# Java/Linux open source

Robert S. Thau (rst@ai.mit.edu)
Sun, 3 Jan 1999 17:24:18 -0500 (EST)

Ron Resnick writes:
> Robert S. Thau wrote:
>
> > Errmmmm --- source code *was* provided to all users of Bell Labs
> > Research Unix at least through v6 or so. (They did not have
> > permission to redistribute, of course, but neither do users of Minix;
> > the Minix license requires royalties for redistribution of source).
> >
>
> So, is that "open source", or "Open Source" :-)..
>
> Seriously, if one buys into Eric Raymond's argument, redistribution is
> a key element of true OSS (I'm not saying I buy this argument
> myself, mind you..). This (redistribution)
> is a key argument made against the new
> Sun SCSL structure, incidentally.

Actually, I was just knee-jerking off at the "no-source-available"
description of the research Unix distributions. However, the
distinction you're drawing between that state of affairs and "open
source software" as we know it today is perhaps a bit ahistorical.
IMHO, the "modern" notion of open source originated in an effort to
codify and preserve the rights of software users (particularly those
in academe) to keep on doing what they had been doing with the
software on their systems all along, without necessarily having
strict legal authority to do so.

The fact is that users of Research Unix could and did redistribute
modified versions of the source code to other members of their
community --- to wit, the (admittedly narrowly defined, and largely
BSD stands for Berkeley's *Distribution* of their own extensions to
the system, with contributions from others. To this extent at least,
define as the benefits of open source. There were similar communities
around the licenses of other systems, such as (for example) the users
of system software for the Lisp Machines derived from MIT's original

Of course, these arrangements were founded less in the legalities
surrounding the licenses than in an unstated web of community ethics
dating back from the time, surprisingly late in the '60s, when source
was routinely supplied with all software, modifications by the end
user were standard practice, and people weren't even sure that
copyrighting and selling software was a sensible (or legal!) thing to
do. And when the software in question starting getting sold for
serious money, comparable to the cost of the hardware, the
arrangements broke down --- which is precisely (in the case of the
truly hideous Lisp Machine mess) what led Stallman to invent the
copyleft and start the modern open-source movement.

(There might be some interesting lessons comparing this whole state of
affairs to the enclosure movement in the English countryside ---
around the turn of the 1800's, I believe --- in which landholders
asserted rights to control the use of what had formerly been, at least
de facto, peasants' common lands. BTW, the landholders ---
counterparts of what Stallman calls "code hoarders", in this analogy
--- won handily. But I digress).

> Had AT&T code been generally, widely, unquestionably
> available throughout the 70s and 80s, would we have had the BSD/SYSV
> split? Would we have had the Unix wars of HPUX/AIX/SCO etc, right
> at the time that MS was about to steal away the whole show?

Unless AT&T invented the copyleft before Stallman got to it, the wars
would surely have gone on exactly as they did --- HP, Sun, IBM, DEC,
distribute their source mods to *each other*, and with rare exceptions
(NFS, NIS, etc.) they didn't.

> For that
> matter, would AT&T have made any less money off their property, thna
> they ultimately did when they sold (to Novell, right? And then on to
> SCO?).

Interesting tea leaves to read here, without pointing to any
particular conclusion:

*) Red Hat CEO Bob Young's oft-repeated quip that his company is
trying to *reduce* the size of the market it's in by a factor of
ten (by lowering the price of the OS).

*) OSF's failed attempt to start charging royalties for their last
release of X.

There certainly would have been *some* money to make as sources of
"the official Unix" guides and support --- but I doubt AT&T's
executives would have even considered going into that business
(particularly considering the potential size of the market in the
computers-the-size-of-refrigerators, mainframe-dominated IS enviroment
of, say, 1981).

