Open Software Foundation

[Moderator, John S. Quarterman]

From: John T. Nelson <uunet!potomac.ads.com!jtn>

I have mixed feelings about the Open Software Foundation.  On the one
hand I agree with AT&T when they state that the OSF has no track
record in the open software/standards business.  This is true.  The
founders of OSF consist of a handful of large computer companies whose
experience is mostly in marketing hardware and proprietary software,
not designing state-of-the-art operating systems, user interfaces and
environments for the scientist/engineer.

Now the OSF claims that they can pound together a standard which will
appeal to both System V and Berkeley users.  I fear that more than
likely the OSF standard will bind users to a single Unix standard and
thus to a single proprietary product... namely IBM's AIX, and
thus ensure IBM's position in the marketplace.

Making the standard available to everyone does not qualify it as an
"open" standard.  There must be open and equal participation among ALL
users of the system to make both technical and policy decisions when
defining the standard.  The OSF, if it is to live up to its advertised
ideals, must recognize the diverse needs of developers, researchers and
engineers who actually use the system and may prefer a system interface
that is different from what the standard proposes.

There must be diverse community participation.

> Membership
> 
> Foundation members will contribute ideas on both technical and policy
> matters.  They will be informed of foundation activities on a regular
> basis and periodically polled on specific issues.  Membership is open
> to anyone.

... for a price...

I note that individual/educational memberships to the OSF cost $5,000.
Corporate memberships cost $25,000.  Worse, simply being a member does
not allow you to make decisions on ALL policy issues.  Apparently the
OSF founders only want partners who have a signicant financial stake
in Unix.  This means that the individuals who will be most affect by
Unix (hackers, scientists) will be those most excluded from
membership if they cannot find an institution to sponsor them.  Even
then it isn't clear how much of a voice they will have in defining
the standard.

> The foundation's software environment includes a set of application
> programming interfaces to make it easier to write applications for a
> variety of systems.  The initial set of interfaces will support POSIX
> and X/Open specifications, and will be extended to include areas such
> as distributed computing, graphics, and user interfaces.

Sounds pretty good so far though.

> To provide a clear and easy migration path for application developers
> and end users, the foundation's system will include features to
> support current System V- and Berkeley-based UNIX applications.  The
> operating system will use core technology from a future version of
> IBM's AIX(tm) as a development base.

If the corner-stone of the OSF Unix standard is IBM's proprietary
product then how flexible can the standard be to user needs?  If the
new standards require massive rewrites to the AIX kernal will IBM
comply with the OSF's findings and completely rewrite their kernal?

> Specifications supported by the foundation will be publicly available,
> and a set of verification tests for all appropriate facilities will be
> identified or created.  The foundation will license its open system
> software internationally.

I would prefer to see an implementation maintained by an independent
university or the user community at large, not by a handful of
mega-marketing computer companies who have vested interests in products
they have already developed.  I would also like to see source code made
available so that the system can be easily maintained and modified at
each individual site.  Much like Berkeley source code licences.

On the other hand, I welcome the OSF as a positive effect on the
market place.  Adoption of standards that facilitate portability will
stimulate a somewhat stagnant and confused computer software industry.
Also, perhaps now Sun will get serious about their windowing systems
(to name just one of my gripes about Sun computers).  Sunview is a
toy.  NeWS is better and I think this will become a good product.

I'm much more interested in seeing Sun develop an entierly new and
integrated user interface... not just something that runs on top of the
shell.  Rather we should see a highly integrated window system/shell
much like the Symbolics LISP machine's windowing environment.  I
realize this is difficult to implement in Unix thanks to Unix's rather
backward way of thinking about the user interface, but hope springs
eternal.

Volume-Number: Volume 14, Number 19