it's nothing but a marriage of convenience for him.

That's exactly what it is; the Linux kernel is not true GPL (since
Linus permits binary modules), and rms wants to replace it with the
HURD as soon a possible.

    >> (3) The GNU part of the compound system is much more user
    >> visible.

    Simon> Absolutely and utterly untrue. The target audience of the
    Simon> war of words is the new breed of Linux users, not hackers
    Simon> who already know the arguments, and we're talking the
    Simon> people who probably won't use gcc and won't know what libc
    Simon> does. They've far more chance of knowing their kernel
    Simon> version, though.

But they don't know how to use uname, so they don't _really_ know what
kernel is actually running, do they?  The Linux kernel is not
user-visible, not until they learn how to use `cat /proc/whatever'.
The Linux name (and version number) may be user-visible, but I rather
doubt that most users know where in the startup sequence of messages
the kernel messages end and init's children begin.

I find it hard to believe that the typical user you are talking about
could (absent the boot messages and /etc/{issue,motd}) could taste the 
difference between Coke and Pepsi, kaff, kaff, tell whether the system 
they were running was Linux or FreeBSD, except by the presence or
absence of packages that haven't been ported.

    >> three points, the only one which is open for discussion is
    >> number (2), the other two are a matter of fact.

    Simon> Not fact, no.

I'm with Manu on both (1) and (3).

    >> PS: To Scott, maybe rms starts to call GNU/Linux just Linux,
    >> when Linus writes his own C compiler?

    Simon> Oh, you also forget there exist other free libc's. (newlib
    Simon> comes to mind immediately, and I'm sure I heard of others.)

The BSD libc is the relevant one.  newlib would not support a Linux
distribution; it's nowhere near POSIX compliant (although close enough 
to support Cygwin and DJGPP).  It's also something of a bitch to work
with, since it's not reentrant by default you will clobber yourself in 
threaded applications.

    Simon> I'm not sure about compilers, though.

Yeah, can't tell easily from looking at FreeBSD or NetBSD home pages.
Interesting, that.  _None_ of these OSes discuss the part of the
system that rms thinks IS THE SYSTEM on their home pages.  No
programmer's toolchain, no libc!!, no text editors, no admin
utilities.  X is invariably mentioned though, as are application
classes like webserver and various kinds of network hosts.

I guess we're all either kernel hackers or lusers now....

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
_________________  _________________  _________________  _________________
What are those straight lines for?  "XEmacs rules."
-------------------------------------------------------------------
Next Technical Meeting: January 14 (Fri) 19:00
* Topic: "glibc - current status and future developments"
* Guest Speaker: Ulrich Drepper (Cygnus Solutions)
* Place: Oracle Japan HQ 12F Seminar Room (New Otani Garden Court)
-------------------------------------------------------------------
more info: http://www.tlug.gr.jp        Sponsor: Global Online Japan