> I don't think there's much of a point to argue here, frankly -
> technically you're right, but in terms of OSS philosophy, at least
> in the modern 1990s Stallman/Raymond view of it, Unix was
> not "open source". Yet it ultimately led to "true" OSS knockoffs,
> particularly Linux. In that sense, I stand by my point to Dan -
> the fact that some piece of software
> isn't pure OSS doesn't make it evil, in and of itself. The OSS
> bandwagon is very trendy to climb aboard, and I'm a'climbin' along
> with most everyone else these days, but lets keep our heads here!
> Commercial software isn't evil in and of itself

See above remarks on the foundation of the Open Source movement being
Stallman's attempt to recapture what he had come to regard as his
natural rights as a user of software --- he very definitely does
regard closed-source software distribution as a usurpation of those
natural rights, and therefore as evil. And he isn't shy about saying
so, either. (NB don't say "commercial software" when you mean
closed-source; some open-source software is commercial).

NB that Raymond disagrees violently with Stallman on this moral issue;
he takes your position that closed-source distribution isn't
necessarily evil, but simply (for some purposes) less efficient. So
there really is no "Stallman/Raymond" view of OSS philosophy; the two
have *very* different philosophies, even if they use some of the same
means to pursue them.

> If Dan doesn't want to use or reference IBM no-source tools - that's
> his right. But taken to an extreme, this attitude is biting your
> nose to spite your face.

That depends. If your explicit goal (for moral or other reasons) is a
fully open-source computing environment, then adopting closed-source
tools for which open-source alternatives exist can be seen as
backsliding, even if the closed-source options are, momentarily,
technically superior in inessential ways.

> > Do you know if Sun has released the actual text of the license itself
> > (or a draft)? I've seen quite a bit about Sun's *intentions* (open
> > access to source, but no distribution of modified versions unless you
> > pass a compatibility test --- which they haven't made generally
> > available in the past --- and pay royalties), but I haven't yet seen
> > draft text of the license. I may be missing something...
>
>
> Well, you took the liberty of chopping my words above. You appear
> to intimate that Sun's willingness to be a true OSS convert may be
> less than first meets the eye - and I share these suspicions!
> In addition to the words above, I wrote:
> > with some
> > form of "free" source license. The meaning of "free' is still being
> > disputed - check slashdot for lots of discussion of this - but
> > there's no doubt that Java is being strongly pulled in the direction
> > of open source.
>
> There's lots of dispute about just how 'free' the SCSL really will be.
> The fact that Sun appears to reserve the right to charge royalties
> once you go commercial with a product makes the whole thing seem,
> to me, a bit like front-load and back-load mutual fund sales. You
> have to pay a commission - the only question is, do you pay up
> front or at the end?

I've chopped you less this time; my apologies if I distorted what you
wrote; that wasn't my intent. At any rate, what I wrote is a
paraphrase of what I find in the Q&A with Javasoft's marketing manager
--- in particular, this Q&A:

How and when do you charge fees for usage?

The only time a licensing fee is charged is at the point of
distributing the modified source code in a product for either
internal usage or commercial redistribution. This is a major
change from our previous model, which required up-front fees.
We have eliminated the up-front fees; we make money only

As to Sun's sincerity, they aren't *claiming* to be open source...

> (Hey Robert - I've been messing around with latex2html lately,
> and saw your name all over the docs - cool :). You really do
> get into some interesting projects, don't you? latex2html, apache,
> what are you working on these days?)

latex2html --- that was *ages* ago. I was just starting up the MIT AI
lab web site, and I was doing bulk conversions of old LaTeX memoranda
as a way of getting some content --- *any* content --- on the site
itself. The documents in question had a lot of nasty hacks in them,
which made them an excellent torture test for Latex2HTML, and I wound
up fixing a lot of bugs to deal with them. (So you think $...$ math
mode delimiters can't nest? Guess again:

$x + \hbox{ something like$y$} + z$

That was a fun one).

As to my current project, it's thoroughly dull, at least from a
technical perspective --- a reminder and messaging system for patients
to the Children's Hospital (of Boston) emergency room. There are
interesting issues in the construction of such systems, but I'm not
exploring them at the moment. Sic transit gloria mundi...

rst