[linux-elitists] The First-Ever Installfest in Egypt (fwd from brian-slashdotnews@hyperreal.org)

Karsten M. Self kmself@ix.netcom.com
Tue May 4 00:50:18 PDT 2004

on Mon, May 03, 2004 at 12:31:03PM -0400, Jeff Kinz (jkinz@kinz.org) wrote:
> On Mon, May 03, 2004 at 03:54:17PM +0200, Eugen Leitl wrote:

> > Fedora Core 2 test 3 <http://torrent.dulug.duke.edu/> is a genuine
> > improvement in end user usability. The 2.6.x kernel produces a much
> > superior GUI response. Resolution resizing in X doesn't require
> > restarting the server.  Lots of other goodies, too, but those are
> > more under the hood.
> > 
> > Have been attending a road show (Sun dual Opteron servers, Java
> > Desktop System) today. I still don't understand what Sun's story on
> > Linux is. They must know it's eating them alive both at the low end
> > and at the clustering end.
> > 
> > Is Sun confused? Or are they just faking it? Anyone knows?

> Is this a serious question or just a rhetorical one?

> Please note that I think SUN does make some good stuff. This is not a
> rant against SUN. SUN is just at the wrong place of the market at the
> wrong time. If they had followed MS lead and commoditized UNIX in the
> mid to late 1980's they could now be in MS's very enviable position
> and would consequently be much less vulnerable to the impending Linux
> erosion. In fact, a commoditized BSD-based UNIX might have prevented
> the development of Linux at all. Would enough people have worked as
> hard on Linux as they did if there had been Intel-Solaris licenses
> available for $50-$60?

Solaris, probably not.

However, there *were* PC Unices available for around that price-point in
the early 1990s.  One of the items I stumbled across in researching the
Caldera/SCO case early on was ESR's "PC-clone UNIX Software Buyer's
Guide" (no longer maintained ;-):


A few of the last versions of this are still available online, and do an
interesting job of cataloging ESR's own awareness, exposure, and
conversion to GNU/Linux.

However, there were other Unices available at the time, several in the
$50-$500 price-point likely appealing to hobbyists and students, at
least for limited (e.g.:  2-user, no development support) systems.

I don't think that would have sufficed to quash GNU/Linux (or an
alternative free Unix).   

ESR noted that many of the technically better systems were those which
were lower priced.  He particularly notes that SCO UNIX was among the
most overpriced, underperforming alternatives.  Marketing (and
partnerships with both hardware and vertical-solution software vendors)
trumped sheer technical capability.

Too:  the market was crowded by small firms barely scrabbling by.

The problem with low-cost versions of *proprietary* software is that
they tend to:

  - Follow rather low, short, tragectories.  Popular for a time on
    price and particular feature sets.  Then they die as economic
    realities come home to roost.

  - Be, remain, and become ever moreso, fragmented.  The applications
    are still fundamentally proprietary, and compete on exclusivity of

I see a lot of similarities between the PC-UNIX proprietary market, the
various commercial GNU/Linux distros, on the one hand, and the shareware
/ inexpensive toolware market for PC utilities.  Some tools persist for
a long time (PKZip, Symantec and Norton AV, more recently some CD
burning tools), but the markets are generally small and short-lived.

GNU/Linux has seen the coming and going of several companies, with
others now struggling or slogging along:  SLS, Yggdrasil, TurboLinux,
Caldera, and SuSE were all one-time leaders or major players now
entirely defunct, struggly badly, or aquired by a larger player.  Most
of these were also noted for having significant proprietary elements,
particularly in management interfaces.  Though this really isn't a
_major_ interoperability issue at the system management level.

Still:  the tendency for such tools and/or distributions is to emphasize
differences rather than strengthening interoperability.  Particularly
with similar competing products.

In a purely free software context, this problem *doesn't* exist.  The
classic 'Nix interoperability interface is the pipe.  More recently,
APIs, or fully embedded implementations (modPerl, python libraries in
Perl, MySQL modules in PHP, and of course Emacs and everything) show
that the boundaries between various free software projects are permiable
and vague at best.

My take, then, is that a cheap Solaris wouldn't have prevented a
GNU/Linux from emerging.  It might have pushed things back a few years,
but RMS would still not have been satisfied, and if not a Finnish grad
student, then a small group with technical scalability, portability, or
distribution issues would have pushed forward.  It *is* possible that
the still-closing window offered by patent law would have made such
development untenable, but I think that this itself is to an extent a
reflection of GNU/Linux and the 'Net revolution it in part enabled.

> By the way, don't blame me for it not happening. :) I suggested
> just that to SUN several times in the 1980's and 1990's. Nobody was
> interested and now its far too late.

One of the more significant such proposals was Larry McVoy's free Unix
proposal, "The Sourceware Operating System Proposal".  Still available
on Red Hat's website:



> (Anyone recall "Road Runner"? SUNOS running on a 386?)


> Two years ago, on a New England Linux user's list (either blug or
> gnhlug) I stated that SUN would be the first serious "victim" ("entity
> severely diminished by") of the Linux movement. 

Well.  SCO (in it's original or current incarnation) probably takes that
claim, though they were "serious" in name, rather than revenues.  You
could also argue that DEC/Compaq, HP (HPUX), SGI, and IBM at least made
a tranformative change.  Wind River is an example of a specialty market
converted, as is Symbian, and as will be PalmOS / WinCE.  GNU/Linux has
been undercutting the 'Nix market for the past thirteen years in both
small-scale and special-purpose niches.  It's taking on both the
datacenter and desktop now.

Comments previously made to this list are now part of my "Free Software
Primer" page at TWikIWeThey:


...and the thoughts of this post are really an extension of this, along
the lines of how proprietized software development is inherently at odds
with developing quality, sustainable, interoperable software.


Karsten M. Self <kmself@ix.netcom.com>        http://kmself.home.netcom.com/
 What Part of "Gestalt" don't you understand?
   Iomega:  click of death, Jaz Junk, and now, NAS?  Not!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://allium.zgp.org/pipermail/linux-elitists/attachments/20040504/7706619c/attachment.pgp 

More information about the linux-elitists mailing list