Subject: Re: [was Cygnus] now Ghostscript and Aladdin
From: "Stephen J. Turnbull" <>
Date: Tue, 22 Jun 1999 15:27:54 +0900 (JST)

>>>>> "Brock" == R Brock Lynn <> writes:

    Brock> I think if there were just one ghostscript and it was all
    Brock> licensed under a unified GPL, or similar free software
    Brock> license, there would be more coders who would be willing to
    Brock> do the work to write additional drivers, including maybe

I think this is highly unlikely.  The Aladdin license is quite
permissive enough for Linux _users_; as you should know: Debian
includes Aladdin Ghostscript, albeit in the non-free category.  Peter
certainly accepts non-Aladdin licenses, but very few people insist on,
say, GPL.  If they do, he distributes their code in source form with
the Aladdin package (of course, binaries incorporating them cannot be
redistributed at all---because of the terms of the GPL).

Most Linux users I know don't even realize that Ghostscript isn't free
software.  And when the difference is explained to them, they usually
react "hey, that's a great idea!  Why does GNU do it the GPL way?"
Obviously, developers will have a different set of reflexes, but if it
comes with widely redistributable source, that's "first-approximation
free" for most people, I suspect.

    Brock> even reverse engineering and so forth, for inclusion into
    Brock> ghostscript. I would even say that I'd do the work, if I
    Brock> had the experience and know-how, but because there is a
    Brock> dual license and other iffy things I personally wouldn't

What is this "iffy" BS?  So much for the idea that there's no FUD
applying to non-GPL licenses.[1]


    Brock> bother about doing the work if I knew the "official"
    Brock> version of ghostscript may deviate and leave my drivers in
    Brock> the dust, or other such things.

This is a complete myth.  In fact, I would suspect that dual
(actually, multiple---there are two free public licenses, but there
are also commercial licenses) licensing leads to _more_ stability in
such interfaces, because Peter has contractual commitments to clients
to support their drivers, and would prefer to avoid breaking them
(creating work for himself) by interface changes.

On the other hand, as is quite well known (if you don't know it, just
go look up "Emacs 20" && "Mule" on comp.emacs), a unified-licensed
package under the control of a single maintainer can go off and break
compatibility willy-nilly.

I really doubt that any developer would go off and break compatibility
with his own code without a very good reason.  I think RMS's reason
for incorporating the Mule code was excellent, but many of his users
disagreed because it broke _their_ code (note, not RMS's problem) or
cost them efficiency for no personal benefit.  Peter's contractual
commitments to his clients force him to internalize their concerns,
but RMS can ignore them if he chooses to.  Thus, my conclusion that
Aladdin would probably be more conservative.

"Conservative" may or may not be a good thing.

    Brock> Just knowing there is one unified version, and knowing it's
    Brock> 'free', 

What makes you think either of those is not true about Ghostscript?

Sure, Aladdin Ghostscript is technically non-free, but there's nothing 
to prevent you from contributing to GNU Ghostscript instead, or
insisting that your Aladdin driver be simultaneously licensed GPL and
distributed with the next version of GNU Ghostscript.  As for
"unified," GNU Ghostscript is just Aladdin Ghostscript lagged about a

OTOH "one unified version" is most certainly _not_ true of Emacs, yet
since the fork I would imagine the (code) contributors to the two main
branches combined number in the hundreds.  Grepping a recent archive of
"xemacs-patches" (July 1998--Jan 1999) alone gives 30 contributors, and
there are several names I was surprised to _not_ see, who I know have
contributed code in the last couple of years.  And that's just code
(including documentation patches); not to mention developers emeriti
who show up occasionally to consult on architecture and arcana and of
course scores of beta testers who often critique patches they might
not have been able to write themselves.

    Brock> would give me the green light to feel nice and warm
    Brock> and fuzzy about spending some time to code up additional
    Brock> drivers for the thing.

I don't think "contributing to the support of the GPL" is what drives
development for most people; I work on Ghostscript (very little, these
days) and XEmacs because I live(d) in them.  Actual contribution was
based on the knowledge that it would be redistributed openly, but I
don't need GPL to feel warm and fuzzy about that.  There are lots of
other projects I might want to contribute to; if everyone can read,
modify, and redistribute the code to all users, I won't worry too much
about the detailed vehicle the primary author chose as free public

Just as you are doing, I might advocate changing their licensing
policy, but not on the basis that it would increase the number of
developers willing to work on their project due to "license bias".
Most people have a strong preference for licenses for their own work,
and of course advocate it to others.  But I find it hard to imagine
many refusing to work on or submit a patch due to minor license

    Brock> If perhaps Aladdin would reconsider their stance on
    Brock> licensing, and unify the codebase, if it isn't already, I
    Brock> think a lot more outside occasional hackers would get
    Brock> involved in improving the codebase, and thus benefiting
    Brock> whatever kind of business was built up around the code to
    Brock> support it commercially, in a business fashion.

Probably a rather marginal effect.  Remember that Aladdin, perhaps
uniquely among existing FSBs (probably Cygnus, too), can make strong
alliances with hardware manufacturers.  These are exactly the people
who might prefer that code related to their (unreleased) products
remain closed until release, and so they will not benefit very much
from "occasional hackers."  But I bet they make up a substantial
majority of Ghostscript revenues.

As pointed out elsewhere, the usual problem with drivers is the
difficulty of getting documentation for the target.  Where the model-
specific documents are not available, the drivers don't get written.
This is true for Ghostscript and PCMCIA-cs, with complex licensing,
and it's true for Linux IrDA, which is pure GPL, and for XFree86,
which is XFree86-style but still fully free..

[1]  Please don't read this as an attempt to justify my behavior in
saying that RMS was spreading FUD; it is not.  It is simply a
statement that, taking Brock's words literally, there is FUD out there
about non-GPL licenses.  I'm not accusing anybody of spreading it;
it's quite capable of arising spontaneously out of social dynamics, I

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 two straight lines for?  "Free software rules."