🚀 go-pugleaf

I am in the process of selecting hardware for a free software
development environment.  This is greatly complicated by the fact that
there seem to be no suitable operating systems that are not proprietary
and expensive.  The basic minimum requirements are that the system must
be Unix compatible and it must support GNU Emacs and GCC, as these
will be the most essential basic tools used for porting and developing
other free software.  As I see it, the most reasonable options are:

A) Buy a workstation (Sun, DEC, NeXT, etc.) and live with
   the proprietary operating system supplied for it by its
   manufacturer until a free operating system is available.

B) Buy PC clone style hardware (with a 486, 386 or 386 SX CPU;
   I see no point in wasting efforts on old architectures with
   inadequate address spaces) and buy one of the proprietary Unix
   ports for it to use until a free operating system is available.
 
I am currently leaning strongly toward option B, for several reasons.
First, as much as I like workstations (a major part of my current
position involves supporting Sun and DEC workstations), my experience
with them makes me very hesitant to recommend them for a personal
computer environment; affordable maintenance for them is almost an
oxymoron.  Second, I consider clone hardware to be much more "free" in
the sense that it makes for an extremely open system.  Owners of PC
style systems are free to purchase motherboards, video cards, disks,
monitors, etc. from many different vendors, and there is strong price
competition.  The systems are also very modular, allowing owners to
independently upgrade CPU, memory, disk, and other components as
needed.  This modularity and the existence of multiple parts sources
also cuts maintenance costs drastically.  Finally, this approach
supports compatible systems across a very wide cost and performance
range; on the low end, 386SX systems will probably be available
later this year for less than $1000, and on the high end, 33 MHz 486
systems with big zero wait state caches should be available soon too.

If I knew what hardware platform would be the first to be supported by
a free operating system, I would almost certainly buy such a system.
However, this seems to be a big uncertainty at the moment, and I
expect that any free operating system will quickly be ported to other
platforms anyhow (I would certainly be willing to work on such a port).

I am aiming for a system price of about $5000 or so.  I am very
unhappy with the thought of paying a significant fraction of the cost
(it looks like a least $800 at the moment) for a proprietary version
of Unix, particularly when a major goal of the overall system is to
support work to eliminate exactly this type of proprietary software.

I would be interested to hear the comments of other people who
have faced this dilemma.  I would be particularly interested
in hearing from the Free Software Foundation people.  What
does FSF develop on?  Has FSF made any plans to focus on
particular hardware platforms for initial kernel support?

----------  Don Winsor -----------  don@eecs.umich.edu  ----------
University of Michigan Electrical Engineering and Computer Science